- Home
- Apostolos Kotsialos

# Apostolos Kotsialos

25.95

· PhD, MSc, Dipl.-Eng.About

81

Research items

9,745

Reads

2,641

Citations

Introduction

**Skills and Expertise**

Research Experience

Sep 2016 - Aug 2017

Sep 2005 - Aug 2016

Network

Cited

Followers

Following

Projects

Projects (1)

Project

Research

Research Items (81)

- May 2019
- HELORS 8th International Symposium & 30th National Conference on Operational Research

This paper builds on the earlier work of Kotsialos (2013, 2014), where the Resilient BackPROPagation (RPROP) algorithm is recast as a search direction and step selection method for solving smooth and non-smooth, unconstrained and simply-bounded nonlinear optimisation problems. Although RPROP was originally proposed as a learning algorithm for neural networks training, it has been shown that it is competitive and sometimes can outperform highly sophisticated algorithms, especially for solving non-smooth problems. The simplicity of implementation, the relatively low computation effort, as only one function and one gradient evaluation are required per iteration, and the good convergence properties make RPROP a highly efficient algorithm for large scale problems. RPROP can tolerate errors in the gradient evaluation, since it is based on the partial derivatives' sign rather than their values, allowing its application to non-smooth Lipschitz continuous objective functions. The constrained optimisation problems considered here are transformed to unconstrained problems by use of exact ℓ1 as well as quadratic penalty functions and combinations thereof. RPROP is applied to the unconstrained problems within a simple multi-start framework for improving convergence. Penalty-barrier as well as augmented Lagrangian methods may be considered in this setting as well. A set of benchmark academic problems available from the literature are used for demonstrating and evaluating the overall algorithm's performance. Detailed computational results and the impact of the main RPROP parameters, the directional step increase rate and the restart frequency, are reported. Hence, this paper extents the investigations reported in Kotsialos (2013, 2014) for nonlinear constrained problems.

- Jul 2018
- 2018 IEEE Congress on Evolutionary Computation (CEC)

- Nov 2017
- 2017 IEEE International Conference on Rebooting Computing (ICRC)

- Aug 2017

This paper is concerned with the problem of macroscopic traffic flow model validation for ring-road shaped large-scale motorway networks. The calibration optimization problem is solved by a gradient-based algorithm combining into a single software package METANET (traffic simulator), RPROP (resilient backpropagation search heuristic), and ADOL-C (automatic differentiation library) and by a separate implementation of particle swarm optimization with METANET. These model validation packages are applied to the motorway network around the city of Manchester, U.K. The total road length of the site is 186 km considering the traffic flow on both directions of each modeled motorway. For this large scale network, a single optimization problem is formed for calibrating METANET, i.e., for identifying its parameters. Three different data sets are used and the corresponding optimal parameter sets are obtained. The results show that the combined METANET-RPROP-ADOL-C package is able to calibrate the large scale motorway network model with very good accuracy. The optimal parameter sets, where the optimization algorithm converged for a particular data set, are verified by running METANET simulations using the other two data sets. Results do show the expected degradation of the parameter set's quality, but the essential network wide dynamics of congestion are retained.

This paper presents a series of experiments demonstrating the capacity of single-walled carbon-nanotube (SWCNT)/liquid crystal (LC) mixtures to be trained by evolutionary algorithms to act as classifiers on linear and nonlinear binary datasets. The training process is formulated as an optimisation problem with hardware in the loop. The liquid SWCNT/LC samples used here are un-configured and with nonlinear current-voltage relationship, thus presenting a potential for being evolved. The nature of the problem means that derivative-free stochastic search algorithms are required. Results presented here are based on differential evolution (DE) and particle swarm opti-misation (PSO). Further investigations using DE, suggest that a SWCNT/LC material is capable of being reconfigured for different binary classification problems, corroborating previous research. In addition, it is able to retain a physical memory of each of the solutions to the problems it has been trained to solve.

Question - Convert a Constrained Optimal Control Problem to an Optimization Problem?

- Jan 2017

Answer

You may also want to check Betts' book

…

Macroscopic traffic flow model calibration is an optimisation problem typically solved by a derivative-free population based stochastic search methods. This paper reports on the use of a gradient based algorithm using automatic differentiation. The ADOL-C library is coupled with the METANET source code and this system is embedded within an optimisation algorithm based on RPROP. The result is a very efficient system which is able to be calibrate METANET’s second order model by determining the density and speed equation parameters as well as the fundamental diagrams used. Information obtained from the system’s Jacobian provides extra insight into the system dynamics. A 22 km site is considered near Sheffield, UK and the results of a typical calibration and validation process are reported.

The problem of validating the Modéle d’Écoulement de Trafic sur Autoroute NETworks (METANET) model of a motorway section is considered. Model calibration is formulated as a least squares error minimisation problem with explicit penalisation of fundamental diagram parameter variation. The Automatic Differentiation by Overloading in C++ (ADOL-C) library is incorporated into the METANET source code and is coupled with the Resilient Back Propagation (RPROP) heuristic for solving the minimisation problem. The result is a very efficient system which is able to be calibrate METANET by determining the density and speed equation parameters as well as the fundamental diagrams used. Information obtained from the system's Jacobian provides extra insight into the dynamics showing how sensitivities propagate into the network. A 22 km site near Sheffield, UK, using data from three different days is considered. In addition to the ADOL-C/RPROP system, three particle swarm optimisation algorithms are used for solving the calibration problem. In all cases, the optimal parameter sets found are verified on data not used during calibration. Although, all three sets of data display a similar congestion pattern, the verification process showed that only one of them is capable of leading to parameter sets that capture the underlying dynamics of the traffic flow process.

