R. Gouriveau - Academia.edu (original) (raw)

Papers by R. Gouriveau

hal.inria.fr

Page 1. Spécification d'un système neuro-flou de prédiction de défaillances à moyen terme Sp... more Page 1. Spécification d'un système neuro-flou de prédiction de défaillances à moyen terme Specification of a neuro-fuzzy system for a mid-term failure prediction GOURIVEAU Rafael EL-KOUJOK Mohamed ZERHOUNI Noureddine ...

8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes, 2012

Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at p... more Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at predicting the life time at which a component or a system will be unable to perform a desired function. This task is achieved through essential steps of data acquisition, feature extraction and selection, and prognostic modeling. This paper emphasizes on the selection phase and aims at showing that it should be performed according to the predictability of features: as there is no interest in retaining features that are hard to be predicted. Thereby, predictability is defined and a feature selection procedure based on this concept is proposed. The effectiveness of the approach is judged by applying it on a real-world case: through comparison is made in order to show that the better predictable features lead to better RU L estimation.

2012 IEEE Conference on Prognostics and Health Management, 2012

Although efforts in this field are significant around the world, real prognostics systems are sti... more Although efforts in this field are significant around the world, real prognostics systems are still scarce in industry. Indeed, it is hard to provide efficient approaches that are able to handle with the inherent uncertainty of prognostics and nobody is able to a priori ensure that an accurate prognostic model can be built. As for an example of remaining problems, consider datadriven prognostics approaches: how to ensure that a model will be able to face with inputs variation with respect to those ones that have been learned, how to ensure that a learned-model will face with unknown data, how to ensure convergence of algorithms, etc. In other words, robustness, reliability and applicability of a prognostic approach are still open areas. Following that, the aim of this paper is to address these challenges by proposing a new neural network (structure and algorithm) that enhances reliability of RUL estimates while improving applicability of the approach. Robustness, reliability and applicability aspects are first discussed and defined according to literature. On this basis, a new connexionist system is proposed for prognostics: the Improved-Extreme Learning machine (Imp-ELM). This neural network, based on complex activation functions, enables to reduce the influence of human choices and initial parameterization, while improving accuracy of estimates and speeding the learning phase. The whole proposition is illustrated by performing tests on a real industrial case of cutting tools from a Computer Numerical Control (CNC) machine. This is achieved by predicting tool condition (wear) in terms of remaining cuts successfully made. Thorough comparisons with adaptive neuro fuzzy inference system (ANFIS) and existing ELM algorithm are also given. Results show improved robustness, reliability and applicability performances.

2011 Prognostics and System Health Managment Confernece, 2011

According to Condition Based Maintenance practitioners, various activities, ranging from data col... more According to Condition Based Maintenance practitioners, various activities, ranging from data collection through the recommendation of specific maintenance actions, must be carried out to perform predictive maintenance. Nevertheless, in practice, (and in spite of recommendations like those ones of the OSA-CBM standard), defining and implementing a computer software system for CBM is not a trivial task. That can be mostly explained by the necessity of providing a distributed application that enables to share data and information in an easy but effective manner in-between various actors from various industrial plants. Following that, the aim of the paper is to describe a collaborative software that has been developed in the society e-m@systec. Its simple architecture, as well as its evolving and customizable capabilities make the global information system as useful for distributed applications. The usage of JEE technology improves the portability of the system. This software is well adapted to support predictive maintenance strategies. Thereby and as for an illustration, an example related to a prognostic problem is also described.

Neurocomputing, Aug 1, 2010

Nonlinear autoregressive moving average with exogenous inputs (NARMAX) models have been successfu... more Nonlinear autoregressive moving average with exogenous inputs (NARMAX) models have been successfully demonstrated for modeling the input-output behavior of many complex systems. This paper deals with the proposition of a scheme to provide time series prediction. The approach is based on a recurrent NARX model obtained by linear combination of a recurrent neural network (RNN) output and the real data output. Some prediction metrics are also proposed to assess the quality of predictions. This metrics enable to compare different prediction schemes and provide an objective way to measure how changes in training or prediction model (Neural network architecture) affect the quality of predictions. Results show that the proposed NARX approach consistently outperforms the prediction obtained by the RNN neural network.

