Conference PaperPDF Available

Machine-Learned Emission Model for Diesel Exhaust On-Board Diagnostics and Data Flow Processor as Enabler


Abstract and Figures

Conventional methods of physicochemical models require various experts and a high measurement demand to achieve the required model accuracy. With an additional request for faster development time for diagnostic algorithms, this method has reached the limits of economic feasibility. Machine learning algorithms are getting more popular in order to achieve a high model accuracy with an appropriate economical effort and allow to describe complex problems using statistical methods. An important point is the independence from other modelled variables and the exclusive use of sensor data and actuator settings. The concept has already been successfully proven in the field of modelling for exhaust gas aftertreatment sensors. An engine-out nitrogen oxide (NOX) emission sensor model based on polynomial regression was developed, trained, and transferred onto a conventional automotive electronic control unit (ECU) and also proves real-time capability. Within this study several approaches are demonstrated to show the impact on model accuracy with varying numbers of inputs and polynomial structures. However, with increasing accuracy demands, these models require additional resources of storage and computational performance. One solution is the use of a data flow processor (DFP, developed by DENSO), which can process parallelizable algorithms very efficiently and only requires a fraction of the computation time compared to a conventional ECU. It has been shown that the computation time of the proven polynomial regression model on the DFP can be reduced by a factor of 20. This paper provides an overview of applications of machine learning algorithms and offers an outlook of the advantages of a DFP.
Content may be subject to copyright.
2021-01-5108 Published 17 Dec 2021
Machine-Learned Emission Model for Diesel
Exhaust On-Board Diagnostics and Data Flow
Processor as Enabler
Marco Moser and Steve Kipping IAV GmbH
Hiroyuki Hirayama NSITEXE Inc.
Citation: Moser, M., Kipping, S., Higuchi, K., and Hirayama, H., “Machine-Learned Emission Model for Diesel Exhaust On-Board
Diagnostics and Data Flow Processor as Enabler,” SAE Technical Paper 2021-01-5108, 2021, doi:10.4271/2021-01-5108.
Conventional methods of physicochemical models
require various experts and a high measurement
demand to achieve the required model accuracy. With
an additional request for faster development time for diag-
nostic algorithms, this method has reached the limits of
economic feasibility.
Machine learning algorithms are getting more popular
in order to achieve a high model accuracy with an appropriate
economical eort and allow to describe complex problems
using statistical methods. An important point is the indepen-
dence from other modelled variables and the exclusive use of
sensor data and actuator settings.
e concept has already been successfully proven in the
eld of modelling for exhaust gas aertreatment sensors. An
engine-out nitrogen oxide (NOX) emission sensor model based
on polynomial regression was developed, trained, and
transferred onto a conventional automotive electronic control
unit (ECU) and also proves real-time capability. Within this
study several approaches are demonstrated to show the impact
on model accuracy with varying numbers of inputs and poly-
nomial structures. However, with increasing accuracy
demands, these models require additional resources of storage
and computational performance.
One solution is the use of a data ow processor (DFP,
developed by DENSO), which can process parallelizable algo-
rithms very eciently and only requires a fraction of the
computation time compared to a conventional ECU. It has
been shown that the computation time of the proven poly-
nomial regression model on the DFP can be reduced by a
factor of 20.
This paper provides an overview of applications of
machine learning algorithms and oers an outlook of the
advantages of a DFP.
The rising number of powertrain components leads to
an increasing demand for on-board diagnostics (OBD)
and, as a consequence, to the higher complexity of the
algorithms in automotive electronic control units (ECUs),
which need to beperformed in real time. On top of the more
stringent OBD limits and the upcoming on-board monitoring
(OBM) in future emission, legislations all over the world
demand an even further increased accuracy and therefore add
to this challenge.
Sensors are part of these powertrain components and are
needed to control processes or to detect failures. Sensor
models are required to diagnose these sensors or to provide
an alternative value if no sensor is available. Former sensor
models for the automotive exhaust line were very complex.
Exhaust lines and control strategies get even more complex
(i.e., due to dierent heating modes), emission standards get
tighter; hence, more sensors and models are needed, and
sensor models need to get more accurate while
maintaining robustness.
In addition, there is a high demand for a faster develop-
ment process, which is even more serious for the OBD, since
it is calibrated at the end of the development cycle.
Machine learning algorithms are getting more into the
focus and are increasingly used in ECUs in order to achieve
a high model accuracy with an appropriate economical eort.
An engine-out nitrogen oxide (NOX) emission sensor
model based on polynomial regression was successfully devel-
oped, and a soot model is currently under development. is
paper demonstrates the impact on model accuracy of dierent
© 2021 Moser, Kipping, Higuchi, Hirayama; Published by SAE International. This Open Access article is published under the terms of the Creative Commons Attribution
License (, which permits distribution, and reproduction in any medium, provided that the original author(s) and the source
are credited.
Downloaded from SAE International, Friday, May 13, 2022
approaches with dierent numbers of inputs, polynomial
degrees, a nd polynomia l combinations of two and three inputs.
It has been proven that it is possible to use machine
learning models in different domains within the ECU.
However, with increasing accuracy demands these models
require additional inputs and enhanced polynomial
complexity and therefore need additional resources of storage
and computational performance of the already limited ECU.
For such purpose DENSO developed a data ow processor
(DFP), which can process parallelizable algorithms very e-
ciently and only requires a fraction of the computation time
compared to a conventional ECU. An evaluation of the poly-
nomial regression model on the DFP shows that the computa-
tion time can bereduced by a factor of 20 and allows to easily
implement multiple models to beperformed in parallel.
With a high number of machine-learned models which
are mostly developed in a highly parallelizable structure, opti-
mized ECUs could bevery helpful to enhance the overall
computation performance.
Literature Review
A function development team of the Hubergroup developed
a neural network for the NOX sensor upstream of the Selective
Cataly tic Reduction (SCR) catalyst since a physical NOX model
did not show sucient accuracy to replace the NOX sensor
and therefore save hardware costs. e trained model is fed
with real data, which were not used for training. e accuracy
of the model increases with an increase of inputs. Failure
tolerances were not considered yet as well as the transfer to
dierent vehicle platforms [1, 2].
In 2016, the Innovative Transportation Research Institute
worked on dierent machine learning approaches to estimate
idling emissions in a gasoline vehicle during a mini durability
run of 1600km with dierent driving patterns. Data from a
portable emission measurement system was used to train CO2,
CO, HC, and NOX emissions. Among the ve dierent models
investigated, the boosted bagged decision tree was identied
to bethe best-t approach [3].
