Collaborative fault tolerant control for stochastic distribution.pdfVIP

Proceedings of the 2011 International Conference on Advanced Mechatronic Systems, Zhengzhou, China, August 11-13, 2011 Collaborative fault tolerant control for stochastic distribution systems Yuwei Ren, Aiping Wang, and Hong Wang, Senior Member, IEEE Ahstract-This paper is concerned with the collaborative fault tolerant controller design problem for m non-Gaussian collaborative stochastic distribution systems which connect in series. The main idea for this paper is that when fault occurs in one of the subsystems, a collaborative fault tolerant controller will be designed for one of the rest healthy subsystems based on the non-fault controller and fault diagnosis value. This controller can accommodate the fault and guarantee that the whole systems still have a good tracking performance. I. INTRODUCTION T o improve the reliability of control systems, fault diagnosis and fault tolerant control techniques have long been regarded as an important and integrated part in modern control systems [1-3]. For stochastic systems, in the past two decades many significant approaches have been presented and applied in practical processes successfully [4-6]. A common feature in these works is that all variables in the system are supposed to obey a Gaussian distribution and only the output mean and variance are controlled. Different from any other previous stochastic control approaches, the variables can be non-Gaussian in references [7-9] and the concerned output is, in fact, the whole shape of the output probability distribution function (PDF). The objective of the controller design is therefore, to find a crisp control input, namely the PDF control, so that the shape of output PDF can follow a given distribution. On the other hand, for complex stochastic processes, nonlinearity will generally lead to non-Gaussian outputs even if the stochastic inputs are Gaussian. If the output obeys an asymmetric PDF with multiple peaks, theoretically the classic