This paper deals with the presentation of an experimental platform called PRONOSTIA, which enable... more This paper deals with the presentation of an experimental platform called PRONOSTIA, which enables testing, verifying and validating methods related to bearing health assessment, diagnostic and prognostic. The choice of bearings is justified by the fact that most of failures of rotating machines are related to these components. Therefore, bearings can be considered as critical as their failure significantly decreases availability and security of machines. The main objective of PRONOSTIA is to provide real data related to accelerated degradation of bearings performed under constant and/or variable operating conditions, which are online controlled. The operating conditions are characterized by two sensors: a rotating speed sensor and a force sensor. In PRONOSTIA platform, the bearing's health monitoring is ensured by gathering online two types of signals: temperature and vibration (horizontal and vertical accelerometers). Furthermore, the data are recorded with a specific sampling frequency which allows catching all the frequency spectrum of the bearing during its whole degradation. Finally, the monitoring data provided by the sensors can be used for further processing in order to extract relevant features and continuously assess the health condition of the bearing. During the PHM conference, a "IEEE PHM 2012 Prognostic Challenge" is organized. For this purpose, a web link to the degradation data is provided to the competitors to allow them testing and verifying their prognostic methods. The results of each method can then be evaluated regarding its capability to accurately estimate the remaining useful life of the tested bearings.

International Journal of Hydrogen Energy, 2015

Mathematics and Computers in Simulation, 2015

International Journal of Hydrogen Energy, 2014

ABSTRACT This paper studies the prediction of the output voltage reduction caused by degradation ... more ABSTRACT This paper studies the prediction of the output voltage reduction caused by degradation during nominal operating condition of a PEM fuel cell stack. It proposes a methodology based on Adaptive Neuro-Fuzzy Inference Systems (ANFIS) which use as input the measures of the fuel cell output voltage during operation. The paper presents the architecture of the ANFIS and studies the selection of its parameters. As the output voltage cannot be represented as a periodical signal, the paper proposes to predict its temporal variation which is then used to construct the prediction of the output voltage. The paper also proposes to split this signal in two components: normal operation and external perturbations. The second component cannot be predicted and then it is not used to train the ANFIS. The performance of the prediction is evaluated on the output voltage of two fuel cells during a long term operation (1000 h). Validation results suggest that the proposed technique is well adapted to predict degradation in fuel cell systems.

Fuel Cells, 2014

ABSTRACT Hybridization of proton exchange membrane fuel cells (PEMFC) and ultra capacitors (UC) a... more ABSTRACT Hybridization of proton exchange membrane fuel cells (PEMFC) and ultra capacitors (UC) are considered as an alternative way to implement high autonomy, high dynamic, and reversible energy sources. PEMFC allow high efficiency and high autonomy, however their dynamic response is limited and this source does not allow recovering energy. UC appears to be a complementary source to fuel cell systems (FCS) due to their high power density, fast dynamics, and reversibility. A direct hybridization of these sources could allow reducing the number of power converters and then the total cost of the hybridized system. Simulations show the behavior of the hybrid source when the fuel cell and ultra capacitors are interconnected and the natural energy management when a charge is connected. The results show that the magnitude of the transient current supplied by the fuel cell to charge the UC can be much higher than its nominal value. An experimental setup is implemented to study the effects of these high currents in a PEMFC. This is done by imposing a controlled short-circuit between the electrodes. The PEMFC degradation is quantified by using electrochemical impedance spectroscopy.

2014 IEEE Vehicle Power and Propulsion Conference (VPPC), 2014

The Hydrogen energy vector is one of the possible solutions to overcome future energy crisis anno... more The Hydrogen energy vector is one of the possible solutions to overcome future energy crisis announced by the International Energy Agency. However, various bottleneck, whether technological or societal, slow the industrial interest for this technology and therefore the mass production of fuel cells. Among these locks that may be mentioned one relating to the still limited useful lifetime of the fuel cells. To improve this lifetime, one of the existing approaches is to use the discipline of PHM (for Prognostics and Health Management). This discipline aims to improve the efficiency of control and maintenance operations on the system by using diagnostic or prognostics algorithms. This article covers the prognostics aspect of PHM applied to a PEMFC using an algorithm based on a tool from the reservoir computing discipline to predict the Remaining Useful Lifetime.