Bosch developed a method to integrate machine learning
know-how into the design of experiments steps to identif y the
most relevant combinations of inputs and achieved an accurate
machine learning model. e Gaussian process regression has
proven to bethe most favorable approach due to their natural
smoothness assumptions and the capability to consider the
temporal history of inputs. Bosch developed a new generation
of control units (Bosch MDG1, from device 3 upwards,
including an Adva nced Modelling Unit) to beable to calculate
the Gaussian regression model in real time [4].
In 2019, PSA Group and IFPEN developed in cooperation
a neural network and an empirical correlation to estimate
NOX emission during real driving conditions to substitute the
upstream NOX sensor [5]. e empirical model was used to
describe the combustible composition in the combustion
chamber and therefore to reduce the number of inputs into
the neural network. e validation was based on a wide
measurement campaign. e design of experiments was used
to train the model with 12 inputs. A network of one layer with
ve neurons was derived to prevent overtting and achieve
sucient accuracy [6].
e dissertation of Lumpp in 2011 describes the develop-
ment of an empirical NOX model with high accuracy. e basis
of the modelling here is a consideration of the processes in
the engine and the formation of relevant variables in the
combustion chamber by modelling with existing measured
variables. ese models are then used as features for the
empirical model. Global expert knowledge is therefore
required to create this NOx model. Furthermore, the inves-
tigations are based on test bench measurements of an engine
test bench. Its signicance for a eet is, therefore, limited [7].
In 2017, the University of Salerno investigated the possi-
bility of using a recurrent neural network (RNN) to model
NOX formation in the engine. e model was designed to adapt
to an existing NOX sensor over runtime (aging). An oxygen
model was created as an important input. ey also dealt with
a compensation of the temporal shis between the sensors.
e test environment consisted of a vehicle on a chassis dyna-
mometer and mainly standard cycles—casting doubt on the
robustness over all environmental variations [8].
Sequenz and Isermann discussed 2011in their paper the
use of “global-local” structures in ECUs for modelling emis-
sions. e aim was to replace the state-of-the-art two-dimen-
sional (2D) map-based approaches. Splitting the feature
dimensions into areas with good predictability is the idea
behind this. e Result is a model of models with dierent
parameter sets. For this evaluation it leads to a too complex
soware st ructure and is therefore not investigated f urther [9].
Outline of this Paper
e objective of this paper was to study the possibilities of
simple polynomial models for exhaust gas sensors that were
parameterized and optimized by machine learning algo-
rithms. Independence from other modelled variables and the
exclusive use of sensor data and actuator settings was a key
point. e goal was to t this model into the most commonly
used automotive control units.
Furthermore, the eciency impact of a new DFP with
such a model was studied.
is paper shows the possibilities of the interplay between
optimizations in the model structure and the machine
learning process and the development of suitable hardware
in order to make optimal use of the new functionalities. A
holistic approach to the interaction of hardware, soware,
and optimization processes is the key to improvements in
automotive control units.
Conventional Approach
e development process for new ECU functionalities was
performed in Matlab/Simulink®. Exist ing models can beeasily
adapted or new ideas can besketched and investigated. ECU
measurements were used to validate the simulation environ-
ment. Small errors were tolerated due to the fact that the ECU
functionalities were implemented in a fixed-point
Downloaded from SAE International, Friday, May 13, 2022
environment on the ECU, which is a dierent hardware
compared to a discrete calculation on a conventional
personal computer.
However, these models are used to create and analyze
new concepts and also to derive an initial calibration.
Dependent on the data basis and the state of the calibration,
it is also possible to use the models to show the robustness of
a calibration. e bigger the data basis the more reliable is the
information regarding robustness. It helps to have data from
dierent vehicles, driver and test cycles, as well as test trips
from extreme conditions such as altitude and heat.
Especially for OBD reasons, the simulation environment
can also beused to perform tolerance analysis on relevant
sensor signals to prove diagnosis robustness against false
detection. In general, a Monte Carlo simulation is performed
with a huge number of runs to catch the most relevant combi-
nations to sensor tolerances. Dependent on the data basis and
the number of runs, this can bea time-consuming task, which
can take days or even weeks.
Once the models are successfully validated, they can
beimplemented, which means the models are compiled into
machine code and tested furthermore via various test cases.
ECU Code in Python
e current development process for OBD functions consists
of the development, testing, calibration, validation, and trans-
lation of functionalities for an ECU up to the point when it is
ready for implementation. Hence, the compiled les for several
ECU functionalities are already available. So the idea was to
use the machine learning environment for the NOX model,
which is in Python already, instead of getting the calibration
into Matlab and a separate model.
Cython, which combines Python and C and acts as an
interface, was used to implement the existing source code into
Python. e original source les do not need to bechanged
at all. What needs to bedone is the creation of a setup le,
which “cythonizes” the source les and also contains compiler
information. It also handles the data transfer, all the datat ypes,
and the calibration. So the source code needs to bewell known
regarding inputs and outputs and their dataty pes. Aer execu-
tion the output file can be imported in Python as a
regular module.
e advantage of using this approach is that Cython code
itself is approximately three to four times faster than pure
Python code [10]. Compared to the conventional approach, it
is also approximately three to four times faster. So the ultrafast
simulation time (1 h measurement is performed in 0.5 sec)
helps especially for a huge number of data and runs. A
comparison against an ECU measurement reveals that the
accuracy is even increased compared to the conventional
approach. is is because the source code is executed in the
same manner as on the ECU with the correct datatype
and arithmetic.
It also allows a high potential for automation, especially
for auto-generated source code. Nevertheless, the analysis
capabilities in Python are almost endless due to the great
open-source community. Not to mention are the nonexistent
license costs compared to, for example, Matlab®.
However, the drawback is the initial setup, which can
be quite time consuming due to poor documentation.
Furthermore, this approach lacks the ability to change the
functionalities aerward once the source code is dened and
imported. Regarding measurement points, youare limited to
what is dened in t he source code, and it is not possible to add
additional signals as needed.
e relevant ECU functionalities considered for this project
are developed in Matlab/Simulink®. is has the advantage
that new concepts can befast evaluated and benchmarked
against conventional or alternative approaches. It also helps
for further investigations regarding robustness and tolerance
analysis. From there the models are processed with dierent
toolchains into C-Code for ECU integration.
Another benet of Matlab/Simulink® models is that they
can also betransferred easily to rapid prototyping systems
such as ETAS ES910 and therefore betested in a real hardware
environment (Figure 1).
For such purpose, a demonstrator vehicle (mild-hybrid
diesel passenger car with a four-cylinder engine of 1.5 l
displacement) from previous work was used [11].
e aertreatment system consists of a close-coupled (cc)
lean NOX trap (LNT), ccSCR/SDPF combination, and an under-
oor SCR/ASC system and fu llls the EU6d and possible future
legislations. e engine-out NOX sensor, which is modelled
within this investigation, is located upstream of the LNT.