- Sep 2016
- International Conference on Parallel Problem Solving from Nature

The potential of Evolution in Materio (EiM) for machine learning problems is explored here. This technique makes use of evolutionary algorithms (EAs) to influence the processing abilities of an un-configured physically rich medium, via exploitation of its physical properties. The EiM results reported are obtained using particle swarm optimisation (PSO) and differential evolution (DE) to exploit the complex voltage/current relationship of a mixture of single walled carbon nanotubes (SWCNTs) and liquid crystals (LCs). The computational problem considered is simple binary data classification. Results presented are consistent and reproducible. The evolutionary process based on EAs has the capacity to evolve the material to a state where data classification can be performed. Finally, it appears that through the use of smooth signal inputs, PSO produces classifiers out of the SWCNT/LC substrate which generalise better than those evolved with DE.

Evolution-in-materio concerns the computer controlled manipulation of material systems using external stimuli to train or evolve the material to perform a useful function. In this paper we demonstrate the evolution of a disordered composite material, using voltages as the external stimuli, into a form where a simple computational problem can be solved. The material consists of single-walled carbon nanotubes suspended in liquid crystal; the nanotubes act as a conductive network, with the liquid crystal providing a host medium to allow the conductive network to reorganise when voltages are applied. We show that the application of electric fields under computer control results in a significant change in the material morphology, favouring the solution to a classification task.

Evolution In Materio (EIM) is concerned with solv- ing computational problems by exploiting the physical prop- erties of materials. This paper presents the results of using a particle swarm optimisation (PSO) algorithm for evolving logic circuits in single-walled carbon nanotubes (SWCNT) based composites on a special custom made platform. The material used is a composite of SWCNT dispersed randomly in a polymer forming a complex conductive network. Follow- ing the EIM methodology the conductance of the material is manipulated for evolving threshold based logic circuits. The problem is formulated as a constrained, mixed integer optimisation problem. It is solved using PSO in conjunction with the shortest position value rule. The results showed that the conductive properties of SWCNT can be used to configure these materials to evolve multiple input/ output logic circuits.

- Jul 2016
- International Conference on Unconventional Computation and Natural Computation

Evolution-in-Materio uses evolutionary algorithms (EA) to exploit the physical properties of unconfigured, physically rich materials, in effect transforming them into information processors. The potential of this technique for machine learning problems is explored here. Results are obtained from a mixture of single walled carbon nanotubes and liquid crystals (SWCNT/LC). The complex nature of the voltage/current relationship of this material presents a potential for adaptation. Here, it is used as a computational medium evolved by two derivative-free, population-based stochastic search algorithms, particle swarm optimisation (PSO) and differential evolution (DE). The computational problem considered is data classification. A custom made electronic motherboard for interacting with the material has been developed, which allows the application of control signals on the material body. Starting with a simple binary classification problem of separable data, the material is trained with an error minimisation objective for both algorithms. Subsequently, the solution, defined as the combination of the material itself and optimal inputs, is verified and results are reported. The evolution process based on EAs has the capacity to evolve the material to a state where data classification can be performed. PSO outperforms DE in terms of results’ reproducibility due to the smoother, as opposed to more noisy, inputs applied on the material.

Macroscopic traffic flow model calibration is an optimisation problem typically solved by a derivative-free population based stochastic search methods. This paper reports on the use of a gradient based algorithm using automatic differentiation. The ADOL-C library is coupled with the METANET source code and this system is embedded within an optimisation algorithm based on RPROP. The result is a very efficient system which is able to be calibrate METANET's second order model by determining the density and speed equation parameters as well as the fundamental diagrams used. Information obtained from the system's Jacobian provides extra insight into the system dynamics. A 22 km site is considered near Sheffield, UK and the results of a typical calibration and validation process are reported.

- May 2016
- Autonomic Road Transport Support Systems

One of the most persistent problems that plague modern-day road transport facilities is the quality of service provided. Especially during rush hours, this expensive infrastructure does not operate at capacity nor does it provide the level of service required by its users. Congestion has become a problem with severe economic and environmental repercussions. Hence, efficient road traffic management is more important than ever.

- May 2016
- Autonomic Road Transport Support Systems

This chapter discusses a systems design approach inspired from the autonomic nervous system for intelligent transportation system (ITS) applications. This is done not with reference to the employed computing system but with reference to the requirements of traffic engineering applications. It is argued that the design and development of autonomic traffic management systems must identify the control loop that needs to be endowed with autonomic properties and subsequently use this framework for defining a desired set of self-∗ properties. A macroscopic network modelling application is considered for showing how autonomic system design can be used for defining and obtaining self-∗ properties, with particular emphasis given on self-optimisation. The interpretation of policies followed by network operators regarding route guidance is also discussed from the perspective of autonomic ITS.

- Jan 2016
- Birkhäuser Basel

Question - How can I use the differential evolution technique in matlab to solve multiple differential equations?

- Nov 2015

Answer

Given the way you phrase it, I would say no. You may need to solve systems of ODEs and employ DE on top of that for problems like parameter identification or optimal control; this is something different from solving the ODEs themselves. Using DE as a solving method for systems of ODEs does not make much sense. I would suggest to put down the mathematical problem formulation you need to solve and then get an appreciation of what methods are available.

…

Question - How can I use the differential evolution technique in matlab to solve multiple differential equations?

- Nov 2015

Answer

Differential evolution is a stochastic search algorithm used for optimising an objective function(s); it is not a solver of differential equations. These are two different computational tasks.

