A novel densely connected neural network for proton exchange membrane fuel cell fault diagnosis

It is of great significance to perform proton exchange membrane fuel cell (PEMFC) fault diagnosis and take action timely to mitigate or even eliminate the faults, which can strengthen PEMFC reliability and durability. In previous studies, cell voltage is extensively used for PEMFC fault diagnosis. However, there exists similar cell voltage drop phenomenon as different PEMFC faults occur, especially for faults like flooding and air starvation having extremely similar voltage dynamic variation, which makes it difficult to capture the features sensitive to faults. Moreover, cell voltages collected from different MEAs follow different distributions even in the same operation condition, which challenges the diagnosis consistency of fault diagnosis methods. In this paper, in order to break through the hindrances, a novel densely connected neural network codenamed Inc-DenseNet is proposed for PEMFC fault diagnosis, which integrates advantages of InceptionNet and DenseNet to extract more specific and robust features from cell voltage. In the analysis, the collected PEMFC voltage signal is transformed into 2D image data, which is then used to train the Inc-DenseNet. Results demonstrate that with the trained Inc-DenseNet, the diagnostic accuracy for four PEMFC states of health (normal, flooding, dehydration, air starvation) can reach 95.3%, especially for flooding and air starvation. In addition, by using the voltage datasets collected from two different MEAs, the generalization capacity of the Inc-DenseNet is proved. With the findings, the proposed network Inc-DenseNet can not only achieve high-precision fault diagnosis, but also has a high computing efficiency, which makes it promising in real-time PEMFC fault diagnosis in the future.     

Sensor fault estimation of PEM fuel cells using Takagi Sugeno fuzzy model

This paper presents a sensor fault estimation scheme for polymer electrolyte membrane (PEM) fuel cells using Takagi Sugeno (TS) fuzzy model. First, PEM fuel cell systems with sensor faults are modelled by TS fuzzy model. Next, by adding a first order filter, an augmented TS fuzzy system with actuator fault is obtained. Then, for the augmented system, an unknown input observer (UIO) and a fault estimator are developed. The UIO gains are computed by solving linear matrix equalities (LMEs) and linear matrix inequalities (LMIs). The UIO convergence and stability are analyzed and the performances of the proposed fault estimation scheme is demonstrated by numerical simulations for a PEM fuel cell system with return manifold pressure and hydrogen mass sensors.     

LPV observer design for PEM fuel cell system: Application to fault detection

In this paper, the modelling of an energy generation system based on polymer electrolyte membrane fuel cell (PEMFC) system through a parameter varying approach (LPV model), that takes in to account model parameter variation with the operating point, is presented. This model has been obtained through a Jacobian linearization of the PEMFC non-linear dynamic model that was previously calibrated using real data from lab. In order to illustrate the use of the LPV model obtained its application to model-based fault detection is used. For this purposes a set of common fault scenarios, which could appear during a normal PEMFC operation, is used as case study.     

Actuator fault detection for a PEMFC system based on delta operator approach

This paper proposes an actuator fault detection observer design method for a proton exchange membrane fuel cell (PEMFC) system described by delta operator form. In the first, a nonlinear mathematical model for the PEMFC system is proposed by using the mechanism analysis method, which includes the time delay, external disturbance and uncertainty to reduce the conservatism of the design. Then, in order to detect the actuator fault, the functional observer is constructed as the residual generator, and sufficient conditions for the stability of the error system are obtained by using the Lyapunov-krasovsky functional theory in terms of linear matrix inequalities (LMIs), so as to ensure the design of the required functional observer. Finally, simulation results are provided to show the effectiveness of the approach.     

Fuzzy observer-based fault tolerant control against sensor faults for proton exchange membrane fuel cells

In this paper, robust fault diagnosis problem of Proton Exchange Membrane Fuel Cells (PEMFC) is presented based on Takagi-Sugeno (TS) Fuzzy Unknown Input Observer (FUIO). TS FUIO based on Linear Matrix Equalities (LMEs) and the Linear Matrix Inequalities (LMIs) are design. Firstly, the nonlinear PEMFC system with sensor faults and disturbance is represented by TS fuzzy model. Then, a FUIO and sensor fault estimation algorithm is developed and then a model based Fuzzy Fault Tolerant Controller design uses the concept of Parallel Distributed Compensation (PDC). Sufficient stability conditions are studied based on LMIs and LMEs. In order to verify the proposed approach, a PEMFC system with return manifold pressure and hydrogen mass sensors fault and disturbance was tested to illustrate the effectiveness of the proposed strategy.     

