Weighted Data-Based Fault Detection Approach for Nonlinear Nuclear Power System

Abstract

Mechanism-based and data-based methods are widely used in fault detection of nuclear power plant. Each method has its advantages and disadvantages. In this paper, we propose weighted data-based fault detection method based on data which we called Implicit Model Approach (IMA) for nuclear power plant. By introducing weighted factor and Takagi-Sugeno modeling method, we extend the IMA-based fault detection from linear system to nonlinear system. At local operating point, linear IMA is utilized to generate sub-system residual function. Inspired by Parallel Distributed Control (PDC) method, we generate overall residual function and overall residual evaluation mechanism. Simulation analysis has verified the validity of the proposed algorithm. The significance of the weighted IMA lies in establishing a data-based and mechanism-friendly approach. This work can provide new route for fault detection of nonlinear nuclear power system.