2014 IEEE Vehicle Power and Propulsion Conference (VPPC), 2014

IECON 2013 - 39th Annual Conference of the IEEE Industrial Electronics Society, 2013

One remaining technological bottleneck to develop industrial Fuel Cell (FC) applications resides ... more One remaining technological bottleneck to develop industrial Fuel Cell (FC) applications resides in the system limited useful lifetime. Consequently, it is important to develop failure diagnostic and prognostic tools enabling the optimization of the FC. Among all the existing prognostics approaches, datamining methods such as artificial neural networks aim at estimating the process' behavior without huge knowledge about the underlying physical phenomena. Nevertheless, this kind of approach needs huge learning dataset. Also, the deployment of such an approach can be long (trial and error method), which represents a real problem for industrial applications where realtime complying algorithms must be developed. According to this, the aim of this paper is to study the application of a reservoir computing tool (the Echo State Network) as a prognostics system enabling the estimation of the Remaining Useful Life of a Proton Exchange Membrane Fuel Cell. Developments emphasize on the prediction of the mean voltage cells of a degrading FC. Accuracy and time consumption of the approach are studied, as well as sensitivity of several parameters of the ESN. Results appear to be very promising.

2014 International Conference on Prognostics and Health Management, 2014

In the perspective of decreasing polluting emissions and developing alternative energies, fuel ce... more In the perspective of decreasing polluting emissions and developing alternative energies, fuel cells, and more precisely Proton Exchange Membrane Fuel Cells (PEMFC), represent a promising solution. Even if this technology is close to being competitive, it still suffers from too short life duration. As a consequence, prognostic seems to be a great solution to anticipate PEMFC stacks degradation. However, a PEMFC implies multiphysics and multiscale phenomena making the construction of an aging model only based on physics very complex. One solution consists in using a hybrid approach for prognostics combining the use of models and available data. Among these hybrid approaches, particle filtering methods seem to be really appropriate as they offer the possibility to compute models with time varying parameters and to update them all along the prognostics process. But to be efficient, not only should the prognostics system take into account the aging of the stack but also external events influencing this aging. Indeed, some acquisition techniques introduce disturbances in the fuel cell behavior and a voltage recovery can be observed at the end of the characterization process. This paper proposes to tackle this problem. First, PEMFC fuel cells and their complexities are introduced. Then, the impact of characterization of the fuel cell behavior is described. Empirical models are built and introduced in both learning and prediction phases of the prognostics model by combining three particle filters. The new prognostic framework is used to perform remaining useful life estimates and the whole proposition is illustrated with a long term experiment data set of a PEMFC in constant load solicitation and stable operating conditions. Estimates can be given with an error less than 5% for life durations of more than 1000 hours. Finally, the results are compared to a previous work to show that introducing a disturbance modeling can dramatically reduce the uncertainty coming with the predictions.

2010 11th International Conference on Control Automation Robotics & Vision, 2010

The growing interest in predictive maintenance makes industrials and researchers turning themselv... more The growing interest in predictive maintenance makes industrials and researchers turning themselves to artificial intelligence methods for fulfilling the tasks of condition monitoring and prognostics. Within this frame, the general purpose of this paper is to investigate the capabilities of an Evolving eXtended Takagi Sugeno (exTS) based neuro-fuzzy algorithm to predict the tool condition in high-speed machining conditions. The performance of evolving Neuro-Fuzzy model is compared with an Adaptive Neuro-Fuzzy Inference System (ANFIS) and a Multiple Regression Model (MRM) in term of accuracy and reliability through a case study of tool condition monitoring. The reliability of exTS also investigated.

IEEE Transactions on Industrial Electronics, 2015

Performances of data-driven prognostics approaches are closely dependent on form, and trend of ex... more Performances of data-driven prognostics approaches are closely dependent on form, and trend of extracted features. Indeed, features that clearly reflect the machine degradation, should lead to accurate prognostics, which is the global objective of the paper. This paper contributes a new approach for features extraction / selection: the extraction is based on trigonometric functions and cumulative transformation, and the selection is performed by evaluating feature fitness using monotonicity and trendability characteristics. The proposition is applied to time-frequency analysis of non-stationary signals using Discrete Wavelet Transform. The main idea is to map raw vibration data into monotonic features with early trends, which can be easily predicted. To show that, selected features are used to build a model with a data-driven approach namely, the Summation Wavelet-Extreme Learning Machine, that enables a good balance between model accuracy and complexity. For validation and generalization purpose, vibration data from two real applications of Prognostics and Health Management challenges are used: 1) cutting tools from Computer Numerical Control (CNC) machine (2010), and 2) bearings from platform PRONOSTIA (2012). Performances of the proposed approach are thoroughly compared with the classical approach by performing: feature fitness analysis, cutting tool wear "estimation" and bearings "long-term predictions" tasks, which validates the proposition.

Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at p... more Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at predicting the life time at which a component or a system will be unable to perform a desired function. This task is achieved through essential steps of data acquisition, feature extraction and selection, and prognostic modeling. This paper emphasizes on the selection phase and aims at showing that it should be performed according to the predictability of features: as there is no interest in retaining features that are hard to be predicted. Thereby, predictability is defined and a feature selection procedure based on this concept is proposed. The effectiveness of the approach is judged by applying it on a real-world case: through comparison is made in order to show that the better predictable features lead to better RU L estimation.

2012 IEEE Conference on Prognostics and Health Management, 2012

Although efforts in this field are significant around the world, real prognostics systems are sti... more Although efforts in this field are significant around the world, real prognostics systems are still scarce in industry. Indeed, it is hard to provide efficient approaches that are able to handle with the inherent uncertainty of prognostics and nobody is able to a priori ensure that an accurate prognostic model can be built. As for an example of remaining problems, consider datadriven prognostics approaches: how to ensure that a model will be able to face with inputs variation with respect to those ones that have been learned, how to ensure that a learned-model will face with unknown data, how to ensure convergence of algorithms, etc. In other words, robustness, reliability and applicability of a prognostic approach are still open areas. Following that, the aim of this paper is to address these challenges by proposing a new neural network (structure and algorithm) that enhances reliability of RUL estimates while improving applicability of the approach. Robustness, reliability and applicability aspects are first discussed and defined according to literature. On this basis, a new connexionist system is proposed for prognostics: the Improved-Extreme Learning machine (Imp-ELM). This neural network, based on complex activation functions, enables to reduce the influence of human choices and initial parameterization, while improving accuracy of estimates and speeding the learning phase. The whole proposition is illustrated by performing tests on a real industrial case of cutting tools from a Computer Numerical Control (CNC) machine. This is achieved by predicting tool condition (wear) in terms of remaining cuts successfully made. Thorough comparisons with adaptive neuro fuzzy inference system (ANFIS) and existing ELM algorithm are also given. Results show improved robustness, reliability and applicability performances.

hal.inria.fr

Page 1. Spécification d'un système neuro-flou de prédiction de défaillances à moyen terme Sp... more Page 1. Spécification d'un système neuro-flou de prédiction de défaillances à moyen terme Specification of a neuro-fuzzy system for a mid-term failure prediction GOURIVEAU Rafael EL-KOUJOK Mohamed ZERHOUNI Noureddine ...

8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes, 2012

Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at p... more Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at predicting the life time at which a component or a system will be unable to perform a desired function. This task is achieved through essential steps of data acquisition, feature extraction and selection, and prognostic modeling. This paper emphasizes on the selection phase and aims at showing that it should be performed according to the predictability of features: as there is no interest in retaining features that are hard to be predicted. Thereby, predictability is defined and a feature selection procedure based on this concept is proposed. The effectiveness of the approach is judged by applying it on a real-world case: through comparison is made in order to show that the better predictable features lead to better RU L estimation.

2012 IEEE Conference on Prognostics and Health Management, 2012

Although efforts in this field are significant around the world, real prognostics systems are sti... more Although efforts in this field are significant around the world, real prognostics systems are still scarce in industry. Indeed, it is hard to provide efficient approaches that are able to handle with the inherent uncertainty of prognostics and nobody is able to a priori ensure that an accurate prognostic model can be built. As for an example of remaining problems, consider datadriven prognostics approaches: how to ensure that a model will be able to face with inputs variation with respect to those ones that have been learned, how to ensure that a learned-model will face with unknown data, how to ensure convergence of algorithms, etc. In other words, robustness, reliability and applicability of a prognostic approach are still open areas. Following that, the aim of this paper is to address these challenges by proposing a new neural network (structure and algorithm) that enhances reliability of RUL estimates while improving applicability of the approach. Robustness, reliability and applicability aspects are first discussed and defined according to literature. On this basis, a new connexionist system is proposed for prognostics: the Improved-Extreme Learning machine (Imp-ELM). This neural network, based on complex activation functions, enables to reduce the influence of human choices and initial parameterization, while improving accuracy of estimates and speeding the learning phase. The whole proposition is illustrated by performing tests on a real industrial case of cutting tools from a Computer Numerical Control (CNC) machine. This is achieved by predicting tool condition (wear) in terms of remaining cuts successfully made. Thorough comparisons with adaptive neuro fuzzy inference system (ANFIS) and existing ELM algorithm are also given. Results show improved robustness, reliability and applicability performances.

