An integrated algorithm on disturbance rejection and fault diagnosis for a class of stochastic distribution systems

An integrated algorithm on system modeling, disturbance rejection and fault diagnosis problem is presented for a class of stochastic distribution control (SDC) systems with exogenous disturbance in this contribution. A 2-step neural network (NN) algorithm is employed to set up the plant. Unlike traditional SDC systems, the driven information is modeled in the first step as a kind of probability density function (PDF) of the output or its monitor information via a static neural network. An adaptive dynamic neural network is employed in the second step to identify the nonlinearity, uncertainty of the system, where the weight matrices of NN and their boundary are designed with adaptive rules. A full order disturbance observer is designed for disturbance rejection purpose while an adaptive 1st-order filter is designed for fault diagnosis purpose. Through such adaptive algorithm, nonlinear parameter estimation, disturbance and sensor fault identification can be well dealt with simultaneously. A sensor compensation rule is consequently given to restore the plant with output feedback controller. Simulation examples are given to verify the effectiveness of the presented algorithm.