A Bayesian network fault diagnostic system for proton exchange membrane fuel cells

This paper considers the effects of different types of faults on a proton exchange membrane fuel cell model (PEMFC). Using databases (which record the fault effects) and probabilistic methods (such as the Bayesian-Score and Markov Chain Monte Carlo), a graphical–probabilistic structure for fault diagnosis is constructed. The graphical model defines the cause-effect relationship among the variables, and the probabilistic method captures the numerical dependence among these variables. Finally, the Bayesian network (i.e. the graphical–probabilistic structure) is used to execute the diagnosis of fault causes in the PEMFC model based on the effects observed.     

Data-driven fault diagnosis for PEMFC systems of hybrid tram based on deep learning

The running state of the hybrid tram and the service life of fuel cell stacks are related to the fault diagnosis strategy of the proton exchange membrane fuel cell (PEMFC) system. In order to accurately detect various fault types, a novel method is proposed to classify the different health states, which is composed of simulated annealing genetic algorithm fuzzy c-means clustering (SAGAFCM) and deep belief network (DBN) combined with synthetic minority over-sampling technique (SMOTE). Operation data generated by the tram are clustered by SAGAFCM algorithm, and valid data are selected as fault diagnosis samples which include the training sample and the test sample. However, the fault samples are usually unbalanced data. To reduce the influence of unbalanced data on the fault diagnosis accuracy, SMOTE is employed to form a new training sample by supplementing the data of the small sample. Then DBN is trained by the new training sample to obtain the fault diagnosis model. In this paper, the proposed method can well distinguish the four health states, which are high deionized water inlet temperature fault, hydrogen leakage fault, low air pressure fault and the normal state, with an accuracy of 99.97% for the training sample and 100% for the test sample.     

Wind turbine asymmetrical load reduction with pitch sensor fault compensation

Offshore wind turbines suffer from asymmetrical loading (blades, tower, etc), leading to enhanced structural fatigue. As well as asymmetrical loading different faults (pitch system faults etc.) can occur simultaneously, causing degradation of load mitigation performance. Individual pitch control (IPC) can achieve rotor asymmetric loads mitigation, but this is accompanied by an enhancement of pitch movements leading to the increased possibility of pitch system faults, which exerts negative effects on the IPC performance. The combined effects of asymmetrical blade and tower bending together with pitch sensor faults are considered as a “co‐design” problem to minimize performance deterioration and enhance wind turbine sustainability. The essential concept is to attempt to account for all the “fault effects” in the rotor and tower systems, which can weaken the load reduction performance through IPC. Pitch sensor faults are compensated by the proposed fault‐tolerant control (FTC) strategy to attenuate the fault effects acting in the control system. The work thus constitutes a combination of IPC‐based load mitigation and FTC acting at the pitch system level. A linear quadratic regulator (LQR)‐based IPC strategy for simultaneous blade and tower loading mitigation is proposed in which the robust fault estimation is achieved using an unknown input observer (UIO), considering four different pitch sensor faults. The analysis of the combined UIO‐based FTC scheme with the LQR‐based IPC is shown to verify the robustness and effectiveness of these two systems acting together and separately.     

Investigation of PEMFC fault diagnosis with consideration of sensor reliability

Despite the wide range of applications for the polymer electrolyte membrane fuel cell (PEMFC), its reliability and durability are still major barriers for further commercialization. As a possible solution, PEMFC fault diagnosis has received much more attention in the last few decades. Due to the difficulty of developing an accurate PEMFC model incorporating various failure mode effects, data-driven approaches are widely used for diagnosis purposes. These methods depend largely on the quality of sensor measurements from the PEMFC. Therefore, it is necessary to investigate sensor reliability when performing PEMFC fault diagnosis.In this study, sensor reliability is investigated by proposing an identification technique to detect abnormal sensors during PEMFC operation. The identified abnormal sensors will be removed from the analysis in order to guarantee reliable diagnostic performance. Moreover, the effectiveness of the proposed technique is investigated using test data from a PEMFC system, where fuel cell flooding is observed. During the test, due to accumulation of liquid water inside the PEMFC, the humidity sensors will give misleading readings, and flooding cannot be identified correctly with inclusion of these humidity sensors in the analysis. With the proposed technique, the abnormal humidity measurements can be detected at an early stage. Results demonstrate that by removing the abnormal sensors, flooding can be identified with the remaining sensors, thus reliable health monitoring can be guaranteed during the PEMFC operation.     