…

Workshop III: Traffic Control (Schedule) - IPAM

Purpose
A process improvement sampling methodology, known as Process Variation Diagnostic Tool
(PROVADT), was proposed by Cox et al (2013). The method was designed to support the
objectivity of Six Sigma projects performing the Measure-Analyse phases of the Define-
Measure-Analyse-Improve-Control (DMAIC) cycle. An issue in PROVADT is that it is
unable to distinguish between measurement and product variation in the presence of a poor
Gage R&R result. In this paper PROVADT’s sampling structure is improved and addresses
this issue by enabling a true Gage R&R as part of its design.
Design/methodology/approach
This paper derives an enhanced PROVADT method by examining the theoretical sampling
constraints required to perform a Gage R&R study. The original PROVADT method is then
extended to fulfil these requirements. To test this enhanced approach, it was applied first to a
simulated manufacturing process and then in two industry case studies.
Findings
The results in this paper demonstrate that enhanced PROVADT was able to achieve a full
Gage R&R result. This required twenty additional measurements when compared to the
original method, but saved up to ten additional products and twenty additional measurements
being taken in future experiments if the original method failed to obtain a valid Gage R&R.
These benefits were highlighted in simulation and industry case studies.
Originality/value
The work into the PROVADT method aims to improve the objectivity of early Six Sigma
analyses of quality issues, which has documented issues.

- Mar 2015
- International Conference on Numerical Analysis and Applied Mathematics 2014 (ICNAAM-2014)
- INTERNATIONAL CONFERENCE OF NUMERICAL ANALYSIS AND APPLIED MATHEMATICS

We demonstrate the manipulation of single-walled carbon nanotube/liquid crystal composites using in-plane electric fields. The conductivity of the materials is shown to be dependant on the application of a DC bias across the electrodes. When the materials are subjected to this in-plane field, it is suggested that the liquid crystals orientate, thereby forcing the SWCNTs to follow in alignment. This process occurs over many seconds, since the SWCNTs are significantly larger in size than the liquid crystals. The opportunity for applying this material to unconventional computing problems is suggested.

This paper explores the use of single-walled carbon nanotube (SWCNT)/poly(butyl methacrylate) composites as a material for use in unconventional computing. The mechanical and electrical properties of the materials are investigated. The resulting data reveal a correlation between the SWCNT concentration/viscosity/conductivity and the computational capability of the composite. The viscosity increases significantly with the addition of SWCNTs to the polymer, mechanically reinforcing the host material and changing the electrical properties of the composite. The electrical conduction is found to depend strongly on the nanotube concentration; Poole-Frenkel conduction appears to dominate the conductivity at very low concentrations (0.11 % by weight). The viscosity and conductivity both show a threshold point around 1 % SWCNT concentration; this value is shown to be related to the computational performance of the material. A simple optimization of threshold logic gates shows that satisfactory computation is only achieved above a SWCNT concentration of 1 %. In addition, there is some evidence that further above this threshold the computational efficiency begins to decrease.

- Feb 2015

We report on the use of a liquid crystalline host medium to align single-walled carbon nanotubes (SWCNTs) in an electric field using an in-plane electrode configuration. Electron microscopy reveals that the nanotubes orient in the field with a resulting increase in the DC conductivity in the field direction. Current versus voltage measurements on the composite show a nonlinear behavior, which was modelled by using single-carrier space-charge injection. The possibility of manipulating the conductivity pathways in the same sample by applying the electrical field in different (in-plane) directions has also been demonstrated. Raman spectroscopy indicates that there is an interaction between the nanotubes and the host liquid crystal molecules that goes beyond that of simple physical mixing.

Question - How can I use/link the MATLAB optimization toolbox for my optimization problem?

- Feb 2015

Answer

The toolbox comes along with some very nice and informative help files. For every solver included you can find the problem formulation and the details of the specific inputs. They are fairly explanatory, You will need to write a couple of scripts to provide the evaluation of the objective function and analytical expressions of the derivatives if you don't want to use the option given for approximating from the solver. You will need to determine the category of problem you have (linear, nonlinear, constrained, unconstrained, smooth, nonsmooth, convex, nonconvex, integer etc) and then select the solver.

…

A practical control chart is introduce, called multivariate Set-Up Process Algorithm (m-SUPA), which can be used to signal when a process is statistically off-target. This control chart uses a traffic light system to provide simple information to an operator about how close a measured part is to its global target. The chart works with a simple rule set resulting in process adjustments at a calculated point, rather than relying on rule-of-thumb methods. A final consideration is calculating the size of process adjustment, when one control adjustment has multiple effects on different design features. Simple feedback controllers are suggested for calculating process adjustments, providing consistency to an action taken. Simulation results suggest that m-SUPA with adjustments based on this kind of controllers is able to steer the process to a desired performance region.

A search method based on the backpropagation rule commonly used for training neural networks is proposed here for the optimisation of smooth nonlinear functions. The use of the Resilient backPROPagation (RPROP) heuristic rule for local minimisation is described. The details of employing the directional step length determined by RPROP along with a simple restarting scheme are provided. In the approach proposed here direct use of the directional step determined by the heuristic without using any line search conditions takes place. The overall algorithm has been tested on a number of benchmark functions found in the literature with very positive results. The test problems’ dimension ranges from 100 to 50,000. The results obtained show that the suggested search direction method results to a highly efficient algorithm suitable for large scale optimisation.