Within this context, the polynomial structure of the
engine-out NOX model was implemented in the demonstrator
vehicle and trained with on-road data. Due to the simple
nature of the polynomial structure (Equation 1), the model
was created within the Matlab/Simulink® environment using
several for-loops in an embedded Matlab function to calculate
the polynomial output instead of writing out the entire poly-
nomial structure. Dependent on the approach and the
maximum number to be combined for each polynomial
product, several for-loops are active.
Various measurements of urban and rural driving were
used to train the model. e preparation and selection of data
for machine learning represents a signicant part of the work
and is not described in this paper. Necessary steps include,
for example, gas-flow-dependent time shifting of sensor
signals to a common point, exclusion of ammonia slip via
 FIGURE 1  Demonstrator vehicle with rapid prototyping
hardware for model implementation.
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
low-pressure EGR, sensor status assessment, and
other considerations.
NOX Sensor
e automotive diesel air path is controlled and diagnosed by
many dierent sensors (e.g., temperature, pressure, mass ow,
and gaseous species). e modelling of mass ow, tempera-
tures, and pressures is fairly simple because of the direct
physical correlation to its surroundings.
For gaseous species the lambda/O2 sensor is most
common. As for the former sensor types, lambda models are
also relatively easy to create. ere are only a few but major
dependencies to values like mass ow, EGR, and fuel mass.
All those sensors can be modelled with sufficient
physical approaches.
e NOX sensor signal on the other hand is much more
complicated to model. e NOX formation is a complex four-
dimensional process with many temporary local eects [12].
Building a reliable model is a costly and time-consuming
eort. e results of the simpler models needed for ECUs are
not sucient for future needs regarding control strategies
and OBD.
Considering this the NOX sensor is chosen to prove the
possibilities of machine-learned models for diesel air and
exhaust path. e acquired insights are applicable to most
other air path sensors.
e NOX sensor used in the demonstrator vehicle is a
state-of-the-art electrochemical sensor [13].
Model Structure
e aim is to install the NOX sensor model in a standard ECU
(e.g., Bosch, DENSO, Delphi, etc.). Knowing the restrictions
of an automotive ECU helps to choose a reasonable model
structure. Some major restrictions for implementation are
e model structure has to bexed. It is not possible to
change it during runtime.
e impact on computation time is very important.
Every module in the ECU soware has a determined
time task to nish a single computation of the
entire function.
e memory is limited. Only a few as possible feedbacks
should beused.
Because the target is to provide an analog output, it is
obvious to choose a regression model for the NOX sensor.
Feedforward models are preferred because of low
memory usage.
With these considerations, polynomial model structures
are used for this investigation. Another benet not to ignore
is the easy disclosure to authorities. As an alternative, neural
networks-like multilayer perceptron (MLP) or recurrent
(RNN) are also possible to implement on a standard ECU, but
they are much more dicult to explain in their behavior and
so they are not the rst choice here. e polynomial structure
used for the regression consists of independent linear terms
whose response to a change in an input is easily understood.
Polynomial models are described by their inputs
(features), the order of the terms, and the input combinations.
A polynomial of at most three input combinations is described
through the equation
ee ij
ee ij
 
12 2
ee ij
ee ijk
e13 34 4
Eq. (1)
x are model inputs
y is the oset
i, j, k are the feature combination
e1, e2, e3, e4 are the polynomial degree
The following settings were used to investigate the
polynomial performance:
e overall number of inputs into the model varied
between 16 and 24.
e number of input combinations used to calculate a
product of the polynomial structure was either two
or three.
e maximum order of the products of the polynomial
structure was set to ve.
Dierent algorithms were taken into account for
model training.
For the size of the model, this results in a wide range. To
visualize the dierent sizes of the model, Tabl e 1 lists the
number of coecients.
All the dierent combinations, which were possible to
beimplemented in the hardware (microcontroller 256 k B) and
able to betrained on the given hardware (PC with 256 GB
RAM), are discussed in the following section.
Model Inputs
e common way to select features for machine learning
models is to take all available signals and do a correlation
TABLE 1 Model setup and related number of coecients as
characteristic value for the model size.
Input combination 2 Input combination 3
order 16 20 24 16 20 24
2152 230 324 152 230 324
3 408 630 900 968 1770 2924
4748 1220 1752 3024 5780 9848
5 1280 2000 2880 6880 13,400 23,120
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
analysis for ltering in respect of the target signal [14]. For
polynomial structures, the additional orders of the signals
and the combination of the signals and their higher orders
are needed for this evaluation.
ere are many signals available in the ECU, which can
beused for features of a NOX model, e.g., raw sensor signals,
ltered sensor signals, modelled sensors, targets and actual
values of actors, modelled mass ows, etc. A sensor is not
available for every useful variable, so it may benecessary to
fall back on a target variable of an actuator or a model. Existing
models are used here as a form of feature engineering, which
bundles information through preprocessing [15].
Fundamental understanding of the physicochemical
system leads to a faster progress in model development, and
that knowledge is available through former modelling
researches [12]. For our investigation, wecount on the most
common ltered sensor signals and target values of actuators.
Signals from models have to be verified. If they are
reliable, they can beused to reduce the dimensions of the
polynomial structure if the model itself combines features of
the machine learning model to a new signal [16]. e use of
models was explicitly not desired for this study, as the models
would have to beevaluated for this purpose.
For the demonstrator car wehave the following veried
signals from the air path to use for modelling the NOX sensor:
Pressures: environment (p_0), inlet (p_ 2), outlet (p_3),
dierence pressure SCR-DPF (dp_SCRF), rail (p_rail)
Temperatures: environment (T_0), intake (T_1), inlet
(T_ 2), outlet (T_3), turbine downstream (T_4), engine
(T_e ng), fuel (T_ fu e l)
Mass ow: fuel (dm_fuel), intake (dm_maf )
Other: rotation (n_eng), torque (M_eng), injection angle
(phi_fuel), oxygen (lambda), NOX SCR-DPF downstream
(r_NOx2), speed (vs)
Target values of valves because of not existent actual
feedback: throttle valve (r_e mtv), high- and low-pressure
EGR valve (r_hp, r_lp), exhaust gas ap (r_et)
Figure 2 shows the result of a simple correlation analysis
of all reliable signals to the NOX sensor signal.
e direct correlation of the second NOX sensor and its
combinations with other features is reasonable. e same
applies to mass airow, fuel mass ow, injection angle, rotation
speed, and others. ese features have similar frequencies.
Other signals (e.g., engine temperature) are known to have a
high physical inuence on NOX formation [12] but, because
of their different frequency, are not detectable as high
performers with correlation analysis.
e next step can bean embedded feature selection to
nd not relevant features through direct model learning.