2011 Prognostics and System Health Managment Confernece, 2011

According to Condition Based Maintenance practitioners, various activities, ranging from data col... more According to Condition Based Maintenance practitioners, various activities, ranging from data collection through the recommendation of specific maintenance actions, must be carried out to perform predictive maintenance. Nevertheless, in practice, (and in spite of recommendations like those ones of the OSA-CBM standard), defining and implementing a computer software system for CBM is not a trivial task. That can be mostly explained by the necessity of providing a distributed application that enables to share data and information in an easy but effective manner in-between various actors from various industrial plants. Following that, the aim of the paper is to describe a collaborative software that has been developed in the society e-m@systec. Its simple architecture, as well as its evolving and customizable capabilities make the global information system as useful for distributed applications. The usage of JEE technology improves the portability of the system. This software is well adapted to support predictive maintenance strategies. Thereby and as for an illustration, an example related to a prognostic problem is also described.

Neurocomputing, Aug 1, 2010

Nonlinear autoregressive moving average with exogenous inputs (NARMAX) models have been successfu... more Nonlinear autoregressive moving average with exogenous inputs (NARMAX) models have been successfully demonstrated for modeling the input-output behavior of many complex systems. This paper deals with the proposition of a scheme to provide time series prediction. The approach is based on a recurrent NARX model obtained by linear combination of a recurrent neural network (RNN) output and the real data output. Some prediction metrics are also proposed to assess the quality of predictions. This metrics enable to compare different prediction schemes and provide an objective way to measure how changes in training or prediction model (Neural network architecture) affect the quality of predictions. Results show that the proposed NARX approach consistently outperforms the prediction obtained by the RNN neural network.

This paper deals with the presentation of an experimental platform called PRONOSTIA, which enable... more This paper deals with the presentation of an experimental platform called PRONOSTIA, which enables testing, verifying and validating methods related to bearing health assessment, diagnostic and prognostic. The choice of bearings is justified by the fact that most of failures of rotating machines are related to these components. Therefore, bearings can be considered as critical as their failure significantly decreases availability and security of machines. The main objective of PRONOSTIA is to provide real data related to accelerated degradation of bearings performed under constant and/or variable operating conditions, which are online controlled. The operating conditions are characterized by two sensors: a rotating speed sensor and a force sensor. In PRONOSTIA platform, the bearing's health monitoring is ensured by gathering online two types of signals: temperature and vibration (horizontal and vertical accelerometers). Furthermore, the data are recorded with a specific sampling frequency which allows catching all the frequency spectrum of the bearing during its whole degradation. Finally, the monitoring data provided by the sensors can be used for further processing in order to extract relevant features and continuously assess the health condition of the bearing. During the PHM conference, a "IEEE PHM 2012 Prognostic Challenge" is organized. For this purpose, a web link to the degradation data is provided to the competitors to allow them testing and verifying their prognostic methods. The results of each method can then be evaluated regarding its capability to accurately estimate the remaining useful life of the tested bearings.

International Journal of Hydrogen Energy, 2015

Mathematics and Computers in Simulation, 2015

International Journal of Hydrogen Energy, 2014

ABSTRACT This paper studies the prediction of the output voltage reduction caused by degradation ... more ABSTRACT This paper studies the prediction of the output voltage reduction caused by degradation during nominal operating condition of a PEM fuel cell stack. It proposes a methodology based on Adaptive Neuro-Fuzzy Inference Systems (ANFIS) which use as input the measures of the fuel cell output voltage during operation. The paper presents the architecture of the ANFIS and studies the selection of its parameters. As the output voltage cannot be represented as a periodical signal, the paper proposes to predict its temporal variation which is then used to construct the prediction of the output voltage. The paper also proposes to split this signal in two components: normal operation and external perturbations. The second component cannot be predicted and then it is not used to train the ANFIS. The performance of the prediction is evaluated on the output voltage of two fuel cells during a long term operation (1000 h). Validation results suggest that the proposed technique is well adapted to predict degradation in fuel cell systems.