This paper presents results of computations based on threshold logic performed by a thin solid film, following the general principle of evolution in materio. The electrical conductivity is used as the physical property manipulated for evolving Boolean functions. The material used consists of a composite of single-wall carbon nanotubes (SWCNTs) and the polymer poly(methyl methacrylate). The SWCNTs are randomly dispersed in the polymer forming a complex conductive network at the nano-scale. The training is formulated as an optimisation problem with continuous and binary constraints and is subsequently solved by two derivative-free algorithms, the Nelder-Mead (NM) and the Differential Evolution (DE) algorithms. This approach has been used to evolve gates and circuits. The NM fails to converge for all computational tasks, whereas the DE is always successful. The computation tasks considered are simple threshold logic gates and more complicated circuits. The thin film composite is very stable and its behavior remains the same after the optimal solution has been achieved.

- Nov 2013
- 2nd International Through-life Engineering Services Conference

The main focus of this paper is to use discrete-event simulation models, to test the robustness of two process control methods against processes with different statistical distributions. The two methods under scrutiny are the Small-Batch X & R chart and the Set-Up Process Algorithm (SUPA). These have been developed for ‘setup dominant processes’, were the major source of product variation is detected between batches. Minimizing this type of variation is critical to ensure spare parts produced at a later date will fit in operating assemblies, maintaining a Through-life Engineering Service. This paper shows their suitability to industry.

Purpose ‐ The purpose of this paper is to examine the efficiency and objectivity of current Six Sigma practices when at the measure/analyse phase of the DMAIC quality improvement cycle. Design/methodology/approach ‐ A new method, named process variation diagnostic tool (PROVADT), demonstrates how tools from other quality disciplines can be used within the Six Sigma framework to strengthen the overall approach by means of improved objectivity and efficient selection of samples. Findings ‐ From a structured sample of 20 products, PROVADT was able to apply a Gage R&R and provisional process capability study fulfilling the pre-requisites of the measure and early analyse phases of the DMAIC quality improvement cycle. From the same sample, Shainin multi-vari and isoplot studies were conducted in order to further the analysis without the need of additional samples. Practical implications ‐ The method was tested in three different industrial situations. In all cases PROVADT's effectiveness was shown at driving forward a quality initiative with a relatively small number of samples. Particularly in the third case, it lead to the resolution of a long standing complex quality problem without the need for active experimentation on the process. Originality/value ‐ This work demonstrates the need to provide industry with new statistical tools which are practical and give users efficient insight into potential causes of a process problem. PROVADT makes use of data needed by quality standards and Six Sigma initiatives to fulfil their requirements but structures data collection in a novel way to gain more information.

- Oct 2013
- 2013 16th International IEEE Conference on Intelligent Transportation Systems - (ITSC 2013)

This paper discusses a systems design approach inspired from the autonomic nervous system for ITS applications. This is done not with reference to the employed computing system, but to the requirements of traffic engineering applications. It is argued that the design and development of autonomic traffic management systems must identify the control loop that needs to be endowed with autonomic properties and subsequently use this framework for defining a desired set of self-* properties. A macroscopic network modelling application is considered for showing how autonomic systems design can be used for defining and obtaining self-* properties, with particular emphasis given in self-optimisation.

This paper presents a new method of process control for set-up dominant processes. This new method known as Set-up Process Algorithm (SUPA) was compared with existing industrial practices and statistical techniques in the literature. To test the method’s robustness, a generic discrete-event simulation model was built. This model was used to test four different statistical approaches to process control. It was concluded that SUPA offers a method of process control for set-up dominant processes, which is easier to apply than classically derived SPC approaches, by using simple rules and a traffic light system based on design specification. Simulation analysis shows that SUPA: is more sensitive, at detecting an incapable process as it will monitor more units when a process is less capable; is more sensitive than PRE-Control at detecting mean shifts in a process. SUPA is also a nonparametric methodology and therefore robust against processes with non-Gaussian distributions.

Deterministic model validation with a new approach to fundamental diagrams of the METANET model Overview Macroscopic traffic models are highly non-linear, as such validating these models is a difficult task. Validation is required for numerous reasons: Evaluation of control strategies Comparing traffic models as no definitive theory [1] Model validation has typically been done on an Ad-hoc basis or limited to theoretical problems and data. The state-of-the-art is advancing towards deterministic methods that can be applied to numerous traffic models and are abstract from the model code. In an attempt to further advance the autonomy of the validation procedure we introduce a method to dynamically determine the placement and composition of the fundamental diagrams within the model. Macroscopic Traffic Models Here we are looking at the calibration and verification of the METANET model [2]. Although, the algorithm utilises the model as a black box and as such other models can easily be substituted in METANET's place. Macroscopic models contain fundamental diagrams which describe the highway characteristics. In first order models this is used directly to calculate the velocity of the flow, in higher order models it is given as a relaxation term. These characteristics can change along a models length depending on the topology of the highway. The positioning and set up of these change points, if any, is usually done by the use of expert knowledge and not necessarily obvious from data observations.

- Jan 2013
- 11th International Conference on Manufacturing Research (ICMR2013)

This paper presents a new method of process control for set-up dominant processes. This new method known as Set-up Process Algorithm (SUPA) was compared with existing industrial practices and statistical techniques in the literature. To test the method's robustness, a generic discrete-event simulation model was built. This model was used to test four different statistical approaches to process control. It was concluded that SUPA offers a method of process control for set-up dominant processes, which is easier to apply than classically derived SPC approaches, by using simple rules and a traffic light system based on design specification. Simulation analysis shows that SUPA: is more sensitive, at detecting an incapable process as it will monitor more units when a process is less capable; is more sensitive than PRE-Control at detecting mean shifts in a process. SUPA is also a nonparametric methodology and therefore robust against processes with non-Gaussian distributions.