Learning algorithms like Lasso Lars use regularization to
penalize not important features to zero. Aer one model
learning, the most penalized features can be excluded.
erefore the number of features can bereduced even further.
e following section considers the variations of the model
structure in terms of number of features, feature combination,
polynomial degree, and choice of optimization algorithm. It
is necessary to solve the conict between the highest possible
accuracy of the model and a resource-saving model structure.
e results show that high accuracy does not necessarily
require the most complex model structure.
Feature Selection
Every evaluated signal enhances the model to more accuracy
but also leads to a bigger structure too—so the number of
features is a trade-o between accuracy and size of the model.
As long as the accuracy of the model is high enough, fewer
inputs are preferable. Correlation analysis, model training
with dierent subsets of features, and expert system analysis
can help to take out the most unimportant features to reduce
the size of the model.
For the evaluation, wechose feature sets of 24, 20, and 16
signals. A set of 24 was the maximum number over all of the
available measurements and 16 was the set of features which
was the minimum set to describe the modelled system.
Figure 3 shows the inuence of the reduction of features
on model size and accuracy. Both models were trained with
 FIGURE 2  Correlation to NOX sensor signal (features 24,
pol. order 3, pol. combinations 2).
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 3  Correlation to NOX sensor signal with varying
features 16 and 24 (pol. degree: 5, pol. combinations: 2).
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
the same dataset and Ridge Regression for the learning algo-
rithm. e model structure was with a combination of two
features and a maximum polynomial degree of 5 for each
polynomial term. e reduction from 24 model inputs to 16
inputs reduces the model size from 2880 to 1280 terms. But,
at the same time, the standard deviation in the NOX model
oset increases from 15.4 ppm to 22.1 ppm.
e nal choice of the model structure depends on the
restrictions of the ECU and the needs for the use case of
the model.
Feature Combination
e correlation analysis shows that the polynomial combina-
tion of features leads to a gain of information for the
NOX model.
e rst rows of the list in Table 2 indicate that a combi-
nation of two or three features to one polynomial term has
high informative content. erefore, it is sensible to use
combinations of at least two features.
For the model integration into the ECU, it is important
not to exceed a certain size of the model. For each combination
order, the number of polynomial terms and thus the structure
doubles (see Equation 1). With the number of features going
up to 24, combinations of 3 features are the acceptable
maximum for a module in a standard ECU to run in the 100
ms time task.
Figure 4 shows an example of a polynomial input combi-
nation of 2 and 3 for 20 features and a polynomial degree of
4. e standard deviation is improved from the combination
of 2 to the combination of 3 features from 21.4 ppm to
15.2 ppm.
e additional polynomial combinations lead to a model
enlargement. For this example the number of polynomial
terms rises from 1220 to 5780. is has to be taken into
account for assessing the best specic solution for the target
ECU. e model with a feature combination of 3 is more than
four times as big as the model with combinations of 2 features,
and for standard ECUs the total number of 5780 terms and
parameters for one module would be a hindrance
for implementation.
Polynomial Degree
e degree of a polynomial is the maximum sum of the expo-
nents of a polynomial term. Table 1 shows a degree of 5in the
third row. Higher degrees are not evaluated because of the
high impact on the size of the polynomial; see Equation 1.
Figure 5 shows an example of a polynomial degree of 2
and 5 for 20 features and a feature combination of 2. e
standard deviation is reduced for the change from degree of
2 to degree of 5 from 31.5 ppm to 19.9 ppm.
e scatter plot of all test points and the histogram of
Figure 5 shows an increase of accuracy—but the higher poly-
nomial degree leads to an increased number of polynomial
terms (degree 2, 230 coecients vs. degree 5, 2000 coe-
cients). On the side the results indicate that, despite the very
small model with a polynomial degree of 2, a usable model
could bebuilt.
is evaluation is a test of dierent useable standard regres-
sion algorithms (e.g., Linear Regression, Ridge Regression,
Lasso Lars, SGD Regressor, and MLP) from Scikit-Learn [17].
e target was not to nd the optimal algorithm, but to prove
that there are different useful algorithms for different
TABLE 2 Sorted correlation to NOX sensor signal (features:
24, pol. order: 5, pol. combinations: 3).
Rank Polynomial term Correlation to NOX
1 r_NOx2 0.85
2 r_et^2*r_NOx2^1 0.84
3 r_et^4*r_NOx2^1 0.84
4 r_et^3*r_NOx2^1 0.84
5 r_et^1*r_1p^1*r_NOx2^1 0.83
6 r_1p^2*r_NOx2^1 0.83
7 r_et^2*r_1p^2*r_NOx2^1 0.83
8 r_et^3*r_1p^1*r_NOx2^1 0.83
9 r_et^1*r_1p^3*r_NOx2^1 0.83
10 r_et^1*r_1p^2*r_NOx2^1 0.83
11 r_1p^4*r_NOx2^1 0.83
12 r_1p^3*r_NOx2^1 0.83
13 r_et^2*r_1p^1*r_NOx2^1 0.83
14 r_et^1*r_NOx2^1 0.83
15 r_1p^1*r_NOx2^1 0.83
16 dm_maf^1*r_et^1*r_1p^1 0.76
17 dm_maf^1*r_et^3*r_1p^1 0.76
18 dm_maf^1*r_1p^2 0.76
19 dm_maf^1*r_et^2*r_1p^2 0.75
20 dm_maf^1*r_et^2*r_1p^1 0.75
21 dm_maf^1*r_et^1*r_1p^2 0.75
22 dm_maf^1*r_et^1*r_1p^3 0.75
23 dm_maf^1*r_1p^3 0.75
24 dm_maf^1*r_1p^4 0.75
25 dm_maf^1*r_et^2 0.74
26 r_lp^1*r_NOx2^2 0.74
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 4  Correlation to NOX sensor signal with varying
polynomial combinations 2 and 3 (features: 20, pol. degree: 4).
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
problems. e dierent algorithms need specic constructions
of the ECU soware related to the specic optimization of the
weighting factors.
To understand the selection of used algorithms here is a
short description of the tested algorithms:
Linear Regression: e weighting factors (coecients) of
the polynomial are optimized to minimize the residual
sum of the error between the target sensor and model
approximation. ere is no limit to the value area of the
weighting factors, so it is not optimal for the planning of
a xed-point-based structure for automotive ECUs.
Besides this, this algorithm is sensitive to random errors
in the target.
Ridge Regression: is algorithm extends the Linear
Regression with an L2-norm regularization. e size of
the coecients is penalized. is results in a set of small
weighting factors, and most of the factors are none zero.
e weak point here is the small quantization dierences
between the coecients which can lead to noise eects
in the model prediction.
Lasso Lars Regression: is algorithm extends the Linear
Regression with an L1-norm regularization. Weighting
factors near zero are penalized and set to zero.
