Fault detection of nonlinear stochastic systems via a dynamic event-triggered strategy

• Published in: Signal processing Vol. 167; p. 107283
• Main Authors: Ning, Zhaoke, Wang, Tong, Song, Xiaona, Yu, Jinyong
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2020
• Subjects: Dynamic event-triggered mechanism; Fault detection; Nonlinear stochastic systems; Packet dropout; Randomly occurring nonlinearity; Fault detection; Nonlinear stochastic systems; Randomly occurring nonlinearity; Dynamic event-triggered mechanism; Packet dropout
• ISSN: 0165-1684; 1872-7557
• DOI: 10.1016/j.sigpro.2019.107283

•An unified fault detection design strategy is firstly investigated for nonlinear stochastic systems subject to random nonlinearity, data transmission delays and packet dropout via a dynamic event-triggered mechanism, which provides an adaptive function to change the threshold in real time.•Two independent Bernoulli stochastic sequences are introduced to describe the model nonlinearity and data packet dropout in probability.•The Wirtinger-based integral inequality is provided to reduce the conservativeness of obtained results. This paper studies the fault detection problem of nonlinear stochastic systems with randomly occurring nonlinearity and missing measurements via a dynamic event-triggered strategy. Firstly, the dynamic event-based data transmission strategy is provided to increase network utilization efficiency, while the data transmission threshold is changed with an adaptive function. This design idea can reduce network transmission pressure more effectively than the traditional one with fixed threshold. Secondly, two independent Bernoulli stochastic sequences are provided to describe the randomly occurring nonlinearity and measured-data packet dropout phenomenon, respectively. By considering time-varying delays, dynamic event-triggered strategy, randomly occurring nonlinearity and packet dropout simultaneously, a unified dynamic model for fault detection is established to generate mean-square asymptotical stability and desirable detection performance. Three examples are finally provided to generate the practicability of proposed technique.