Robust fault diagnosis with disturbance rejection performance for non-Gaussian stochastic distribution systems

In this paper, an enhanced robust fault diagnosis scheme is provided for the non-Gaussian stochastic distribution systems (SDSs). The available driven information for fault diagnosis is the probability density functions (PDFs) or the statistic information set of the output rather than the output value. A mixed neural network (NN) model with modeling error is established, where a static NN is applied to model the output PDFs and a dynamic NN is used to describe the relationships between the input and the weighting. The concerned problem is transformed into the fault diagnosis problem of the weighting system presented by an uncertain nonlinear system with unknown external disturbance. The statistic information driven composite observer for SDSs is constructed by combining a fault diagnosis observer with a disturbance observer, with which the fault can be diagnosed and the disturbance can be rejected simultaneously. Finally, simulations for the particle distribution control problem are given to show the efficiency of the proposed approach.