Fault Diagnosis in Dynamic Systems Using Recurrent Neural Networks

There is an increasing demand for dynamic systems to become safer, more reliable and more economical in operation. This requirement extends beyond the normally accepted safety-critical systems e.g., nuclear reactors, aircraft and many chemical processes, to systems such as autonomous vehicles and some process control systems where the system availability is vital. The field of fault diagnosis for dynamic systems (including fault detection and isolation) has become an important topic of research. Many applications of qualitative and quantitative modelling, statistical processing and neural networks are now being planned and developed in complex engineering systems. Issues of Fault Diagnosis for Dynamic Systems has been prepared by experts in fault detection and isolation (FDI) and fault diagnosis with wide ranging experience.Subjects featured include: - Real plant application studies; - Non-linear observer methods; - Robust approaches to FDI; - The use of parity equations; - Statistical process monitoring; - Qualitative modelling for diagnosis; - Parameter estimation approaches to FDI; - Fault diagnosis for descriptor systems; - FDI in inertial navigation; - Stuctured approaches to FDI; - Change detection methods; - Bio-medical studies. Researchers and industrial experts will appreciate the combination of practical issues and mathematical theory with many examples. Control engineers will profit from the application studies.

This paper suggests an approach for fault detection and diagnosis capable to detect new operation modes online. The approach relies upon an evolving fuzzy classifier able to incorporate new operational information using an incremental unsupervised clustering procedure. The efficiency of the approach is verified in fault detection and diagnosis of an induction machine. Experimental results suggest that the approach is a promising alternative for fault diagnosis of dynamic systems when there is no a priori information about all failure modes. It is also attractive for incremental learning of diagnosis systems with streams of data.KeywordsEvolving Fuzzy SystemsParticipatory LearningAdaptive Fault Detection and Diagnosis

Fault Detection and Diagnosis (FDD) in complex industrial systems requires methods that can handle uncertain operating conditions, soft thresholds, evolving sensor behavior, and increasing volumes of heterogeneous data. Traditional model-based or rule-driven approaches offer interpretability but lack adaptability, while purely data-driven and Large Language Model (LLM)-based methods often struggle with consistency, traceability, and causal grounding. Dynamic Master Logic (DML) provides a causal and temporal reasoning structure with fuzzy rules that capture gradual drift, soft limits, and asynchronous sensor signals while preserving traceability and deterministic evidence propagation. Building on this foundation, this paper presents a DML-LLM hybrid architecture that integrates targeted LLM inference to interpret unstructured information such as logs, notes, or retrieved documents under controlled prompts that maintain domain constraints. The combined system integrates Bayesian updating, deterministic routing, and semantic interpretation into a unified FDD pipeline. In a semiconductor manufacturing case study, the proposed framework reduced time to detection (TTD) from 7.4 h to 1.2 h and improved the F1 score from 0.59 to 0.83 when compared with conventional Statistical Process Control (SPC) and Fault Detection and Classification (FDC) workflows. Provenance completeness increased from 18% to 96%, while engineer triage time was reduced from 72 min to 18 min per event. These results demonstrate that the hybrid framework provides a scalable and explainable approach to anomaly detection and fault diagnosis in sensor-rich industrial environments.

Detection of incipient (slowly developing) faults is crucial in automated maintenance problems where early detection of worn equipment is required. In this paper, a general framework for model-based fault detection and diagnosis of a class of incipient faults is developed. The changes in the system dynamics due to the fault are modeled as nonlinear functions of the state and input variables, while the time profile of the failure is assumed to be exponentially developing. An automated fault diagnosis architecture using nonlinear online approximators with an adaptation scheme is designed and analyzed. A simulation example of a simple nonlinear mass-spring system is used to illustrate the results.

We address the feasibility of the pragmatic implementation of monitoring systems for real-time distributed fault diagnosis in complex processes. We delve into the integration of theoretical distributed estimation methods and practical distributed embedded systems. This study emphasizes the merger of distributed computing and signal processing to augment fault diagnosis in dynamic systems. We introduce a generic mathematical model for distributed fault diagnosis algorithms, laying the groundwork for a reference network computing architecture that details the essential components, organization, and operational characteristics necessary for distributed monitoring systems. Subsequently, we evaluate existing software and hardware solutions that facilitate the realization of these systems. The theoretical framework and system design are empirically validated through an experimental setup involving a liquid-level control system, demonstrating the efficacy and applicability of our approach.

Monte Carlo simulation for model-based fault diagnosis in dynamic systems

In this chapter, selected aspects of the fault diagnosis problem for control systems have been considered with special attention paid to fault detection and, to a lesser extent, fault isolation. In particular, the robustness problem of FDI with respect to the requirement to maximise sensitivity to faults while minimising (or decoupling) the effect of uncertain effects through unknown inputs has been described. This severe robustness challenge accompanying analytical methods of FDI has led to the requirement of developing new methodologies for FDI in which analytical and CI techniques are combined to achieve good global and nonlinear modelling and robustness. The chapter has outlined some of the recent research on FDI and fault diagnosis for dynamic systems, using this integrated approach. The examples have shown that by using evolutionary computing, neural networks and NF modelling structures realistic solutions are achievable. It is hoped that this direction of research will stimulate an increased adoption of real industrial application to make mode-based FDI more usual and effective in real process systems.

