Robust distributed state estimation for Markov coupled neural networks under imperfect measurements

• Published in: Journal of the Franklin Institute Vol. 357; no. 4; pp. 2420 - 2436
• Main Authors: Hu, Xiaohui, Xia, Jianwei, Wang, Zhen, Song, Xiaona, Shen, Hao
• Format: Journal Article
• Language: English
• Published: Elmsford Elsevier Ltd 01.03.2020; Elsevier Science Ltd
• Subjects: Decoupling method; Delta function; Discrete element method; Estimating techniques; Markov analysis; Markov chains; Measurement; Neural networks; Robustness; State estimation
• ISSN: 0016-0032; 1879-2693; 0016-0032
• DOI: 10.1016/j.jfranklin.2020.01.021

In this paper, the issue of robust distributed state estimation is investigated for Markov coupled neural networks in the discrete-time domain. Fully considering network-induced phenomena, the signal quantization and sensor saturation existed in actual measurement are investigated in a unified framework through the Kronecker delta function. Moreover, a Markov chain is used to describe the structural variations in the addressed systems. The main attention of this paper is devoted to designing a mode-dependent estimator to estimate the system states through available output measurements, which ensures that the resulting system is stochastically stable and satisfies strictly dissipative property concurrently. By applying Lyapunov stability theory and a modified matrix decoupling method, some sufficient criteria are derived to obtain an explicit expression of the mode-dependent estimator. Finally, an example is presented to elucidate the validity of the proposed method.