Fuel Cells, 2014

ABSTRACT Hybridization of proton exchange membrane fuel cells (PEMFC) and ultra capacitors (UC) a... more ABSTRACT Hybridization of proton exchange membrane fuel cells (PEMFC) and ultra capacitors (UC) are considered as an alternative way to implement high autonomy, high dynamic, and reversible energy sources. PEMFC allow high efficiency and high autonomy, however their dynamic response is limited and this source does not allow recovering energy. UC appears to be a complementary source to fuel cell systems (FCS) due to their high power density, fast dynamics, and reversibility. A direct hybridization of these sources could allow reducing the number of power converters and then the total cost of the hybridized system. Simulations show the behavior of the hybrid source when the fuel cell and ultra capacitors are interconnected and the natural energy management when a charge is connected. The results show that the magnitude of the transient current supplied by the fuel cell to charge the UC can be much higher than its nominal value. An experimental setup is implemented to study the effects of these high currents in a PEMFC. This is done by imposing a controlled short-circuit between the electrodes. The PEMFC degradation is quantified by using electrochemical impedance spectroscopy.

2014 IEEE Vehicle Power and Propulsion Conference (VPPC), 2014

The Hydrogen energy vector is one of the possible solutions to overcome future energy crisis anno... more The Hydrogen energy vector is one of the possible solutions to overcome future energy crisis announced by the International Energy Agency. However, various bottleneck, whether technological or societal, slow the industrial interest for this technology and therefore the mass production of fuel cells. Among these locks that may be mentioned one relating to the still limited useful lifetime of the fuel cells. To improve this lifetime, one of the existing approaches is to use the discipline of PHM (for Prognostics and Health Management). This discipline aims to improve the efficiency of control and maintenance operations on the system by using diagnostic or prognostics algorithms. This article covers the prognostics aspect of PHM applied to a PEMFC using an algorithm based on a tool from the reservoir computing discipline to predict the Remaining Useful Lifetime.

2014 IEEE Vehicle Power and Propulsion Conference (VPPC), 2014

IECON 2013 - 39th Annual Conference of the IEEE Industrial Electronics Society, 2013

One remaining technological bottleneck to develop industrial Fuel Cell (FC) applications resides ... more One remaining technological bottleneck to develop industrial Fuel Cell (FC) applications resides in the system limited useful lifetime. Consequently, it is important to develop failure diagnostic and prognostic tools enabling the optimization of the FC. Among all the existing prognostics approaches, datamining methods such as artificial neural networks aim at estimating the process' behavior without huge knowledge about the underlying physical phenomena. Nevertheless, this kind of approach needs huge learning dataset. Also, the deployment of such an approach can be long (trial and error method), which represents a real problem for industrial applications where realtime complying algorithms must be developed. According to this, the aim of this paper is to study the application of a reservoir computing tool (the Echo State Network) as a prognostics system enabling the estimation of the Remaining Useful Life of a Proton Exchange Membrane Fuel Cell. Developments emphasize on the prediction of the mean voltage cells of a degrading FC. Accuracy and time consumption of the approach are studied, as well as sensitivity of several parameters of the ESN. Results appear to be very promising.

2014 International Conference on Prognostics and Health Management, 2014

In the perspective of decreasing polluting emissions and developing alternative energies, fuel ce... more In the perspective of decreasing polluting emissions and developing alternative energies, fuel cells, and more precisely Proton Exchange Membrane Fuel Cells (PEMFC), represent a promising solution. Even if this technology is close to being competitive, it still suffers from too short life duration. As a consequence, prognostic seems to be a great solution to anticipate PEMFC stacks degradation. However, a PEMFC implies multiphysics and multiscale phenomena making the construction of an aging model only based on physics very complex. One solution consists in using a hybrid approach for prognostics combining the use of models and available data. Among these hybrid approaches, particle filtering methods seem to be really appropriate as they offer the possibility to compute models with time varying parameters and to update them all along the prognostics process. But to be efficient, not only should the prognostics system take into account the aging of the stack but also external events influencing this aging. Indeed, some acquisition techniques introduce disturbances in the fuel cell behavior and a voltage recovery can be observed at the end of the characterization process. This paper proposes to tackle this problem. First, PEMFC fuel cells and their complexities are introduced. Then, the impact of characterization of the fuel cell behavior is described. Empirical models are built and introduced in both learning and prediction phases of the prognostics model by combining three particle filters. The new prognostic framework is used to perform remaining useful life estimates and the whole proposition is illustrated with a long term experiment data set of a PEMFC in constant load solicitation and stable operating conditions. Estimates can be given with an error less than 5% for life durations of more than 1000 hours. Finally, the results are compared to a previous work to show that introducing a disturbance modeling can dramatically reduce the uncertainty coming with the predictions.

