A hybrid-encoding adaptive evolutionary strategy algorithm for windage alteration fault diagnosis

• Published in: Process safety and environmental protection Vol. 136; pp. 242 - 252
• Main Authors: Huang, De, Liu, Jian, Deng, Lijun
• Format: Journal Article
• Language: English
• Published: Rugby Elsevier B.V 01.04.2020; Elsevier Science Ltd
• Subjects: Accuracy; Adaptive algorithms; Air flow; Algorithms; Computer applications; Diagnostic systems; Euclidean geometry; Evolution & development; Evolution strategy; Evolutionary algorithms; Evolutionary computation; Fault detection; Fault diagnosis; Genetic algorithms; Hybrid-Encoding; Mine ventilation; Objective function; Particle swarm optimization; Robustness (mathematics); Ventilation; Windage alteration faults; Fault diagnosis; Hybrid-Encoding; Evolution strategy; Mine ventilation; Windage alteration faults
• ISSN: 0957-5820; 1744-3598
• DOI: 10.1016/j.psep.2020.01.037

It is critically important that windage alteration faults (WAFs) within mine ventilation systems be quickly identified and mitigated in order to ensure a safe mine production environment. Thus, we propose a Hybrid-Encoding adaptive Evolution Strategy (ES) Algorithm to diagnose the fault’s location and volume quickly and accurately, as it combines classification and regression features. The Euclidean distance between the airflow set calculated via fault diagnosis and the airflow set obtained by the monitoring system was used as the objective function value. Six benchmark functions and one thousand six hundred tests were carried out to verify the feasibility of using Hybrid-Encoding for WAFs diagnosis. The effectiveness of adaptive ES was demonstrated by Genetic Algorithm (GA), Differential Evolution Algorithm (DEA), and Particle Swarm Optimization (PSO). The experimental results fully validate the Hybrid-Encoding adaptive ES superiority in terms of accuracy, precision, diagnostic errors, robustness, computational efficiency, and convergence speed, etc. Diagnostic accuracy and precision during field testing were both 92.5 %, and 93.75 % of the results showed relative errors of < 5 %. Thus, our proposed Hybrid-Encoding adaptive ES Algorithm meets the requirements for fault diagnosis accuracy at the mine production site.