Correlating features are treated equally with
equiangularity between the weighting factors. is
together tends to give a set of good-natured coecients
concerning the value area. e algorithm can
besensitive to noise eects.
SGD Regressor: Stochastic Gradient Descent estimates
the gradient of the model error at every new sample. is
makes it t for large datasets of more than 100 k samples.
e regularization is tunable between L1- und L2-norm
so that the set of resulting coecients is adjustable to the
needs of the soware structure inside the ECU.
MLP: is algorithm is not a polynomial approach like
the others. An MLP is, in general, a neural network. It
serves here as a benchmark.
In the investigations carried out here, the machine
learning algorithms were not optimized with respect to
their parameters so that the results correspond to a basic
series of experiments. The tuning of the so-called hyper-
parameters for the different estimators can be done by
manually varying the parameters or by using approaches
for parameter optimization from Scikit-Learn (e.g.,
GridSearch or HalvingRandomSearch).
For an example this evaluation was done with a polyno-
mial order of 5 and 20 inputs with Ridge Regression and Lasso
Lars as learning algorithms. e 2D histogram in Figure 6
shows a better result for the Ridge Regression algorithm. e
drawback of this algorithm is the granularity of the weighting
factors. Ridge Regression (L2 regu larization) uses much more
weighting factors and tries to use smaller va lues. is can bea
problem for the quantization of computation steps inside
the ECU.
e Lasso Lars algorithm with its L1 regularization tends
to have more weighting factors deactivated with 0—very
important features are favored with higher weighting factors.
e advantage of the algorithm can beseen in Figure 7. Lasso
Lars sets most of the weighting factors (about 10,000) to 0,
which can be seen in the histogram plot (right side, green
plot). The other algorithms show a more or less
equal distribution.
In addition, Lasso Lars a lso provides the smallest average
quantization error (le side, green plot). is means, for the
 FIGURE 6  Correlation to NOX sensor signal with dierent
learning algorithms: Ridge and Lasso Lars (features: 20, pol.
degree: 5, pol. combinations: 2).
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 5  Correlation to NOX sensor signal with varying
polynomial degrees 2 and 5 (features: 20, pol.
combinations: 2).
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 7  Quantization error at ECU implementation.
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
implementation into an ECU, more accuracy of the model
output because of more reliable accuracy of the
model coecients.
Results of Evaluation
Summarizing the different investigations there are the
following results.
Figure 8 shows the accuracy of the dierent learning a lgo-
rithms. Linear Regression, Ridge Regression, Lasso LARS,
and SGD Regressor make use of a polynomial which is easy
to implement in common ECUs.
MLP uses a dierent structure with hidden layers, and
this gives the advantage of much more weighting factors (here
a factor of 3 through 3 hidden nodes). us the algorithm here
primarily serves as a benchmark and is otherwise outside of
the evaluation.
The best results without optimization of the hyper-
parameters produce the Linear Regression algorithm. e
larger the polynomial in terms of degree, feature combination,
and number of inputs, the more accurate the model becomes.
Linear Regression has not had the high number of hyper-
parameters like the other algorithms. us the results of the
learning experiments already represent the optimum of this
algorithm. e disadvantages of the wide value area of the
coecients together with its strong correlation to huge poly-
nomials make it not a favorite for ECU implementation.
Ridge Regression performs even better with the rising
number of inputs, feature order, and input combination.
Depending on the capacity of the ECU, this algorithm has a
good performance for the NOX model, but is not as good as
the simple Linear Regression. e set of hyper-parameters
leaves room for optimization.
Lasso Lars seems to take no advantage of more complex
polynomial structures—the models’ accuracy remains rela-
tively constant. But through its behavior to set many weighting
factors to zero, it is possible to take unimportant inputs out
of the model or reduce computation load by ignoring zero set
computations. ese are the performance gains of Lasso Lars:
The polynomial can remain very simple for the same
model accuracy.
e SGD Regressor has the worst behavior in this test
environment. Because of the large number of hyper-param-
eters and the algorithms’ ability to produce an optimum
between the L1-regularization of Lasso Lars and the
L2-regularization of the Ridge Regression, SGD has the
greatest potential here to provide the optimal
learning procedure.
e MLP shows stable results over the dierent combina-
tions of polynomial order, number of inputs, and input combi-
nations. With enough capacity in the ECU, the algorithm is
a stable candidate for a reliable model.
Figure 9 shows the comparison between the best evalu-
ation performer (polynomial with 24 inputs, degree of 4,
feature combination of 3 and Linear Regression) and worst
performer (polynomial with 16 inputs, degree of 2, feature
combination of 2 and SGD). e big dierence between the
two test sets can beseen in the scatter plot as well as in the
standard deviation (73 ppm vs. 8 ppm). The scatter plot
together with Figure 8 shows that an even larger polynomial
can no longer bring any signicant improvements since the
model already represents the sensor very accurately.
e nal choice of model structure and learning algo-
rithm depends on many factors, and there is more than one
possibility to nd a good model. For this evaluation with no
direct restrictions, the Linear Regression with 24 inputs, input
combination of 3, and polynomial order of 5 leads to the
best result.
For series ECU implementation, a polynomial with 20
inputs, a polynomial degree of 4, and a feature combination
of 2 was favored because of the smaller model structure.
Because of its ability to shrink the parameters to a small
number, Lasso Lars was chosen to calibrate the model.
In Figure 10 a time trace of a measurement of a real street
drive is shown. e high accuracy of the machine-learned
model is clearly visible. e integrated values (“NOX cum”)
show only minor dierences. With such a model, OBD and
system controls are feasible on a high level. is approach is
not limited to engine-out emissions but can also beused for
any other sensor in the powertrain such as an air mass
ow sensor.
 FIGURE 8  Accuracy of dierent machine
learning algorithms.
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 9  Comparison between the best and worst
test performers.
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
Humidity and Air Quality: The
Missing Inputs
It is well known that humidity has an impact on NOX forma-
tion [17, 18]. Especially for high temperatures (>30°C), the
impact can bevery high because the absolute humidity (mass
of water in the intake air) has a wide range. Most of the current
passenger cars do not have a humidity sensor, so this impor-
tant input is missing. e air ow meter itself is sensitive to
absolute humidity, so newer versions of air ow meters have
a humidity sensor implemented [4]. It is very sensible to insta ll
such a sensor in future cars—not only for NOX modelling but
for emission controls too.
Another factor is the intake air quality. Combustion
processes in direct proximity (e.g., trac jam in close areas
of cities or long tunnels) or decomposition processes (e.g.,
swamp areas) can lead to lower oxygen levels which lead to
lower NOX formulation (like an external/outside exhaust gas
recirculation). Current passenger cars are not equipped with
an intake oxygen sensor.
