Jérôme Lacaille

Safran Aircraft Engines · Methods

In the field of aeronautical engineering, understanding and simulating aircraft engine performance is critical, especially for improving operational safety, efficiency, and sustainability. At Safran Aircraft Engines, we were able to demonstrate the effectiveness of using time series collected from the engines after each flight to build a digital tw...

We present AESim, a data-driven Aircraft Engine Simulator developed using transformer-based conditional generative adversarial networks. AESim generates samples of aircraft engine sensor measurements over full flights, conditioned on a given flight mission profile representing the flight conditions. It constitutes an essential tool in aircraft engi...

In this article, we present a statistical model that aims to predict the wear of turbofan components. The model is composed of a neural network calculating the wear of the targeted component and a parametric model describing the exposure of the material. The two parts are trained together using gradient-based optimization. We tested the model on si...

Model selection is a major challenge in non-parametric clustering. There is no universally admitted way to evaluate clustering results for the obvious reason that no ground truth is available. The difficulty to find a universal evaluation criterion is a consequence of the ill-defined objective of clustering. In this perspective, clustering stabilit...

Conditional generation of time-dependent data is a task that has much interest, whether for data augmentation, scenario simulation, completing missing data, or other purposes. Recent works proposed a Transformer-based Time series generative adversarial network (TTS-GAN) to address the limitations of recurrent neural networks. However, this model as...

Exhaust Gas Temperature (EGT) denotes the temperature of the exhaust gas when it leaves the turbine. EGT is an important parameter for measuring the energy efficiency of a turbofan engine. Indeed, the heat energy produced by an aircraft engine corresponds to a loss of power. Therefore, forecasting the exhaust gas temperature is a key task to monito...

Assessing the underlying structure of a dataset is often done by training a clustering procedure on the features describing the data. In practice, while the data may be described by a large number of features, only a minority of them may be actually informative with regard to the structure. Furthermore, redundant features may also bias the clusteri...

Production tests on test-benches are mandatory before aircraft engines are delivered, as the measurements taken during the tests determine whether the engine meets the requirements. They are essential for both aircraft and engine manufacturers and the latter can use them to obtain a comprehensive picture of the performance of each engine. Such test...

A recent research area in unsupervised learning is the combination of representation learning with deep neural networks and data clustering. The success of deep learning for supervised tasks is widely established. However, recent research has demonstrated how neural networks are able to learn representations to improve clustering in their intermedi...

Time series clustering is a challenging task due to the specificities of this type of data. Temporal correlation and invariance to transformations such as shifting, warping or noise prevent the use of standard data mining methods. Time series clustering has been mostly studied under the angle of finding efficient algorithms and distance metrics ada...

In this article, we introduce a deep learning model (denoted thereafter DCM: Deep Contextual Model ) for survival analysis able of predicting the probability that a subject meets an event of interest according to its past life. The subject and the event of interest can be diverse depending on the field of application, thus the model can be applied...

Self-Organizing Map algorithms have been used for almost 40 years across various application domains such as biology, geology, healthcare, industry and humanities as an interpretable tool to explore, cluster and visualize high-dimensional data sets. In every application, practitioners need to know whether they can \textit{trust} the resulting mappi...

Vibration analysis is an important component of industrial equipment health monitoring. Aircraft engines in particular are complex rotating machines where vibrations, mainly caused by unbalance, misalignment, or damaged bearings, put engine parts under dynamic structural stress. Thus, monitoring the vibratory behavior of engines is essential to det...

Engines are verified through production tests before delivering them to customers. During those tests, lot of measures are taken on different parts of the engine, considering multiple physical parameters. Unexpected measures can be observed. For this very reason, it is important to assess if these unusual observations are statistically significant....

The anomaly detection problem for univariate or multivariate time series is a critical question in many practical applications as industrial processes control, biological measures, engine monitoring, supervision of all kinds of behavior. In this paper we propose a simple and empirical approach to detect anomalies in the behavior of multivariate tim...

Model selection is a major challenge in non-parametric clustering. There is no universally admitted way to evaluate clustering results for the obvious reason that there is no ground truth against which results could be tested, as in supervised learning. The difficulty to find a universal evaluation criterion is a direct consequence of the fundament...

Ten years ago we develop a numeric state vector to allow us following the performance of a small fleet of turbofan engines. As engine manufacturer with "flight by the hour" new contracts going we needed to improve our technology, hence building analytic digital-twin solutions. This improvement allows to monitor the engines as well as the behavior o...

How can we tell if a flight is normal or abnormal? In Safran Aircraft Engines, we are interested in the engine behavior. Some data are collected at low frequency between 1Hz up to 66Hz. These data are mainly measurements acquired from engines sensors, information coming from the aircraft that are needed to control the propulsion system and results...