Simultaneous fault diagnosis of proton exchange membrane fuel cell systems based on an Incremental Multi-label Classification Network

Fault diagnosis plays an important role in the operation of proton exchange membrane fuel cell (PEMFC) systems. In some certain working conditions, multiple faults can occur simultaneously. And, to the best of our knowledge, very few studies have yet to design an algorithm specifically for simultaneous fault diagnosis in PEMFC systems. Therefore, a novel simultaneous fault diagnosis algorithm, based on multi-label classifier chain named Incremental Multi-label Classification Network (IMCN), is proposed. To develop and optimize IMCN, a PEMFC model is constructed based on the commercial software AVL CURISE M to simulate data when simultaneous multiple faults occur. To further verify the generalization performance and practical effect of IMCN, a bench experiment using a high-power PEMFC system is conducted, which has similar boundary conditions as the boundary conditions embedded in simulation model. And, whether in experiment or simulation, corresponding verification methods are adopted to verify the success of simultaneous multiple faults embedding. Experimental data testing shows that, the subset accuracy, Hamming loss, Jaccard index, precision and recall of IMCN reaches 0.973, 0.029, 0.921, 0.961 and 0.956 respectively (better than Multi-Label MLP classifier, Label powerset MLP classifier, etc.), and the proposed simultaneous fault diagnosis method has achieved excellent results.     

Sensor and actuator fault diagnosis for wind turbine systems by using robust observer and filter

In this paper, we consider sensor and actuator fault detection and estimation issues for large scale wind turbine systems where individual pitch control (IPC) is used for load reduction. The faults considered are the blade root bending moment sensor faults and blade pitch actuator faults. In the first part, with the aid of a dynamical model of the wind turbine system, a so‐called H_∞/H_? observer in the finite frequency range, is applied to generate the residual for fault detection. The observer is designed to be sensitive to faults but insensitive to disturbances, such as wind turbulence. When there is a detectable fault, the observer sends an alarm signal if the residual evaluation is larger than a predefined threshold. In addition to the fault detection, we also consider the fault estimation problem, where a dynamic filter is used to estimate the fault magnitude. The effectiveness of the proposed approach is demonstrated by simulation results for several fault scenarios. Copyright © 2010 John Wiley & Sons, Ltd.     

Model-based fault diagnosis in PEM fuel cell systems

In this work, a model-based fault diagnosis methodology for PEM fuel cell systems is presented. The methodology is based on computing residuals, indicators that are obtained comparing measured inputs and outputs with analytical relationships, which are obtained by system modelling. The innovation of this methodology is based on the characterization of the relative residual fault sensitivity. To illustrate the results, a non-linear fuel cell simulator proposed in the literature is used, with modifications, to include a set of fault scenarios proposed in this work. Finally, it is presented the diagnosis results corresponding to these fault scenarios. It is remarkable that with this methodology it is possible to diagnose and isolate all the faults in the proposed set in contrast with other well known methodologies which use the binary signature matrix of analytical residuals and faults.     

A discrete hidden Markov model fault diagnosis strategy based on K-means clustering dedicated to PEM fuel cell systems of tramways

To solve the fault classification problems of fuel cell (FC) various health states for tramways, a discrete hidden Markov model (DHMM) fault diagnosis strategy based on K-means clustering is proposed. In this paper, the K-means clustering algorithm is used to filter the sample points which aren't consistent with the actual class labels. The Lloyd algorithm is employed to quantify the sample vector sets and obtain the discrete code combination of training samples and test samples. The Baum-Welch algorithm and forward-backward algorithm are respectively presented to train and deduce the DHMM. The classification results show that the six concerned faults can be detected and isolated. The targeted fault types include low air pressure, deionized glycol high inlet temperature, deionized humidification pump low pressure, deionized glycol outlet temperature signal voltage overrange, normal state and hydrogen leakage. The fault recognition rates with the novel approach are at best 94.17%.     

