Rachid Outbib

A Generic Prognostic Methodology Using Damage Trajectory Models

In modern industries, there is intense pressure to continuously reduce costly, unscheduled maintenance of complex systems. To obtain high availability with reduced life cycle total ownership costs, classical maintenance policies are not optimal. Indeed, these polices do not allow us to perform maintenance only when it is necessary because they are not able to forecast system damage states in the future. To predict precisely the future system damage state, it is necessary to take into account how and where the system will be used. To build incremental damage models, this paper presents a generic methodology and formalism based on the system decomposition in three levels: environment, mission, and process. Predictions are performed via a sequence of known mission parameters, and environmental conditions. This allows for mission and maintenance planning by taking into account the predicted system damages over time.

Damage trajectory analysis based prognostic

To obtain high availability with reduced life cycle total ownership costs, classical maintenance policies are not sufficient. Indeed these polices do not allow us to maintain just when its necessary because they are not available to plan the current system damage state in the future. To predict accurately and precisely the future system damage state, it is necessary to take into account how and where the system is used in order to analyze the damage trajectory. The paper presents a methodology based on the system decomposition in three levels: environment, mission and process, to predict the future damage state by tracking its various damage trajectories and thus to know whether the system is able to accomplish its mission in time by using system current damage state and its future use.

A Result on State Estimation of Nonlinear Systems with Application to Fuel Cell Stacks

Abstract In this paper, the observation issue of the partial pressure of oxygen and nitrogen and the mass flow rate of dry air in the cathode channel of a fuel cell stack is addressed. The proposed approach considers the mass flow rate of dry air as an unknown input and uses the voltage and the total pressure as measurements. By using the Jacobian of the nonlinear functions and the convexity principe, the observer design problem is turned into a LMI feasibility problem. Simulation results with a detailed model show the good convergence properties of the observer.

A new scheme on robust unknown input nonlinear observer for PEM Fuel Cell Stack system

This paper presents an unknown input observer design method for a class of nonlinear systems in the presence of disturbances in both the systems dynamics and the output. The main motivation of this work is to develop a state estimator for PEM-Fuel Cell Stack systems. More precisely, the paper addresses the observation issue of the main variables of the fuel cell systems that are the compressor speed, the supply manifold pressure, the mass of the gas accumulated in the supply manifold volume, the partial pressure of oxygen and the partial pressure of nitrogen in the cathode channel. The proposed approach considers the current as an unknown input and uses the air flow rate through the compressor and the total pressure as measurements. By means of regular transformations, the Jacobian of the nonlinear functions and the convexity principle, the observer design problem is turned into a LMI feasibility problem. Simulation results with a refined model show the good convergence properties of the observer.

Functional observer for linear parameter-varying systems with application to diagnosis of PEM fuel cell

This paper is a contribution to the problem of functional observers for linear parameter-varying () systems. The goal is twofold. First, we propose an extension of some results of literature on functional observers, established originally for linear systems, to systems. The design of these observers are based on linear interpolation principle according to each parameter. Besides, sufficient conditions for asymptotic convergence are expressed in terms of linear matrix inequalities, easily tractable by convex optimisation techniques. Second, we show how this new result can be used to investigate the problem of diagnosis for proton electrolyte membrane fuel cell. More precisely, we deal with the problem of flooding considered as an important issue for water management inside the fuel cell. Finally, simulation results are given in order to highlight the performances of the proposed approach.

Observer design applied to prognosis of system

This paper is dedicated to model-based prognosis to predict remaining useful life of a system. This methodology is applied on multiple time scale systems made up of a slow and a fast dynamic behaviors subsystems, defining damage state and state of system behaviors respectively. Prediction of remaining useful life implies to have a slow dynamic behavior subsystem model. Slow dynamic subsystem behavior is supposed to be unknown, only the structure is assumed to be known a priori. For that, in the fast dynamic behavior subsystem, unmeasured state are estimated based on the design of unknown input observers. Slow dynamic behavior state in the fast dynamic behavior subsystem is led back to an unknown input. High gain observer is used to obtain accurate state and unknown input estimates. Slow dynamic behavior model parameters are then identified with the previous estimates. Prediction of remaining useful life is finally achieved based on relative accuracy of the observer estimates. Pertinence of the proposed methodology is illustrated based on simulation results to an electromechanical oscillator.

Estimation of Damage Behaviour for Model-Based Prognostic

Abstract In this paper, preliminary definitions on damage and damage prognosis will be given. A damage prognosis definition will be translated into a mathematical definition. The main contribution is to introduce a model-based prognosis methodology for reconstructing dynamics of cumulative damage state with only inputs and outputs measurements. The methodology is based essentially on the fact that damage will be lead back to an unknown input. The state will be reconstructed exactly in a predefined finite-time using a finite time unknown input observer. An expression of unknown input containing cumulative damage state information will be estimated. Then damage state estimation becomes an inverse problem. Finally, an identification process is designed to identify reconstructed dynamics of cumulative damage state with a class of nonlinear damage models. Simulation results on an academic automotive suspension system will be given to illustrate the pertinence of the approach.

A Data Driven Prognostic Methodology without a Priori Knowledge

Abstract Nowadays systems are more and more complex, there is intense pressure to continuously reduce and eliminate costly, unscheduled maintenance of these systems. In such case, using physics-based damage model is not adequate in term cost/benefit analysis. While, recent technological advances of new sensors, coupled with robust processing algorithms offer an elegant and theoretically sound approach to Condition-Based Maintenance (CBM)/Prognostic Health Management of such complex systems. A new strategy based on forecasting of system degradation through a prognostic data-driven method is required. This paper introduces the development of a data-driven methodology to predict remaining useful life (RUL) of an unspecified complex system. Remaining useful life prediction is performed by recent machine learning techniques without including any system or domain specific informations. The solution is efficient and easy to implement and has the potential to be applicable to a variety of complex systems (automobiles, aerospace systems).

State estimation for a class of time-varying linear systems: Application to fuel cell systems

Abstract This paper is a contribution to the problem of state estimation.More precisely, it concerns the case of time-varying linear systems. The goal is twofold. First to improve some results on state estimation. Second, to use this strategy to investigate the problem of diagnosis for Pi;M fuel cells process. Simulations results are given in order to illustrate the performances of the proposed approach.

Fault diagnosis and novel fault type detection for PEMFC system based on spherical-shaped multiple-class support vector machine

In this paper, a data-based strategy is proposed for PEMFC (polymer electrolyte membrane fuel cell) diagnosis. In the strategy, the feature extraction method Fisher Discriminant Analysis (FDA) is used firstly to extract the features from individual cell voltages. After that, the classification method Spherical-Shaped Multiple-class Support Vector Machine (SSM-SVM) is used to classify the extracted features to various classes related to health states. The potential novel failure mode can be detected in the procedure. Experiments on a 40-cell stack are dedicated to verify the approach.