Both disturbances of the NOX formulation result in devia-
tions of the oxygen level in the exhaust gas. It is possible to
detect this deviation and adjust the computation of the NOX
model. Because of insucient data of such environments, it
is not possible to show the eect here.
For ozone, it is possible to have an eect too [19]. e high
reactive gas species could lead to dierent combustion eects.
Because of a missing sensor, this could not be evaluated in
this research.
Adoption of Other Sensors
e model structure is usable for other air path sensors too.
Air ow meter or pressure or temperature sensors are candi-
dates for this kind of modelling.
As an example Figure 11 shows the modelling of a
pressure sensor upstream of the engine. For this model the
specic pressure sensor switched its input status with the NOX
sensor. e accuracy of this model is very high.
e same result is valid for other sensors or actors in the
air path, Figure 12 shows an example for the air ow sensor.
So the same model structure is usable for dierent sensors,
and the building of soware with many parallel models is
easily feasible.
In the end, there is a way to use many machine-learned
models with thousands of parallelizable computational steps.
The disadvantage here is the high load on the central
processing unit (CPU) that has to behandled.
About DFP
In the automotive eld in recent years, there has been a need
for processors that enable unprecedentedly sophisticated and
complex operations. e CPU, which is widely used today, is
connected to storage devices and I/O devices and is able to
handle various types of control instructions. However, due to
serial processing, there is a problem that real-time perfor-
mance is impaired as the amount of calculation increases. On
the other hand, a graphics processing unit (GPU), which is
 FIGURE 11  Model accuracy for pressure sensor
(engine inlet).
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 10  Time trace of NOX model accuracy.
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 12  Model accuracy for the mass air flow sensor.
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
becoming more and more popular for image processing, is a
processor that can process large-scale routine operations in
parallel at high speed. However, GPU tends to belarge in size
and consumes a large amount of power. In this context, DFP
has been developed as a processor that realizes high-speed
and power-saving processing while minimizing the size by
maximizing the utilization eciency of the parallel processing
mechanism. DFP is designed to beimplemented in a micro-
computer for automotive applications, and it is expected to
beone of the technologies to achieve an unprecedented high-
performance ECU by sharing the role with the CPU.
Architecture and Mechanism
of Parallel Processing
As shown in Figure 13, a DFP consists of three types of
processor cores: four scalar processing units (SPUs), a vector
processing unit (VPU), and a control core unit (CCU).
e SPU is an operations core and is able to execute
dierent processes independently in parallel (a function
similar to that of a multi-CPU). e VPU is a 512-bit wide
vector operations core, and it can execute up to 16 operations
concurrently if the data length is 32-bit (a function similar to
that of a GPU). e CCU is responsible for dynamically allo-
cating task processing to free SPUs to improve computing
eciency. Fi gu re 14 shows the outline of the parallel processing
ow in the DFP.
Each SPU executes independently assigned task
processing in parallel, and if there is a vector operation
instruction in it, the VPU executes the process in place of the
SPU. Sharing the VPU with each SPU in this way makes it
possible to achieve high throughput. By combining the task
parallelism and the data parallelism by vector operations, the
DFP achieves highly ecient processing while maintaining
the compact size and lower power consumption required for
use in a microcomputer for automotive applications. For the
polynomial models used in this paper, the operation of each
term and the sum of all terms were parallelized and processed
by the VPU. Moreover the entire process was divided into
multiple tasks that were assigned to and executed by indi-
vidual SPUs. is improved the utilization eciency of the
VPU and increased execution speeds.
Execution Time Assessment
e DFP execution time was measured by using the evaluation
board equipped with DFP and by means of clock counter value
at the measurement points on the soware framework. In the
same manner, as a benchmark, the execution time was
measured with an ECU equipped with an existing microcom-
puter for automotive applications (CPU clock frequency of
400 MHz). Figure 15 shows the relationship between predic-
tion error (Root Mean Square Error, RMSE) of NOX model
with dierent polynomial inputs and order and the execution
times of DFP and CPU.
Prediction error decreases by increasing the number of
inputs and order of polynomial model, but as a trade-o, it
can beseen that the CPU execution time increases exponen-
tially. is conrms that the calculation order O of the poly-
nomial model is O((np)c) for the combination of the number
of inputs n, the order p, and the number of combinations c.
On the other hand, the DFP keeps the execution time low by
performing parallel processing even though it was for the
same calculation order. Focusing on the polynomial model
with RMSE = 25 ppm from the viewpoint of OBD purpose,
the CPU execution time was 250 μs, while the DFP execution
time was about 1/20 of this, at 12 μs. is result shows that
the DFP enables large-scale processing such as polynomial
 FIGURE 13  Architecture of DFP.
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 14  Overview of parallelization process.
© Moser, Kipping, Higuchi, Hirayama.
 FIGURE 15  Result of execution time (DFP vs. CPU).
© Moser, Kipping, Higuchi, Hirayama.
Downloaded from SAE International, Friday, May 13, 2022
models to beexecuted in the ECU w ithout impairing real-time
performance. Furthermore, the result shows that an even more
complex and larger number of models can also behandled.
e target of this work was to evaluate the possibilities of
bringing a machine-learned emission model with high
accuracy to a standard control unit of a passenger car. e
challenge here is to nd an optimum of the conict between
the accuracy of the model and the correlating complexity of
its structure.
is paper shows the possibilities of simple polynomial
models for exhaust gas sensors that were parameterized and
optimized by machine learning algorithms. e goal was to
t this model into area-wide used automotive control units.
Furthermore, the eciency impact of a new DFP with
such a model was studied, and its benets were shown.
For the upcoming regulations for OBM, the need for very
accurate models of engine emissions is increasing. Just as with
the models for air mass and intake pressure already shown
here, models for soot, hydrocarbons, carbon monoxide, form-
aldehyde, carbon dioxide, and oxygen can also becreated with
appropriate emission measurement technology. e problem
here is the measurement data collection of emissions without
existing regular sensors.
e most important recommendation to future implementa-
tions of machine-learned models into automotive control
units is to make use of the developments of processor units
that are optimized for the needs of parallelized structures.
ere is a huge potential to make use of machine learning in
the automotive sector, but the standard control units have too
many restrictions. e real benet from machine-learned
models is to have a high range of structural possibilities and
a high accuracy of mathematical functions (e.g., exponential,
logarithmic, tangential functions).
To enhance the accuracy of machine-learned emission
models, more attention should bepaid to the distribution
of the model features and the target value. The distribu-
tion of input values from sensors and especially from
actors are not always ideal for machine learning. In the
best case, every feature and the target value will beindi-
vidually transformed to get as near as possible to Gaussian
normal distribution.
Optimization of learning algorithms is the next going on
from this evaluation. It is very important to know well the
capabilities and limitations of each learning algorithm to
guarantee a stable model outside the trained data space.