Recent research has demonstrated how deep neural networks are able to learn representations to improve data clustering. By considering representation learning and clustering as a joint task, models learn clustering-friendly spaces and achieve superior performance, compared with standard two-stage approaches where dimensionality reduction and cluste...

For aircraft engineers, detecting abnormalities in a large dataset of recorded flights and understanding the reasons for these are crucial development and monitoring issues. The main difficulty comes from the fact that flights have unequal lengths, and data is usually high dimensional, with a variety of recorded signals. This question is addressed...

The anomaly detection problem for univariate or multivariate time series is a critical question in many practical applications as industrial processes control, biological measures, engine monitoring, supervision of all kinds of behavior. In this paper we propose an empirical approach to detect anomalies in the behavior of multivariate time series....

The anomaly detection problem for univariate or multivariate time series is a critical question in many practical applications as industrial processes control, biological measures, engine monitoring, supervision of all kinds of behavior. In this paper we propose an empirical approach to detect anomalies in the behavior of multivariate time series....

Poster for the USPN Galilée graduate school day 2019 (Journée de l'école doctorale Galilée)

In the wake of recent advances in joint clustering and deep learning, we introduce the Deep Embedded Self-Organizing Map, a model that jointly learns representations and the code vectors of a self-organizing map. Our model is composed of an autoencoder and a custom SOM layer that are optimized in a joint training procedure, motivated by the idea th...

Every day, several aircraft engines exit Safran plant in the south of Paris. Each engine is assembled and sent for a last bench test before shipment to the aircraft. Among all operations implemented during this hour length phase and after the first run-in, we realize a slow acceleration and a slow deceleration. During those two steps the engine is...

Recent research has demonstrated how deep neural networks are able to learn representations to improve data clustering. By considering representation learning and clustering as a joint task, models learn clustering-friendly spaces and achieve superior performance, compared with standard two-stage approaches where dimensionality reduction and cluste...

Airlines main possession cost factors are fuel burn and maintenance (1). Lately the technics of Structural Health Monitoring (SHM) sounded promising about enabling gain, not limited to ground diagnostics’ field. During the last two decades, structural health assessments, from non-destructive tests for diagnostics to real time monitoring, were invol...

Context Thrust is the primary functional characteristics of turbofan engines. It is essentially produced by the fan. The normalized thrust margin measurements during the final acceptance tests are affected by long term evolution and high level scatter.  The normalized thrust is the measured thrust transposed into standard atmosphere conditions....

L'invention concerne un procédé d'équilibrage d'un ensemble d'aubes (5) destinées à être disposées sur un disque nu (7) d'un moteur d'aéronef, le disque nu (7) comportant un nombre déterminé d'alvéoles (ai) numérotées destinées à recevoir le même nombre déterminé d'aubes qui peuvent présenter une dispersion de masse, ledit procédé étant comportant...

Thrust is the main performance figure for a turbofan. The engine is sold for a given thrust and cannot be delivered under a minimum thrust level (EASA, 2010), requirement CS-E 40 (f). Hence it is fundamental to accurately evaluate thrust. All individual engines are verified before delivering to the customer during pass-off tests. However, those tes...

Analysing multivariate time series created by sensors during a flight or a test bench represents a new challenge for engineers. Each of them can be decomposed univariately into series of stabilised phases, well known by the expert, and transient phases that are merely explored but very informative when the engine is running. Our project aims at con...

A method for measuring a thrust margin of a turbomachine, in which data are acquired including the thrust margin which is determined as a function of a specified thrust and a measured thrust, the measured thrust being determined on a measuring bench which includes a bench equipment and on which the turbomachine is, wherein a time evolution of the t...

Multiple signals are measured by sensors during a flight or a test bench and their analysis represent a big interest for engineers. These signals are actually multivariate time series created by the sensors present on the aircraft engines. Each of them can be decomposed into series of stabilized phases, well known by the experts, and transient phas...

The main concern for airlines is fuel consumption but also a long-term expectancy about the engine cost during its full life. This includes maintenance frequency and shop costs. In the next five years, Snecma, as engine manufacturer, needs to be able to collect more than one gigabyte of data per flight and per engine. This becomes huge as the flow...

Engine Health Management (EHM) is the up to date solution that is used by Aircraft Engine Manufacturers in order to maintain an engine operative through a reduction of operational events that impact its availability for end customers. The aim of EHM systems is to monitor and forecast the health status of an engine based on operational data in order...

In this paper, vibration analysis of civil aircraft engines in a test-bench to perform anomaly detection is considered. High bandwidth vibration measurements contain essential mechanical information regarding the condition of the engine and the localisation of damage, if present. In this case, vibration data are represented by spectrograms in the f...

To ensure the liability of civil aircrafts, engines have to be tested after their production. Vibrations are one of the most informative measures to diagnose some damages in the engine if any. The representation of these vibrations as spectrograms provides visual signatures related to damages. However, this representation is noisy and high-dimensio...