- Jan 2013
- Procedia CIRP, 2nd International Through-life Engineering Services Conference

The main focus of this paper is to use discrete-event simulation models, to test the robustness of two process control methods against processes with different statistical distributions. The two methods under scrutiny are the Small-Batch X ¯ & R chart and the Set-Up Process Algorithm (SUPA). These have been developed for 'setup dominant processes', were the major source of product variation is detected between batches. Minimizing this type of variation is critical to ensure spare parts produced at a later date will fit in operating assemblies, maintaining a Through-life Engineering Service. This paper shows their suitability to industry. © 2013 The Authors. Published by Elsevier B.V. Selection and peer-review under responsibility of the International Scientific Committee of the "2 nd International Through-life Engineering Services Conference" and the Programme Chair – Ashutosh Tiwari.

This paper seeks to highlight often overlooked techniques of process control, Pre-Control, Pairs and Acceptance Control Charts, which are becoming increasingly relevant in the context decreasingly small production runs and an emphasis on high precision and high value product. A methodology is built on these techniques, known as Set-up Process Algorithm (SUPA), to assist operators make adjustments to Set-up dominant processes in a statistically valid framework. The performance of these methods is then assessed analytically and through simulation to demonstrate their respective sensitivity at detecting process deviations from the specified target.

This paper examines the efficiency and objectivity of current Six Sigma practices when at the Measure/Analyse phase of the DMAIC process improvement cycle. A method, known as the Process Variation Diagnosis Tool (PROVADT), is introduced to demonstrate how tools from other quality disciplines can be used within the Six Sigma framework to strengthen the overall approach by means of improved objectivity and efficient selection of samples. From a structured sample of products, PROVADT is able to apply a Gage R&R and Provisional Process Capability study fulfilling pre-requisites of the Measure and early Analyse phases of the DMAIC process improvement cycle. From the same samples a Shainin Multi-Vari study and Isoplot can be obtained in order to further the analysis without additional samples. The latter quality techniques are associated with the "Clue Generation" phase of the Shainin System. The PROVADT method is tested in industry case studies to demonstrate its effectiveness of driving forward a process improvement initiative with a relatively small number of samples, which is particularly important for low volume high value manufacturing. Case studies were conducted at a leading manufacturer of microprocessor based electric motor control systems, a global technology, manufacturing and service company that provide advanced systems in the automotive industry and a furniture manufacturer. Using PROVADT and sample sizes of 20 units it was possible in all cases to validate the measurement system and gain an early objective insight into potential root causes of variation, leading to significant cost savings for both companies.

This paper describes a non-smooth optimization method based on a backpropagation search method. More specifically, the Resilient backPROPagation (RPROP) algorithm, used extensively for neural network training, is employed here. RPROP is recast in terms of numerical optimization and used as a step-finding method. The result is the fast search algorithm RPROP, which avoids expensive line searches and performs one function and one gradient evaluation per iteration. Furthermore, only the gradient's sign is used rather than its value. RPROP is applied to a set of 10 test problems, which have unconstrained and simply bounded constrained versions. The results are discussed and assessed against a set of reported results based on bundle methods. It is shown that RPROP is able to deal efficiently with very large non-smooth problems.

This paper examines the efficiency and objectivity of current Six Sigma practices when at the Measure/Analyse phase of the DMAIC process improvement cycle. A method, known as the Process Variation Diagnosis Tool (PROVADT), is introduced to demonstrate how tools from other quality disciplines can be used within the Six Sigma framework to strengthen the overall approach by means of improved objectivity and efficient selection of samples. From a structured sample of products, PROVADT is able to apply a Gage R&R and Provisional Process Capability study fulfilling pre- requisites of the Measure and early Analyse phases of the DMAIC process improvement cycle. From the same samples a Shainin Multi-Vari study and Isoplot can be obtained in order to further the analysis without additional samples. The latter quality techniques are associated with the “Clue Generation” phase of the Shainin System. The PROVADT method is tested in industry case studies to demonstrate its effectiveness of driving forward a process improvement initiative with a relatively small number of samples, which is particularly important for low volume high value manufacturing. Case studies were conducted at a leading manufacturer of microprocessor based electric motor control systems, a global technology, manufacturing and service company that provide advanced systems in the automotive industry and a furniture manufacturer. Using PROVADT and sample sizes of 20 units it was possible in all cases to validate the measurement system and gain an early objective insight into potential root causes of variation, leading to significant cost savings for both companies.

A nonlinear model-predictive hierarchical control approach is presented for coordinated ramp metering of freeway networks. The utilized hierarchical structure consists of three layers: the estimation/prediction layer, the optimization layer and the direct control layer. The previously designed optimal control tool AMOC (Advanced Motorway Optimal Control) is incorporated in the second layer while the local feedback control strategy ALINEA is used in the third layer. Simulation results are presented for the Amsterdam ring-road. The proposed approach outperforms uncoordinated local ramp metering and its efficiency approaches the one obtained by an optimal open-loop solution. It is demonstrated that metering of all on-ramps, including freeway-to-freeway intersections, with sufficient ramp storage space leads to the optimal utilization of the available infrastructure.

This paper is concerned with a fluid approach towards modelling of production networks. It focuses on developing a macroscopic
modelling framework that is able to describe the average dynamics of the manufactured items’ movement in a production network
taking under consideration possible non-linearities that occur due to the operation and interaction of workcells. The notion
of clearing functions is used in a similar manner as the fundamental diagram of traffic engineering is used for modelling
vehicular traffic flow. The system dynamics are represented by the macroscopic variables of items’ density, flow, speed and
queue length, which are associated through the use of a conservation law. These variables are further differentiated with
respect to the customer order they belong to, providing this way a more detailed description of system dynamics. The suggested
approach is a computationally efficient scheme that can be used for modelling and control purposes in a manner similar to
macroscopic freeway traffic flow models.

