Random wavelet kernels for interpretable fault diagnosis in industrial systems

"En el presente trabajo se presenta un estudio sobre la aplicación de estrategias bioinspiradas para la optimización al diagnóstico de fallos en sistemas industriales. El objetivo principal es establecer una base para el desarrollo de nuevos y viables métodos de diagnóstico de fallos basados en modelos que permitan mejorar las dificultades de los métodos actuales. Estas dificultades están relacionadas, fundamentalmente, con la sensibilidad ante la presencia de fallos y la robustez ante perturbaciones externas. En el estudio se consideraron los algoritmos Evolución Diferencial y Optimización por Colonia de Hormigas. La efectividad de la propuesta es analizada mediante experimentos con el conocido problema de prueba de los dos tanques. Los experimentos consideraron presencia de ruido en la información y fallos incipientes de manera que fuera posible analizar las ventajas de la propuesta en cuanto a diagnóstico robusto y sensible. Los resultados obtenidos indican que el enfoque propuesto y, principalmente, la combinación de los dos algoritmos, caracterizan una metodología prometedora para el diagnóstico de fallos."

Most of the existing artificial neural network models use a significant amount of information for their training. The need for such information could be an inconvenience for its application in fault diagnosis in industrial systems, where the information, due to different factors such as data losses in the data acquisition systems, is scarce or not verified. In this chapter, a diagnostic system based on a Hopfield neural network is proposed to overcome this inconvenience. The proposal is tested using the development and application of methods for the actuator diagnostic in industrial control systems (DAMADICS) benchmark, with successful performance.

This chapter presents the motivation for developing fault diagnosis application in industrial systems. Fault diagnosis methods can be broadly categorized into model-based and data-driven. Model-based strategies are briefly discussed while highlighting the increasing tendency to the use of data-driven methods given the increasing data available from process operations. The classic data-driven fault diagnosis loop is presented and each task is described in detail. A procedure is presented for the systematic design of data driven fault diagnosis methods. Finally, the fault diagnosis problem for multimode processes is briefly discussed.

A method for fault diagnosis of industrial systems is presented. Plant devices, sensors, actuators and diagnostic tests are described as stochastic finite-state machines. A formal composition rule of these models is given to obtain: 1) the set of admissible fault signatures; 2) their conditional probability given any fault; and 3) the conditional probability of a fault given a prescribed signature. The modularity and flexibility of this method make it suitable to deal with complex systems made by a large number of components. The method is used in an industrial automotive application, specifically the diagnosis of the throttle body and of the angular sensors measuring the throttle plate angle is described in detail.

A variant of the particle swarm optimization for the improvement of fault diagnosis in industrial systems via faults estimation

Self-adaptive selection graph pooling based fault diagnosis method under few samples and noisy environment

In this work we present a study on the application of bio-inspired strategies for optimization to Fault Diagnosis in industrial systems. The principal aim is to establish a basis for the development of new and viable model-based Fault Diagnosis Methods which improve some difficulties that the current methods cannot avoid. These difficulties are related with fault sensitivity and robustness to external disturbances. To get start the study, we consider the Differential Evolution and the Ant Colony Optimization algorithms. This application is illustrated using simulation data of the Two Tanks System benchmark. In order to analyze the merits of these algorithms to obtain a diagnosis which needs to be sensitive to faults and robust to external disturbances, some experiments with incipient faults and noisy data have been simulated. The results indicate that the proposed approach, basically the combination of the two algorithms, characterizes a promising methodology for Fault Diagnosis.

The paper presents the application of various classification schemes for actuator fault diagnosis in industrial systems. The main objective of this study is to compare either single or meta-classification strategies that can be successfully used as reasoning means in off-line as well as on-line diagnostic expert systems. The applied research was conducted on the assumption that only classic and well-practised classification methods would be adopted. The comparison study was carried out within the DAMADICS benchmark problem which provides a popular framework for confronting different approaches in the development of fault diagnosis systems.