Reformer faults in SOFC systems: Experimental and modeling analysis,and simulated fault maps

The effects of fuel processor faults in an solid oxide fuel cell (SOFC) system are analyzed. Focusing on a laboratory-size SOFC system, a reformer fault is investigated both experimentally and through a model; comparison between experimental and modeling results is presented and discussed. The results show that some types of reformer faults can be dangerous, because they can give rise to local thermal gradients as large as 10–20·10² K/m or more in the SOFC stack. Simulation results show that SOFC stacks employing metallic interconnects are expected to withstand faults of larger magnitude than SOFC stacks employing ceramic interconnects. Fault maps are presented and discussed, which can be the basis for the development of a fault detection and isolation (FDI) tool.     

PEMFC stack voltage singularity measurement and fault classification

The study summarized in this paper deals with non-intrusive fault diagnosis of Polymer Electrolyte Membrane Fuel Cell (PEMFC) stack. In the proposed approach, the diagnosis operation is based on the stack voltage singularity measurement and classification. To this aim, wavelet transform-based multifractal formalism, named WTMM (Wavelet Transform Modulus Maxima), and pattern recognition methods are combined to realize the identification of the PEMFC faults. The proposed method takes advantage of the non-linearities associated with discontinuities introduced in the dynamic response data resulting from various failure modes. Indeed, the singularities signature of poor operating conditions (faults) of the PEMFC is revealed through the computing of multifractal spectra. The obtained good classification rates demonstrate that the multifractal spectrum based on WTMM is effective to extract the incipient fault features during the PEMFC operation. The proposed method leads to a promising non-intrusive and low cost diagnostic tool to achieve on-line characterizations of dynamical FC behaviors.     

On-line fault diagnostic system for proton exchange membrane fuel cells

In this paper, a supervisor system, able to diagnose different types of faults during the operation of a proton exchange membrane fuel cell is introduced. The diagnosis is developed by applying Bayesian networks, which qualify and quantify the cause–effect relationship among the variables of the process. The fault diagnosis is based on the on-line monitoring of variables easy to measure in the machine such as voltage, electric current, and temperature. The equipment is a fuel cell system which can operate even when a fault occurs. The fault effects are based on experiments on the fault tolerant fuel cell, which are reproduced in a fuel cell model. A database of fault records is constructed from the fuel cell model, improving the generation time and avoiding permanent damage to the equipment.     

An adaptive sliding mode observer based near-optimal OER tracking control approach for PEMFC under dynamic operation condition

Tracking control of oxygen excess ratio (OER) is crucial for dynamic performance and operating efficiency of the proton exchange membrane fuel cell (PEMFC). OER tracking errors and overshoots under dynamic load limit the PEMFC output power performance, and also could lead oxygen starvation which seriously affect the life of PEMFC. To solve this problem, an adaptive sliding mode observer based near-optimal OER tracking control approach is proposed in this paper. According to real time load demand, a dynamic OER optimization strategy is designed to obtain an optimal OER. A nonlinear system model based near-optimal controller is designed to minimize the OER tracking error under variable operation condition of PEMFC. An adaptive sliding mode observer is utilized to estimate the uncertain parameters of the PEMFC air supply system and update parameters in near-optimal controller. The proposed control approach is implemented in OER tracking experiments based on air supply system of a 5 kW PEMFC test platform. The experiment results are analyzed and demonstrate the efficacy of the proposed control approach under load changes, external disturbances and parameter uncertainties of PEFMC system.     

发动机异响故障的模糊专家诊断方法研究

基于模糊综合评价诊断方法,建立了发动机异响故障的模糊综合诊断模型,并开发了计算机辅助诊断系统.研究结果表明,发动机异响故障的模糊综合诊断模型充分考虑了实际诊断中的各种因素,开发的计算机辅助诊断系统可以为检测诊断人员指明方向,提高发动机异响故障诊断的工作效率,具有一定的应用价值.