A combined canonical variate analysis and Fisher discriminant analysis (CVA-FDA) approach for fault diagnosis

Benben Jiang, Xiaoxiang Zhu, Dexian Huang, Joel A. Paulson, Richard D. Braatz

Abstract: Highlights•A new fault diagnosis algorithm for serially correlated data is proposed.•Canonical variate analysis (CVA) is coupled to Fisher discriminant analysis (FDA).•The proposed approach is demonstrated on the Tennessee Eastman process.•Simulation results demonstrate improved fault diagnosis for serially correlated data.•The method outperforms dynamic FDA in discriminatory power and computational time.