Design of Protocol-Based Finite-Time Memory Fault Detection Scheme With Circuit System Application

The protocol-based memory fault detection (FD) problem under finite-time constraint is investigated for nonlinear networked systems, where the round-robin protocol is adopted to identify which sensor node can release the measured value at each transmission instant. In particular, the past states of the filter are exploited to develop the memory fault detection filter (FDF). Subsequently, the Finsler lemma and the Lyapunov approach are combined to analyze the finite-time stability of the addressed system. Then, the specific expression of the memory FDF gain matrices is obtained by using the solutions to certain matrix inequalities. In addition, for the aim of verifying that the memory FD approach has higher detection accuracy, a memoryless FDF is also designed in comparison with the memory FDF. In the end, the illustrative simulation studies are given to demonstrate the applicability of developed FD filtering algorithms in resistance–inductance–capacitance circuit system and the superiority of the designed memory FD scheme.

I. Introduction

With the development of science and technology, the structure of large-scale automatic equipment becomes more complex and the degree of precision is getting higher accordingly, which improves the production efficiency. However, the performance of the equipment will inevitably decline with the increase of the system uptime, which will lead to decrease or loss of the equipment expected function, or even malfunctioning [1], [2]. Once the fault occurs, it will affect the product quality, reduce the production efficiency, cause property damage, and life-threatening [3], [4]. Consequently, effective FD is essential to guarantee reliable operation of the systems. Up to date, various FD methods have been presented in available literature [5], [6]. Nevertheless, notice that most FD results are developed based on the memoryless filtering algorithms. In order to reduce the conservativeness, it is more reasonable and meaningful to consider the historical information when constructing the filter model (i.e., memory filter) [7], [8]. In other words, the working mechanism of the memory filtering algorithm is to update the filter’s states by means of the historical states of the filter on a moving window with finite-time horizon [9]. However, it is relatively rare to study the FTMFD issues for nonlinear networked systems, which constitutes a motivation for the current study.