Resilient fault detection for industrial process using adaptive Fisher discriminant analysis with wavelet denoising

In this paper, fault detection in industrial process using adaptive fisher discriminant analysis (AFDA) with optimization of wavelet denoising (WD) parameters is discussed. Particle swarm optimization (PSO) is used to set the wavelet denoising parameters. Among several important factors in effective fault detection, false alarm rate (FAR) and missed detection rate (MDR) are prominent. The efficacy of the fault detection system has been demonstrated by applying it to a rotating machinery test rig developed at the Pakistan Institute of Engineering and Applied Sciences (PIEAS). Results have shown that the proposed technique can reduce the false alarm rate and missed detection rate to 1% as compared to the adaptive fisher discriminant analysis and fisher discriminant analysis. • Proposed an optimized AFDA-WD technique for robust industrial fault detection. • Utilized a PSO to tune wavelet denoising parameters for improved sensitivity. • Achieves fault detection with < 1% false alarm and missed detection rates. • Developed a rotating machinery test rig to test and validate the proposed framework. • Proposed is Outperformed conventional FDA and AFDA in accuracy and noise robustness.