- Sep 2007
- 4th International Conference on Digital Enterprise Technology

This paper is concerned with a fluid approach towards modelling of production networks. It focuses on developing a macroscopic modelling framework that is able to describe the average dynamics of the manufactured items' movement in a production network taking under consideration possible non-linearities that occur due to the operations and interactions between workcells. The notion of clearing functions is used in a similar manner as the fundamental diagram of traffic engineering is used for modelling vehicular traffic flow. The system dynamics are represented by the macroscopic variables, which are associated through the use of a conservation law. Three different scenarios are presented where serial and parallel manufacturing lines are considered. The suggested approach is a computationally efficient scheme that can be used for modelling and control purposes in a manner similar to macroscopic motorway traffic flow models.

- Jan 2007
- Modélisation, Information et Contrôle dans les Systèmes de Transports Intelligents

One of the most persistent and potentially far reaching, in terms of industrial benefits, problem in new product development, is the parallel and synchronous design and evaluation of the product, the production processes and the production network. The proposed theoretical framework for collaborative design and production network development is based on the concept of Digital Enterprise Technology (DET) and facilitates the integration of design and resource aware planning with aspects of network design and logistics. The controlling cycle of the framework is the DET-enabled, human centric evaluation of products, plans and network configurations that gives rise to an emergent synthesis environment. Early testing using aerospace products has been very encouraging.

Ramp metering, if properly applied, is a direct and efficient means to avoid or reduce the space-time extent of motorway congestion and sensibly to improve the merging conditions. Regardless of the ramp metering algorithm employed, the metering signals may be operated in various ways based on the ramp metering policy adopted. Ramp metering policies include traffic signal cycle, 2- or n-cars per green, and discrete release rates. In the latter policy, a number of discrete release rates are prespecified, each implemented with a specific cycle and green phase. This approach allows for short green phases (small platoon releases) whenever possible but also for high ramp flows when necessary. We address the problem of determining the lowest required number of release rates that will not affect ramp metering operation compared with the theoretical case of any (even decimal) release rate. Results from investigations using the ALINEA (Asservissement Linéaire d'Entrée Autoroutière) ramp metering algorithm and the METANET (Modèle d'Ecoulement du Trafic Autoroutier: NETwork) macroscopic traffic simulator are reported and discussed in detail. Finally, recommendations are provided concerning the lowest required number of release rates and the discretisation scheme to be used.

A nonlinear rolling-horizon model-predictive hierarchical coordinated ramp metering scheme is presented. The hierarchical control structure consists of three layers: the estimation/prediction layer, the optimization layer and the direct control layer. The second layer incorporates the previously designed optimal control tool AMOC while the local feedback strategy ALINEA is used in the third layer. Simulation results are presented for the Amsterdam ring-road. It is shown that control of all on-ramps including freeway intersections leads to the optimal utilization of the available infrastructure.

In this article a nonlinear model predictive control approach to the problem of coordinated ramp metering is presented. The
previously designed optimal control tool Advanced Motorway Optimal Control (AMOC) is used within the framework of a
hierarchical control structure which consists of three basic layers: the estimation/prediction layer, the optimization layer,
and the direct control layer. More emphasis is given to the last two layers where the control actions on a network-wide and
on a local level, respectively, are decided. The hierarchical control strategy combines AMOC’s coordinated ramp metering
control with local feedback Asservissement LIn´eaire d’Entr´e Autorouti`ere (ALINEA) control in an efficient way. Simulation
investigations for the Amsterdam ring-road are reported whereby the results are compared with those obtained by applying
ALINEA as a stand-alone strategy. It is shown that the proposed control scheme is efficient, fair, and real-time feasible.

The problem of medium to long-term sales forecasting raises a number of requirements that must be suitably addressed in the design of the employed forecasting methods. These include long forecasting horizons (up to 52 periods ahead), a high number of quantities to be forecasted, which limits the possibility of human intervention, frequent introduction of new articles (for which no past sales are available for parameter calibration) and withdrawal of running articles. The problem has been tackled by use of a damped-trend Holt-Winters method as well as feedforward multilayer neural networks (FMNNs) applied to sales data from two German companies. Copyright © 2005 John Wiley & Sons, Ltd.

- Jul 2005
- American Control Conference, 2005. Proceedings of the 2005

A nonlinear model-predictive control approach to the problem of coordinated ramp metering is presented. The previously designed optimal control tool AMOC is used within the framework of a hierarchical control structure which consists of three control layers: the estimation/prediction layer, the optimization layer and the direct control layer. More emphasis is given to the last two layers where the control actions on a network-wide and on a local level, respectively, are decided. The hierarchical control strategy combines AMOC's coordinated ramp metering control with local feedback (ALINEA) control in an efficient way. Simulation investigations for the Amsterdam ring-road are reported whereby the results are compared with those obtained by applying ALINEA as a stand-alone strategy. It is demonstrated that the proposed control scheme is efficient, fair and real-time feasible.

