Least Squares and Contribution Plot Based Approach for Quality-Related Process Monitoring

In this paper, a new contribution plots approach based on least squares is proposed to realize fault detection and diagnosis for quality-related sensor faults in industrial processes. The proposed approach can achieve the quality-related fault detection and diagnosis simultaneously using an indicator of contribution plots. The process variables are decomposed into two orthogonal subspaces, which are quality-related and quality-unrelated. Then, the variable contributions to statistics are calculated in each subspace, and quality-related fault detection and diagnostics are achieved by analyzing the contribution plots of all variables. Finally, the validity of the fault diagnosis method proposed in this paper is verified by simulation analysis of the Tennessee-Eastman process.