Knowledge about the behavior of interpolation and extrapola-
tion in areas outside the known data range is essential for an
evaluation of the usability of the model in an
automotive application.
For emission models the use of polynomial structures is
relatively safe in the context of approval of emissions regula-
tion from the authorities. For other machine learning
approaches like neural networks (MLP in this evaluation),
there is no procedure at the moment to get an approval. Here
lies a major problem to use machine-learned models. A
process must beagreed upon with the authorities in order to
obtain approval for dierent soware structures. On the
developer side, it should beclaried in advance how such
structures can bevalidated and secured.
anks to DENSO for providing the evaluation results for the
DFP, AECC and IPA for providing the car to collect the data
and IAV for supporting this paper.
Contact Information
Marco Moser
Sr. Smart Data Analyst for OBD Functions
IAV GmbH, Carnotstr. 1, 10587 Berlin
Tel: +49(0)30 -399789176
Steve Kipping
Function Developer OBD
IAV GmbH, Carnotstr. 1, 10587 Berlin
Tel: +49(0)30-399789247
Kazuhiro Higuchi
Powertrain System Development Div.
1-1, Showa-cho, Kariya, Aichi 448-8661, Japan
Tel: +81(0)566-61-4441
Hiroyuki Hirayama
Processor Development Dept.
2-16-4, Kounan, Minato-ku, Tokyo, 108-0075, Japan
Tel: +81(0)3-5783-2415
1. Huber Group, “Neuronale Netze in Automobilen sollen
NOX-Sensor vor Katalysator in SCR-Systemen ersetzen,”
2009, accessed May 2021, https://www.elektronikpraxis.
2. Huber Group, “Neural Networks in the Automotive
Industry: Huber Group Is Working on Substituting a NOx
Sensor with a Neural Network/Learning Network to Replace
Physical Model,” 2010, accessed August 2021, https://www. l-net works-in-
Downloaded from SAE International, Friday, May 13, 2022
© 2021 Mos er, Kipping, Higu chi, Hirayama; Pu blished by SAE Inte rnational. Thi s Open Access art icle is publishe d under the terms of t he Creative Common s Attribution
License (, which permits distribution, and reproduction in any medium, provided that the original author(s) and the
source are credited.
Positions and opinions advanced in this work are those of the author(s) and not ne cessarily those of SAE International. Respo nsibility for the content of the work lies
solely with the author(s).
ISSN 0148-7191
3. Li, Q., Qiao, F., and Yu, L., “A Machine Learning Approach
for Light-Duty Vehicle Idling Emission Estimation Based on
Real Driving and Environmental Information,” Environ
Pollut Climate Change 1 (2016): 106.
4. BOSCH, “Modeling a Physical System for System Control and
Calibration,” accessed May 2021,
5. BOSCH, “NOx sensor – Regulation and diagnosis of
cataly tic nitrogen oxide reduction,” accessed August 2021,
6. Rudlo, J., Leroy, T., Ngo, C., and Denys, F., “Schätzung der
NOX-Emissionen moderner Dieselmotoren unter realen
Fahrbedingungen,” MTZ- Motortechnische Zeitschri 80,
no. 2 (2019): 58-62.
7. Lumpp, B.C., “Echtzeitfähige Stickoxidmodellierung zur
Integration im Steuergerät eines Nutzfahrzeug-
Dieselmotors,” Technischen Universität München
(Technical University Munich), 2011, http://nbn-resolving.
di ss-2 0111212-1083664 -1-3.
8. Arsie, I., Cricchio, A., De Cesare, M., Lazzarini, F. et al.,
“Neural Network Models for Virtual Sensing of NOx
Emissions in Automotive Diesel Engines with Least Square-
Based Adaptation,” Control Engineering Practice 61 (2017):
11-20 ,
9. Sequenz, H. and Isermann, R., “Emission Model Structures
for an Implementation on Engine Control Units,” IFAC
Proceedings Volumes 44, no. 1 (2011): 11851-11856, https://
doi.or g/10.3182 /2 0110828- 6 -IT-10 02.03131.
10. Cython, “Faster Code via Static Typing,” accessed August
11. Demuynck, J., Favre, C., Bosteels, D., Randlshofer, G. etal.,
AECC, IPA, IAV, “Integrated Diesel System Achieving Ultra-
Low Urban and Motorway NOX Emissions on the Road,” in
Wiener Motorensymposium, Vienna, 2019.
12. Rezaei, R., Dinkelacker, F., Tilch, B., Delebinski, T. et al.,
“Phenomenological Modeling of Combustion and NOX
Emissions Using Detailed Tabulated Chemistry Methods in
Diesel Engines,” International Journal of Engine Research 17
(2016): 846-856.
13. Todo, Y. et al., “Development of High Accuracy and Quick
Light-o NOx Sensor,” SAE Technical Paper 2018 -01-0334 ,
14. Guyon, I. and Elissee, A., “An Introduction to Variable and
Feature Selection,” J. Mach. Learn. Res. 3 (20 03): 1157-1182 .
15. Zheng, A. and Casari, A., Feature Engineering for Machine
Learning: Principles and Techniques for Data Scientists
(Germany: O’Reilly Media, 2018)
16. Scikit-Learn, “User Guide,”
user_guide.html, accessed May 2021.
17. ENVIRON International Corporation, “Humidity and
Temperature Correction Factors for NOx Emissions from
Diesel Engines,” Final Report, 2003.
18. McCormick, R.L., Graboski, M.S., Newlin, A.W., and Ross, J.D.,
“Eect of Humidity on Heavy-Duty Transient Emissions from
Diesel and Natural Gas Engines at High Altitude,” Journal of the
Air & Waste Management Association 47, no. 7 (1997): 784-791,
htt ps:// 473289.1997.10463937.
19. Liu, Y., Wang, Z., Li, L., Wan, K. et al., “Reaction Mechanism
Reduction for Ozone-Enhanced CH4/Air Combustion by a
Combination of Directed Relation Graph with Error
Propagation, Sensitivity Analysis and Quasi-Steady State
Assumption,Energies 11 (2018): 1470, https://doi.
org/10. 3390/en11061470.
Downloaded from SAE International, Friday, May 13, 2022
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
In this study, an 18-steps, 22-species reduced global mechanism for ozone-enhanced CH4/air combustion processes was derived by coupling GRI-Mech 3.0 and a sub-mechanism for ozone decomposition. Three methods, namely, direct relation graphics with error propagation, (DRGRP), sensitivity analysis (SA), and quasi-steady-state assumption (QSSA), were used to downsize the detailed mechanism to the global mechanism. The verification of the accuracy of the skeletal mechanism in predicting the laminar flame speeds and distribution of the critical components showed that that the major species and the laminar flame speeds are well predicted by the skeletal mechanism. However, the pollutant NO was predicated inaccurately due to the precursors for generating NO were removed as redundant components. The laminar flame speeds calculated by the global mechanism fit the experimental data well. The comparisons of simulated results between the detailed mechanism and global mechanism were investigated and showed that the global mechanism could accurately predict the major and intermediate species and significantly reduced the time cost by 72%.