The problem of medium to long term sales forecasting raises a number of re- quirements that must be suitably addressed in the design of the employed forecasting methods. These include long forecasting horizons (up to 52 pe riods ahead), a high num- ber of quantities to be forecasted, which limits the po ssibility of human intervention, as well as frequent introduction of new articles (for whic h no past sales are available for pa- rameter calibration) and withdrawal of running articles . The problem has been tackled by use of a modified Holt-Winters method as well as Feedfor ward Multilayer Neural Net- works (FMNN) applied to sales data from two German companies. Copyright © 2005 IFAC

- Jan 2005
- Advances in Control, Communication Networks, and Transportation Systems

Recurrent and non-recurrent congestions on freeways may be substantially reduced if today’s “spontaneous” infrastructure utilisation
is replaced by an orderly, controllable operation via comprehensive application of ramp metering and freeway-to-freeway control,
combined with powerful optimal control techniques. This chapter first explains why ramp metering can lead to a dramatic amelioration
of traffic conditions on freeways. Subsequently, a large-scale example demonstrates the high potential of advanced ramp metering
approaches. It is demonstrated that the proposed control scheme is efficient, fair and real-time feasible.

The Advanced Motorway Optimal Control (AMOC) strategy for optimal freeway network-wide ramp metering is applied to the ring-road of Amsterdam, The Netherlands, in the aim of investigating some important and interesting problems arising in ubiquitous ramp metering systems. A number of suitably chosen scenarios along with a thorough analysis, interpretation, and suitable visualization of the obtained results provide a basis for the better understanding of some complex interrelationships of competing performance criteria. More precisely, the strategy’s efficiency and equity properties as well as their trade-off are studied and their partially competitive behaviour is discussed. This trade-off is implicitly addressed by the AMOC strategy through consideration of the available ramp storage space and may be used as a tool to establish a desired policy of the system’s efficiency versus equity.

The goal of this paper is to describe a generic approach to the problem of optimal coordinated ramp metering control in large-scale motorway networks. In this approach, the traffic flow process is macroscopically modeled by use of a second-order macroscopic traffic flow model. The overall problem of coordinated ramp metering is formulated as a constrained discrete-time nonlinear optimal control problem, and a feasible-direction nonlinear optimization algorithm is employed for its numerical solution. The control strategy's efficiency is demonstrated through its application to the 32-km Amsterdam ring road. A number of adequately chosen scenarios along with a thorough analysis, interpretation, and suitable visualization of the obtained results provides a basis for the better understanding of some complex interrelationships of partially conflicting performance criteria. More precisely, the strategy's efficiency and equity properties as well as their tradeoff are studied and their partially competitive behavior is discussed. The results of the presented approach are very promising and demonstrate the efficiency of the optimal control methodology for motorway traffic control problems.

This paper reviews the methods used for the design of motorway network traffic control strategies. Various control strategies for the available control measures are discussed in terms of their design and their operating features, including ramp metering, route guidance, and link control. An example of an advanced coordinated ramp metering control strategy which is based on large-scale optimisation and is applied to the Amsterdam ring-road is presented.

A nonlinear rolling-horizon model-predictive hierarchical coordinated ramp metering scheme is presented. The hierarchical control structure consists of three layers: the estimation/prediction layer, the optimization layer and the direct control layer. The second layer incorporates the previously designed optimal control tool AMOC while the local feedback strategy ALINEA is used in the third layer. Simulation results are presented for the Amsterdam ring-road. It is shown that control of all on-ramps including freeway intersections leads to the optimal utilization of the available infrastructure.

Traffic congestion in urban road and freeway networks leads to a strong degradation of the network infrastructure and accordingly reduced throughput, which can be countered via suitable control measures and strategies. After illustrating the main reasons for infrastructure deterioration due to traffic congestion, a comprehensive overview of proposed and implemented control strategies is provided for three areas: urban road networks, freeway networks, and route guidance. Selected application results, obtained from either simulation studies or field implementations, are briefly outlined to illustrate the impact of various control actions and strategies. The paper concludes with a brief discussion of future needs in this important technical area.

This paper employs previously developed modeling, validation, and stimulation tools to address, for the first time, the realistic macroscopic simulation of a real large-scale motorway network. More specifically, the macroscopic simulator METANET, involving a second-order traffic flow model as well as network-relevant extensions, is utilized. A rigorous quantitative validation procedure is applied to individual network links, and subsequently a heuristic qualitative validation procedure is employed at a network level. The large-scale motorway network around Amsterdam, The Netherlands, is considered in this investigation. The main goal of the paper is to describe the application approach and procedures and to demonstrate the accuracy and usefulness of macroscopic modeling tools for large-scale motorway networks.

Recurrent and nonrecurrent congestion on freeways may be alleviated if today's "spontaneous" infrastructure utilization is replaced by an orderly, controllable operation via comprehensive application of ramp metering and freeway-to-freeway control, combined with powerful optimal control techniques. This paper first explains why ramp metering can lead to a dramatic amelioration of traffic conditions on freeways. An overview of ramp metering algorithms is provided next, ranging from early fixed-time approaches to traffic-responsive regulators and to modern sophisticated nonlinear optimal control schemes. Finally, a large-scale example demonstrates the high potential of advanced ramp metering approaches.

The problem of designing integrated trac control strategies for motorway networks with the use of ramp metering, route guidance, and motorway-to-motorway control measures is considered in this paper. A generic problem formulation is presented in the format of a discrete-time optimal control problem whose numerical solution is achieved by use of a feasible-direction algorithm. As an illustrative example, a rel-atively simple motorway network is considered under dierent control scenarios. In each case the optimal control strategy is discussed along with its eect on the trac ¯ow process. The results demonstrate the eciency of the proposed approach as well as the genuinely intelligent behaviour of the designed control strategy.

Recurrent and non-recurrent congestion on freeways may be alleviated if today's `spontaneous' infrastructure utilization is replaced by an orderly, controllable operation via comprehensive application of ramp metering and freeway-to-freeway control, combined with powerful optimal control techniques. This paper first explains why ramp metering can lead to a dramatic amelioration of traffic conditions on freeways. An overview of ramp metering algorithms is provided next, ranging from early fixed-time approaches to traffic-responsive regulators and to modern sophisticated nonlinear optimal control schemes. Finally, a large-scale example demonstrates the high potential of advanced ramp metering approaches.