2010 11th International Conference on Control Automation Robotics & Vision, 2010

The growing interest in predictive maintenance makes industrials and researchers turning themselv... more The growing interest in predictive maintenance makes industrials and researchers turning themselves to artificial intelligence methods for fulfilling the tasks of condition monitoring and prognostics. Within this frame, the general purpose of this paper is to investigate the capabilities of an Evolving eXtended Takagi Sugeno (exTS) based neuro-fuzzy algorithm to predict the tool condition in high-speed machining conditions. The performance of evolving Neuro-Fuzzy model is compared with an Adaptive Neuro-Fuzzy Inference System (ANFIS) and a Multiple Regression Model (MRM) in term of accuracy and reliability through a case study of tool condition monitoring. The reliability of exTS also investigated.

IEEE Transactions on Industrial Electronics, 2015

Performances of data-driven prognostics approaches are closely dependent on form, and trend of ex... more Performances of data-driven prognostics approaches are closely dependent on form, and trend of extracted features. Indeed, features that clearly reflect the machine degradation, should lead to accurate prognostics, which is the global objective of the paper. This paper contributes a new approach for features extraction / selection: the extraction is based on trigonometric functions and cumulative transformation, and the selection is performed by evaluating feature fitness using monotonicity and trendability characteristics. The proposition is applied to time-frequency analysis of non-stationary signals using Discrete Wavelet Transform. The main idea is to map raw vibration data into monotonic features with early trends, which can be easily predicted. To show that, selected features are used to build a model with a data-driven approach namely, the Summation Wavelet-Extreme Learning Machine, that enables a good balance between model accuracy and complexity. For validation and generalization purpose, vibration data from two real applications of Prognostics and Health Management challenges are used: 1) cutting tools from Computer Numerical Control (CNC) machine (2010), and 2) bearings from platform PRONOSTIA (2012). Performances of the proposed approach are thoroughly compared with the classical approach by performing: feature fitness analysis, cutting tool wear "estimation" and bearings "long-term predictions" tasks, which validates the proposition.

Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at p... more Prognostic aims at estimating the remaining useful life (RU L) of a degrading equipment, i.e at predicting the life time at which a component or a system will be unable to perform a desired function. This task is achieved through essential steps of data acquisition, feature extraction and selection, and prognostic modeling. This paper emphasizes on the selection phase and aims at showing that it should be performed according to the predictability of features: as there is no interest in retaining features that are hard to be predicted. Thereby, predictability is defined and a feature selection procedure based on this concept is proposed. The effectiveness of the approach is judged by applying it on a real-world case: through comparison is made in order to show that the better predictable features lead to better RU L estimation.

2012 IEEE Conference on Prognostics and Health Management, 2012

Although efforts in this field are significant around the world, real prognostics systems are sti... more Although efforts in this field are significant around the world, real prognostics systems are still scarce in industry. Indeed, it is hard to provide efficient approaches that are able to handle with the inherent uncertainty of prognostics and nobody is able to a priori ensure that an accurate prognostic model can be built. As for an example of remaining problems, consider datadriven prognostics approaches: how to ensure that a model will be able to face with inputs variation with respect to those ones that have been learned, how to ensure that a learned-model will face with unknown data, how to ensure convergence of algorithms, etc. In other words, robustness, reliability and applicability of a prognostic approach are still open areas. Following that, the aim of this paper is to address these challenges by proposing a new neural network (structure and algorithm) that enhances reliability of RUL estimates while improving applicability of the approach. Robustness, reliability and applicability aspects are first discussed and defined according to literature. On this basis, a new connexionist system is proposed for prognostics: the Improved-Extreme Learning machine (Imp-ELM). This neural network, based on complex activation functions, enables to reduce the influence of human choices and initial parameterization, while improving accuracy of estimates and speeding the learning phase. The whole proposition is illustrated by performing tests on a real industrial case of cutting tools from a Computer Numerical Control (CNC) machine. This is achieved by predicting tool condition (wear) in terms of remaining cuts successfully made. Thorough comparisons with adaptive neuro fuzzy inference system (ANFIS) and existing ELM algorithm are also given. Results show improved robustness, reliability and applicability performances.