The problem of fault detection and diagnosis (FDD) in dynamic systems has received considerable attention in last decades due to the growing complexity of modern engineering systems and ever increasing demand for fault tolerance, cost efficiency, and reliability (Willsky, 1976; Basseville, 1988). Existing FDD approaches can be roughly divided into two major categories including model-based and knowledge-based approaches (Venkatasubramanian et al., 2003a; Venkatasubramanian et al., 2003b). Model-based approaches make use of the quantitative analytical model of a physical system. Knowledgebased approaches do not need full analytical modeling and allow one to use qualitative models based on the available information and knowledge of a physical system. Whenever the mathematical models describing the system are available, analytical model-based methods are preferred because they are more amenable to performance analysis. Generally, there are two steps in the procedure of model-based FDD. First, on the basis of the available observations and a mathematical model of the system, the state variable x and test statistics are required to be obtained. Then, based on the generated test statistics, it is required to decide on the potential occurrence of a fault. For linear and Gaussian systems, the Kalman filter (KF) is known to be optimal and employed for state estimation. The innovations from the KF are used as the test statistics, based on which hypothesis tests can be carried out for fault detection (Belcastro & Weinstein, 2002). In reality, however, the models representing the evolution of the system and the noise in observations typically exhibit complex nonlinearity and non-Gaussian distributions, thus precluding analytical solution. One popular strategy for estimating the state of such a system as a set of observations becomes available online is to use sequential Monte-Carlo (SMC) methods, also known as particle filters (PFs) (Doucet et al., 2001). These methods allow for a complete representation of the posterior probability distribution function (PDF) of the states by particles (Guo & Wang, 2004; Li & Kadirkamanathan, 2001). The aforementioned FDD strategies are single-model-based. However, a single-model-based FDD approach is not adequate to handle complex failure scenarios. One way to treat this

This paper investigates the problem of fault detection and diagnosis (FDD) problem for non-gausian singulat stochastic distribution control (SDC) systems via the output probability density functions(PDFs). The PDFs can be approximated by using square-root B-spline expansion,via this expansions to represent the dynamics weighting systems between the system input and the weights related to the output PDFs. In this works ,an optimal fault detection and diagnosis algorith is present by introducing the parameter-updating .when the fault occurs, an adaptive network parameter-updating law is designed to approximated the fault. Finally, the simulation result are given to show that the approach can detect faults and estimate the size of fault.

On-Line Residual Compensation in Robust Fault Diagnosis of Dynamic Systems

This paper presents a novel approach for Fault detection and diagnosis (FDD) using distributed Neural Networks (DNN) to enhance system performance. FDD plays an important role in industrial applications, to guarantee system performance by detecting and analyzing faults to prevent system failures. Various approaches have been adopted for FDD, among which Artificial Intelligence (AI) has shown promising results. Artificial Neural Networks (ANN), a key AI approach, are extensively utilized nowadays in FDD to monitor system efficiency. This paper introduces a DNN approach that determines the model structure and estimates the results using the Back Propagation (BP) Algorithm. The DNN where used in this work not only to identify multiple faults for DC motors (e.g., broken bar, dynamic strangeness… etc.) but also to determine fault type and location inside a distributed drive network. Several tests of DC engine faults were conducted to assess the execution of the proposed approach. The results revealed a significant improvement in system performance using the proposed approach compared to the multilayer perceptron (MLP) method. The proposed approach enhanced the system's stability and performance by reducing faults from the default value.

Optimal auxiliary input design for fault diagnosis

Cyber-physical systems such as autonomous spacecraft, power plants and automotive systems become more vulnerable to unanticipated failures as their complexity increases. Accurate tracking of system dynamics and fault diagnosis are essential. This paper presents an efficient state estimation method for dynamic systems modeled as concurrent probabilistic automata. First, the Labeled Uncertainty Graph (LUG) method in the planning domain is introduced to describe the state tracking and fault diagnosis processes. Because the system model is probabilistic, the Monte Carlo technique is employed to sample the probability distribution of belief states. In addition, to address the sample impoverishment problem, an innovative look-ahead technique is proposed to recursively generate most likely belief states without exhaustively checking all possible successor modes. The overall algorithms incorporate two major steps: a roll-forward process that estimates system state and identifies faults, and a roll-backward process that analyzes possible system trajectories once the faults have been detected. We demonstrate the effectiveness of this approach by applying it to a real world domain: the power supply control unit of a spacecraft.

Three methods for the fault diagnosis of dynamic systems are described. The first method adopts the multiple hypothesis test to locate the faulty parameter from the deviations of identified system models. This method can be applied not only to a single parameter fault but also to multiple faults. The second method combines this numerical method with the intelligent method of expert systems. The third adopts multilayer-type neural networks to identify the fault parameter. The neural network makes possible a simpler and quicker diagnosis method. The performances of these diagnosis methods have been tested by simulations.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

In recent years, the data-driven based FDD (Fault Detection and Diagnosis) of high-speed train electric traction systems has made rapid progress, as the safe operation of traction system is closely related to the reliability and stability of high-speed trains. The internal complexity and external complexity of the environment mean that fault diagnosis of high-speed train traction system faces great challenges. In this paper, a wavelet transform-based FNR (Fault to Noise Ratio) enhancement is realised to highlight incipient fault information and a Deep PCA (Principal Component Analysis)-based diagnosability analysis framework is proposed. First, a scheme for FNR enhancement-based fault data preprocessing with selection of the intelligent decomposition levels and optimal noise threshold is proposed. Second, fault information enhancement technology based on continuous wavelet transform is proposed from the perspective of energy. Further, a Deep-PCA based incipient fault detectability and isolatability analysis are provided via geometric descriptions. Finally, experiments on the TDCS-FIB (Traction Drive Control System–Fault Injection Benchmark) platform fully demonstrate the effectiveness of the method proposed in this paper.