Full-text available
The conventional models for idling emission estimation are mainly based on ambient temperature and the status of vehicle itself, such as vehicle type/size, age and accumulated mileage and fuel type. Instant vehicle activity information is seldom taken into account. In this research, a machine learning approach is proposed to dynamically estimate vehicle emission rates while idling, based on real-world driving tests on more than 1,600 km highways in the State of Texas in the USA. One driver drove a dedicated light-duty gasoline vehicle on various types of roads, including interstate freeways, farm roads, state highways, and arterial road. During each episode of idling, rates of vehicle exhaust emissions, including carbon dioxide (CO2), carbon monoxide (CO), hydrocarbon (HC) and nitrogen oxides (NOx) were measured by a Portable Emission Measurement System (PEMS). Meanwhile, the real-time vehicle engine information of the test vehicle, such as revolutions per min, intake air temperature, and environmental information (e.g. ambient temperature), were collected through the On-board Diagnosis II port. Five machine learning algorithms were applied to build up idling emission models to illustrate the nature of emission patterns. Results show that Boosted and Bagged Decision Trees (BBDT) based idling emission model was identified as the best-fit ones for dynamic idling emissions with better prediction performance.
Full-text available
The Code of Federal Regulations (CFR 40 Part 86 Subpart N) defines a universal humidity correction for NOx emissions from heavy–duty diesel and alternative fueled engines measured by the heavy–duty transient test. This testing procedure and humidity correction apply to any heavy–duty engine subject to regulation of particulate matter emissions, including spark ignited engines. This correction has been evaluated for a 1988 Detroit Diesel Series 60 engine and a 1995 Cummins B5.9G natural gas engine. The correction at high altitude for the diesel engine is in excellent agreement with the correction published in the Code of Federal Regulations. In addition, humidity is found to affect particulate matter in agreement with the Engine Manufacturers Association correction factor. This suggests that these results, acquired at high altitude, are generally applicable to all altitudes. CO emissions are also correlated with humidity. Emission traces show that humidity affects NOx uniformly for the diesel engine. A modification to the CFR correction method to allow accounting for the effect of humidity changes during the transient test on NOx emissions is proposed. The NOx correction factor for the natural gas engine is much more significant than for the diesel engine. Total hydrocarbon, principally methane, also correlates with humidity. The development of a natural gas correction, rather than the CFR diesel correction, for NOx and hydrocarbon emissions is suggested.
Full-text available
Variable and feature selection have become the focus of much research in areas of application for which datasets with tells or hundreds of thousands of variables are available. These areas include text processing of internet documents, gene expression array analysis, and combinatorial chemistry. The objective of variable selection is three-fold: improving the prediction performance of the predictors, providing faster and more cost-effective predictors, and providing a better understanding of the underlying process that generated the data. The contributions of this special issue cover a wide range of aspects of such problems: providing a better definition of the objective function, feature construction, feature ranking, multivariate feature selection, efficient search methods, and feature validity assessment methods.
To meet current Diesel engine pollutant legislation, it is important to manage after-treatment devices. The paper describes the development of Neural Network based virtual sensors used to estimate NOx emissions at the exhaust of automotive Diesel engines. Suitable identification methodologies and experimental tests were developed with the aim of meeting the conflicting needs of feasible on-board implementation and satisfactory prediction accuracy. In addition, since the prediction of control-oriented models is typically affected by engine aging and production spread as well as components drift, least square technique features were exploited in order to overcome these issues by adapting the virtual sensor output. The NOx adaptive virtual sensor was tested via comparison with experimental data, measured at the engine test bench on a turbocharged common-rail automotive Diesel engine. Furthermore, besides model validation, the experimental measurements were modified to simulate a sensor drift in order to enable full assessment of the proposed LS-based algorithm adaptation capabilities.
Enhancing the predictive quality of engine models, while maintaining an affordable computational cost, is of great importance. In this study, a phenomenological combustion and a tabulated NOx model, focusing on efficient modeling and improvement of computational effort, is presented. The proposed approach employs physical and chemical sub-models for local processes such as injection, spray formation, ignition, combustion, and NOx formation, being based on detailed tabulated chemistry methods. The applied combustion model accounts for the turbulence-controlled as well as the chemistry-controlled combustion. The phenomenological combustion model is first assessed for passenger car application, especially with multiple pilot injections and high exhaust gas recirculation ratios for low-load operating points. The validation results are presented for representative operating conditions from a single-cylinder light-duty diesel engine and over the entire engine map of a heavy-duty diesel engine. In the second part of this study, a novel approach for accurate and very fast modeling of NO formation in combustion engines is proposed. The major focus of this study is on the development of a very fast-running NO mechanism for usage in the next generation of the engine control units. This approach is based on tabulation of a detailed chemical kinetic mechanism and is validated against the detailed chemical reaction mechanism at all engine-relevant conditions with the variation in pressure, temperature, and air–fuel ratio under stationary and ramp-type transient conditions in a perfectly stirred reactor. Using this approach, a very good match to the results from calculations with the detailed chemical mechanism is observed. Finally, the tabulated NOx kinetic model is implemented in the combustion model for in-cylinder NOx prediction and compared with the experimental engine measurement data.
Conference Paper
The number of model-based approaches in modern engine control units (ECUs) increases permanently with the increase of engine complexity. Therefore the efficient storage and the fast computation of these models is a challenging task. Due to their fast computation properties most ECUs utilize 2-dimensional grid map structures. Higher dimensional relations need then to be mapped with nested structures of several 2-dimensional grid maps. The determination of suited structures is a time-consuming task and the number of utilized 2-dimensional grid maps may be large for higher dimensional relations. Therefore in the following a grid map structure, a LOLIMOT structure, a Kernel method and an adaptive polynomial approach are compared with respect to model accuracy, computation time and required memory. Conclusions about the implementation on state of the art ECUs are given. As model output the NOx emissions of a DTH Z19 Opel CR-Diesel engine with VGT-turbocharger and exhaust gas recirculation are regarded. Model structures are presented for a global model approach comprising the engine operation point as input, and for a global-local approach predicting the output by weighting local models.
Neuronale Netze in Automobilen sollen NOX-Sensor vor Katalysator in SCR-Systemen ersetzen
  • Huber Group
Huber Group, "Neuronale Netze in Automobilen sollen NOX-Sensor vor Katalysator in SCR-Systemen ersetzen," 2009, accessed May 2021, https://www.elektronikpraxis.