An LWPR-Based Data-Driven Fault Detection Approach for Nonlinear Process Monitoring

This paper presents a data-driven method for the task of fault detection in nonlinear systems. In the proposed approach, locally weighted projection regression (LWPR) is employed to serve as a powerful tool for modeling the nonlinear process with locally linear models. In each local model, partial least squares (PLS) regression is performed and PLS-based fault detection scheme is applied to monitor the regional model. The diagnosis for the global process is based on the normalized weighted mean of all the local models. Both conventional and quality-related statistical indicators are designed to compute the test statistics. Two nonlinear systems, a numerical one and a benchmark, are used to demonstrate the effectiveness of the proposed method.