Advanced Statistical and Meta-Heuristic Based Optimization Fault Diagnosis Techniques in Complex Industrial Processes: A Comparative Analysis

Abstract

Industrial processes are nonlinear and complicated, requiring accurate fault identification to minimize performance deterioration and respond quickly to emergencies. This work investigates industrial process defect identification and isolation, which is analytically difficult owing to their complexity. This paper carefully analyzes four design methods for flaw identification and isolation based on Principal Component Analysis (PCA), Fisher Discriminant Analysis (FDA), Kernel Fisher Discriminant Analysis (KFDA), and Sequential quadratic programming (SQP). Our study includes the Tennessee Eastman Process (TEP) and the Penicillin Fermentation Process (PFP), among other comparable methods. We assess the proposed fault detection and isolation methods through detailed analysis and comparison. The simulation findings from our extensive investigation provide remarkable insights. Simulation findings show that FDA and KFDA work well in fault identification and isolation, but PCA has certain limits. We also considered SQP as a TEP fault detection and isolation improvement tool. SQP is noted for its success in nonlinear and restricted optimization problems, making it ideal for fault identification and isolation in complicated industrial processes. Data-driven design approaches increase problem identification in complicated industrial processes with greater reliability and efficiency than PCA-based methods. This study also shows that advanced data-driven techniques can improve industrial fault diagnosis, improving operational safety and system performance by leveraging the FDA, KFDA, and SQP.

An adaptive metaheuristic optimization approach for Tennessee Eastman process for an industrial fault tolerant control system

“…This adaptive modification ensures that the chosen attributes are in line with the changing traits of the TEP, hence improving the overall resilience of our approach. Comparative assessments comparing statistical approaches in [24,50,51,56] procedures with ACSA demonstrate the shortcomings of traditional methods in adjusting to dynamic industrial processes, hence emphasizing the suitability of ACSA in our specific situation.…”

“…As shown in Fig 7, the optimizer consistently improves the fitness values over iterations, indicating its effectiveness in achieving the desired objectives. Compared with other well-established optimizing approaches such as SQP in [56], the proposed ACSA quickly converge to optimal solution with finer iteration count. The ACSA adaptive nature and intelligent search mechanisms enable it to swiftly converge towards optimal solutions, resulting in a high fault detection rate while minimizing false alarms.…”

“…The convergence analysis of the optimal solution in the ACSA in comparison to SQP reveals notable distinctions in their convergence behaviors. ACSA, characterized by its bio-inspired optimization approach, demonstrates superior performance in terms of early convergence when contrasted with SQP [56]. This adaptability allows ACSA to efficiently explore the solution space, facilitating a swift convergence towards the optimal solution.…”

Enhancing Pneumonia Detection using Vision Transformer with Dynamic Mapping Re-Attention Mechanism

“…Advancements in technology drive the growing necessity to integrate cutting-edge methods AI and ML into fault diagnosis processes (Safari, Mehralian and Teshnehlab, 2020) and (Safari, Khalfalla and Imani, 2022). (Faizan-E-Mustafa et al, 2023) explored four distinct strategies to identify and isolate faults within the TEP system. Their methodologies encompass principal component analysis, kernel Fisher discriminant analysis, and sequential quadratic programming.…”