Numerous sensors on SNECMA's engines capture a considerable amount of data during tests or flights. In order to detect potential crucial changes of characteristic variables, it is relevant to develop powerful statistical algorithms. This manuscript is devoted to offline change-point detection, in piecewise linear models with an unknown number of ch...

Every day, new engine configurations or engine parts are tested in Snecma's test benches. During each test, up to two thousand sensors capture every bit of information generated by the engine or the bench cell itself. It is extremely difficult to manually analyze all this data. Due to the huge amount of data and their diversity, specialists who ana...

L'invention concerne un procédé et un système d'analyse de données d'essais relatives à un composant moteur, comportant : - un dispositif d'acquisition (3) configuré pour acquérir un signal temporel multivarié (15) de données d'essais relatives audit composant moteur (9), - un dispositif de traitement (5) configuré pour identifier des motifs (21) d...

A flight data evaluation system to optimise operations of an aircraft, comprising: an acquisition circuit (3) configured to collect observation data (D1−Dn) related to a fleet of aircraft, and a processing circuit (5) configured: to assign quality values (Qi) to the observation data by applying predetermined learning models (M1−Mn) to them correspo...

What about a software tool that behaves like a gauge able to estimate the quantity of information contained in a group of measurements? Then if we have a performance indicator or a defect rate, how may we compute the maximum performance explanation contained in our dataset? The first question may be answered by entropy and the second with mutual in...

What about a software tool that behaves like a gauge able to estimate the quantity of information contained in a group of measurements? Then if we have a performance indicator or a defect rate, how may we compute the maximum performance explanation contained in our dataset? The first question may be answered by entropy and the second with mutual in...

A decision aid system, method and computer program product for the maintenance of a machine, including anomaly detection modules to determine health indicators on the basis of measurements of physical parameters of the machine, a calculator to compute an operating diagnosis on the basis of health indicators by applying a decision model capable of l...

Thousands of aircraft fly with engines built by Snecma or CFM International (a joint venture between Snecma and GE). Every day, data from these engines are broadcast to their ground systems. When airline companies ask their engine manufacturer to help them manage their fleet and plan maintenance actions, data are analysed by experts using health mo...

We compare in this paper several feature selection methods for the Naive Bayes Classifier (NBC) when the data under study are described by a large number of redundant binary indicators. Wrapper approaches guided by the NBC estimation of the classification error probability out-perform filter approaches while retaining a reasonable computational cos...

A method of estimating future change in operation of a monitored aircraft (A), including the following steps performed by a computer on board the monitored aircraft, to calculate a current state of the monitored aircraft (ECA) from measurements (VFA) of variables related to operation of the monitored aircraft, to send a request to analyze the simil...

L'invention propose un procédé de détection d'anomalie d'une turbomachine, le procédé étant mis en œuvre à partir d'un spectrogramme caractérisant un comportement vibratoire de la turbomachine en fonction d'une vitesse de rotation d'un arbre de la turbomachine, caractérisé en ce qu'il comprend la mise en œuvre des étapes consistant à : - subdiviser...

A system for pooling observation data relating to aircraft engines includes a receiver adapted for recovering the observation data from distinct entities, a processor adapted for describing the observation data in a metric space by transforming them into measurable observation states, and a database adapted for storing therein the observation state...

Detecting early signs of failures (anomalies) in complex systems is one of the main goal of preventive maintenance. It allows in particular to avoid actual failures by (re)scheduling maintenance operations in a way that optimizes maintenance costs. Aircraft engine health monitoring is one representative example of a field in which anomaly detection...

A method of analysis of the state of operation of a machine including a learning step supplementing a reference database with one or more thresholds for one or more indicators calculated on the basis of signals delivered by a sensor associated with the machine, the learning step including the following operations implemented by a computer processin...

A tool for validation of a system for monitoring at least one piece of equipment in an aircraft engine, also comprising a computer configured to: collect observation data related to the equipment, calculate a current value of at least one quality indicator on a current quantity of observation data, estimate the probability that the current value of...

To understand the behavior of a turbofan engine, one first needs to deal with the variety of data acquisition contexts. Each time a set of measurements is acquired, and such set may account for tens of parameters, the aircraft evolves in a specific flight mode. A diagnostic of the engine behavior models the observations and tests if anything appear...

Automatic anomaly detection is a major issue in various areas. Beyond mere detection, the identification of the source of the problem that produced the anomaly is also essential. This is particularly the case in aircraft engine health monitoring where detecting early signs of failure (anomalies) and helping the engine owner to implement efficiently...

Aircraft engine manufacturers collect large amount of engine related data during flights. These data are used to detect anomalies in the engines in order to help companies optimize their maintenance costs. This article introduces and studies a generic methodology that allows one to build automatic early signs of anomaly detection in a way that is u...