- Sep 2001

One the most important components of supply chains is sales forecasting. The problem of sales forecasting considered in this
paper raises a number of requirements that must be observed in order for the long-term planning of the supply chain to be
realized successfully. These include long forecasting horizons (up to 52 periods ahead), a high number of quantities to be
forecasted, which limits the possibility of human intervention, and frequent introduction of new articles (for which no past
sales are available for parameter calibration) and withdrawal of running articles. The problem has been tackled by use of
the Holt-Winters method and by use of Feedforward Multilayer Neural Networks (FMNN) applied to sales data from two German
companies.

- Jul 2001

The goal of this paper is to describe a generic approach to the problem of optimal coordinated ramp metering control in large-scale motorway networks. In this approach, the traffic flow process is macroscopically modelled by use of a second-order macroscopic traffic flow model. The overall problem of coordinated ramp metering is formulated as a constrained discrete-time nonlinear optimal control problem, and a feasible-direction nonlinear optimisation algorithm is employed for its numerical solution. The control strategy's efficiency is demonstrated through its application to the 32-km Amsterdam ring-road for a number of different scenarios

The problem of traffic congestion in modern day motorways calls for the design and implementation of efficient control strategies. It is argued in this paper that in order to have efficient, generic, and systematic solutions to a wide range of traffic control problems, macroscopic motorway traffic flow models in state-space form, that are relevant for the control problem and computationally non-intensive, are most appropriate. Such models allow the exploitation of available powerful, systematic, and theoretically supported automatic control concepts. Based on these concepts an Extended Kalman Filter for traffic state estimation, a multivariable LQI controller for coordinated ramp metering on a motorway stretch, and an integrated optimal control strategy for motorway networks are shortly presented. The criteria of a model's relevance for a given traffic control problem and its computational requirements are subsequently examined. Finally, the application of an advanced coordinated ramp metering control strategy, based on the optimal control approach, to the ring-road of Amsterdam, The Netherlands, is provided as an illustrative example.

- Feb 2001
- Intelligent Transportation Systems, 2001. Proceedings. 2001 IEEE

The advanced motorway optimal control (AMOC) strategy for optimal
network-wide ramp metering is applied to the ring-road of Amsterdam, the
Netherlands, with the aim of investigating some important and
interesting problems arising in ubiquitous ramp metering. A number of
adequately chosen scenarios along with a thorough analysis,
interpretation, and suitable visualisation of the obtained results
provides a basis for the better understanding of some complex
interrelationships of partially conflicting performance criteria. More
precisely, the strategy's efficiency and fairness properties as well as
their trade-off are studied and their partially competitive behaviour is
discussed. This trade-off is implicitly addressed by the AMOC strategy
through consideration of the available ramp storage space, something
which may be used as a tool to establish a desired policy of the
system's efficiency versus fairness

A generic approach to the problem of optimal coordinated ramp metering control in large-scale motorway networks is described that is implemented in the software tool Advanced Motorway Optimal Control. In this approach, the traffic flow process is modeled by use of a second-order macroscopic traffic flow model. The overall problem of coordinated ramp metering is formulated as a constrained discrete-time nonlinear optimal control problem, and a feasible-direction nonlinear optimization algorithm is employed for its numerical solution. The control strategy's efficiency is demonstrated through its application to the 32-km long Amsterdam ring road. A number of different scenarios with regard to the number of controlled ramps and the available storage space are discussed in some detail. The results of the presented approach are very promising and demonstrate the high efficiency and general applicability of the optimal control methodology for motorway traffic control problems.

- Jul 2000

The goal of this paper is to describe a generic approach to the problem of optimal coordinated ramp metering control in large-scale motorway networks. In this approach, the traffic flow process is macroscopically modelled by use of a second-order macroscopic traffic flow model. The overall problem of coordinated ramp metering is formulated as a constrained discrete-time nonlinear optimal control problem, and a feasible-direction nonlinear optimisation algorithm is employed for its numerical solution. The control strategy's efficiency is demonstrated through its application to the 32 km Amsterdam ring-road for a number of different scenarios. The results of the presented approach are very promising and demonstrate the efficiency of the optimal control methodology for motorway traffic control problems.

- Sep 1999
- TRAIL Expert Seminar on Recent Advances in Traffic Flow Modelling and Control

- Feb 1999
- American Control Conference, 1999. Proceedings of the 1999

The problem of designing integrated traffic control strategies for
motorway networks with the use of ramp metering, motorway-to-motorway
control, and route guidance is considered. A generic problem formulation
is presented in the format of a discrete-time optimal control problem
whose numerical solution is achieved by use of a nonlinear optimisation
method. As an illustrative example, a relatively simple motorway network
is considered under different control scenarios. The results demonstrate
the efficiency of the proposed approach as well as the intelligent
behaviour of the designed control strategy

- Jul 1998
- 8th IFAC/IFORS/IMACS/IFIP Symposium on Large Scale Systems: Theory and Applications

- Jul 1998

This paper presents an approach for the macroscopic modelling of traffic flow on large-scale motorway networks. First the utilised traffic flow model is presented, along with the additional concept of store-and-forward links. Then we describe the modelling of the large-scale motorway network around Amsterdam. The model validation, procedure and results based on real traffic measurements from the Amsterdam network are presented.

- Feb 1998
- DACCORD Workshop on Short Term Traffic Forecasting

Co-authors

**Top co-authors**

**All co-authors (50)**