Fault detection of networked dynamical systems (NDSs) has attracted ever-increasing attention since it can maintain high-quality products as well as operational safety. Considering the utilisation ...

https://www.tandfonline.com/doi/pdf/10.1080/00207721.2021.1998722

Fault detection of networked dynamical systems: a survey of trends and techniques

Wireless Sensor Networks (WSNs) are vulnerable to faults because of their deployment in unpredictable and hazardous environments. This makes WSN prone to failures such as software, hardware, and communication failures. Due to the sensor’s limited resources and diverse deployment fields, fault detection in WSNs has become a daunting task. To solve this problem, Support Vector Machine (SVM), Convolutional Neural Network (CNN), Stochastic Gradient Descent (SGD), Multilayer Perceptron (MLP), Random Forest (RF), and Probabilistic Neural Network (PNN) classifiers are used for classification of gain, offset, spike, data loss, out of bounds, and stuck-at faults at the sensor level. Out of six faults, two of them are induced in the datasets, i.e., spike and data loss faults. The results are compared on the basis of their Detection Accuracy (DA), True Positive Rate (TPR), Matthews Correlation Coefficients (MCC), and F1-score. In this paper, a comparative analysis is performed among the classifiers mentioned previously on real-world datasets. Simulations show that the RF algorithm secures a better fault detection rate than the rest of the classifiers.

Fault Detection in Wireless Sensor Networks through the Random Forest Classifier

Autoencoder (AE) is a deep neural network that has been widely utilized in process industry owing to its superior abilities of feature extraction and data reconstruction. Recently, assuming the latent variables to be random variables, a probabilistic variant of it called variational autoencoder (VAE) has achieved a major success in different applications. In this article, we develop two novel submodels based on deep VAEs (DVAE), which are further utilized to establish a soft sensor framework. By the use of our first submodel known as supervised DVAE (SDVAE), the distribution information of latent features can be obtained. This is used as a prior of the second submodel known as the modified unsupervised DVAE (MUDVAE). Then, a new soft sensor framework can be constructed by combing the encoder of SDVAE with the decoder of MUDVAE. Since our designed VAE has superior ability in data reconstruction, it also works well under the missing data situation which is common in process industries due to sensor failures. Thus, we extend the proposed soft sensor framework to handle the missing data situation. The effectiveness of our proposed soft sensor frameworks is finally demonstrated via an industrial polymerization dataset.

Supervised Variational Autoencoders for Soft Sensor Modeling With Missing Data

A Tuned classification approach for efficient heterogeneous fault diagnosis in IoT-enabled WSN applications

This article investigates the problem of sliding mode control (SMC) for semi-Markov switching systems (S-MSSs) with quantized measurement in finite-time level. The transition between different subsystems obeys a stochastic semi-Markov process related to nonexponential distribution. Additionally, due to the sensor information constraints, the state vectors are not always measurable in practice. Moreover, compared with existing results in literature, the output quantization is first considered for finite-time SMC problem via a logarithmic quantizer. Our attention is to design an appropriate finite-time SMC law to attenuate the influences of parametrical uncertainty and external disturbance onto the overall performance of the system under consideration. First, by the key points of stochastic semi-Markov Lyapunov function and observer design theory, a desired SMC law is constructed to guarantee that the system trajectories can arrive at the specified sliding surface (SSS) within an assigned finite-time level. Then, ST-dependent sufficient conditions are established to ensure the required finite-time boundedness performance including both reaching phase and sliding motion phase. Finally, the applicability of the proposed results is demonstrated by a single-link robot arm model.

Finite-Time Observer-Based Sliding Mode Control for Quantized Semi-Markov Switching Systems With Application

In this paper, a distributed filtering scheme is presented to deal with the fault detection problem of nonlinear stochastic systems with wireless sensor networks (WSNs). The nonlinear stochastic systems, which are of discrete-time form, are represented by interval type-2 (IT2) Takagi–Sugeno (T–S) fuzzy models. Each sensor of the WSN can receive measurements from itself and its neighboring sensors subject to a deterministic interconnection topology. Independent random variables obeying the Bernoulli distribution are formulated to characterize the randomly occurred packet losses between the WSN and the filter unit. To generate residual signals for evaluation functions of the fault detection mechanism, a novel type of IT2 T–S fuzzy distributed fault detection filter is proposed corresponding to each sensor node. Additionally, a fault reference model is adopted for improving the performance of the fault detection system. A new overall fault detection system is formulated in an IT2 T–S fuzzy model framework. Applying Lyapunov functional approach, we concentrate on the analysis of stability and performance of the resulting fault detection system. New techniques are utilized to handle the decoupling problem in design procedure. The desired parametric matrices of the fuzzy filters are designed subject to a developed criterion, which is a sufficient condition of the robust mean-square asymptotic stability for the overall fault detection system with a disturbance attenuation performance. Finally, a truck-trailer system with a four-node WSN is established for simulation validation. In simulations, the mincx function of the MatLab 2017a in Windows 10 OS is used to optimize the level of the disturbance attenuation performance, and to obtain the filter gains for the established system. By comparing the different time instants when the residual evaluation functions exceed their respective thresholds, simulation results successfully validate the effectiveness and applicability of the presented distributed fault detection scheme.

Fu Xiao

Papers

Distributed Soft Fault Detection for Interval Type-2 Fuzzy-Model-Based Stochastic Systems With Wireless Sensor Networks