Online automatic fault diagnosis in industrial systems is essential for guaranteeing safe, reliable and efficient operations. However, difficulties associated with computational overload, ubiquitous uncertainties and insufficient fault samples hamper the engineering application of intelligent fault diagnosis technology. Geared towards the settlement of these problems, this paper introduces the method of dynamic uncertain causality graph, which is a new attempt to model complex behaviors of real-world systems under uncertainties. The visual representation to causality pathways and self-relied “chaining” inference mechanisms are analyzed. In particular, some solutions are investigated for the diagnostic reasoning algorithm to aim at reducing its computational complexity and improving the robustness to potential losses and imprecisions in observations. To evaluate the effectiveness and performance of this method, experiments are conducted using both synthetic calculation cases and generator faults of a nuclear power plant. The results manifest the high diagnostic accuracy and efficiency, suggesting its practical significance in large-scale industrial applications.

Valves and other actuators may fail and cause economic losses or safety accidents. To ensure the stable operation of a control system, it is necessary to identify the failures of various valves and carry out the corresponding maintenance. Several methods are designed and implemented for valve fault diagnosis in this paper. In particular, a novel fault diagnosis method based on a time-series convolution network (FDM-TSCN) is proposed, which is built on a time-series data feature extracting and convolutional neural network. FDM-TSCN can classify 18 out of 19 types of fault, while many other methods cannot. This algorithm is presented in detail and implemented as a prototype system. Comprehensive simulations are performed on valve fault datasets that are generated by the development and application of methods for actuator fault diagnosis in industrial systems (DAMADICS). The simulation results prove the effectiveness and superiority of the proposed FDM-TSCN method. All of the source codes and related data in the paper are made available, which enables other researchers to verify the work easily and may inspire them to carry out more informed research.

With the lack of failure data, class imbalance has become a common challenge in the fault diagnosis of industrial systems. The oversampling methods can tackle the class-imbalanced problem by generating the minority samples to balance the training set. However, one of the main challenges of the existing oversampling methods is how to generate high-quality minority samples. Traditional oversampling methods regard all synthetic samples as minority ones to be added to the training set without filtering. The low-quality synthetic samples would distort the distribution of the dataset and worsen the classification performance. In this article, we propose a weakly supervised oversampling method that treats all synthetic samples as unlabeled samples and develops a graph semisupervised learning algorithm to select high-quality synthetic samples, adding into the final training set as minority samples. To improve the quality of synthetic samples, we propose a cost-sensitive neighborhood component analysis dimensionality reduction method to enhance domain information validity in high-dimensional datasets. Finally, combining a boosting-based ensemble framework, we propose a new imbalanced learning framework suitable for high dimensionality and highly imbalanced fault diagnosis in industrial systems. The experimental validation is performed on five real-world wind turbine blade cracking failure datasets and compared to 15 benchmark methods. The experimental results show that average performances and robustness of the proposed framework are significantly better than those of the benchmark methods.

International audience

This paper explores the application of bioinspired cooperative strategies for optimization on Fault Diagnosis in industrial systems. As a first step, the Differential Evolution and Ant Colony Optimization algorithms are considered. Both algorithms have been applied to a benchmark problem, the two tanks system. The experiments have considered noisy data in order to compare the robustness of the diagnosis. The preliminary results indicate that the proposed approach, basically the combination of the two algorithms, characterizes a promising methodology for the Fault Detection and Isolation problem.KeywordsDifferential EvolutionFault DiagnosisDifferential Evolution AlgorithmIndustrial SystemTank SystemThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Fault Diagnosis on Industrial Systems Based on a Multiple Model Approach

The paper presents the application of various classification schemes for actuator fault diagnosis in industrial systems. The main objective of this study is to compare either single or meta-classification strategies that can be successfully used as reasoning means in the diagnostic expert system that is realized within the frame of the DISESOR project. The applied research was conducted on the assumption that classic as well as soft computing classification methods would be adopted. The comparison study was carried out within the DAMADICS benchmark problem which provides a popular framework for confronting different approaches in the development of fault diagnosis systems.KeywordsFault DetectionFault DiagnosisBase ClassifierSingle ClassifierRule InductionThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.