An improved quantitative recurrence analysis using artificial intelligence based image processing applied to sensor measurements

摘要

Artificial intelligence hasbeenwidely used in reliability analysis for industrial equipment.Thegeartransmission systems are the most common components inminingmachines.Asimple fault in thegearboxmay break down the mining machine for couple of days, resulting in enormous economicloss. Condition monitoring techniques can prevent unscheduled failures in the gear transmissionsystems. Although many techniques have been developed for gearbox fault diagnosis, onechallenging task that the conditionmonitoring still faces ishowto extractquantitative fault indicators.To this end, this paper proposes an improved quantitative recurrence analysis (IQRA) basedon artificial intelligence theory. This new method takes advantages of chaos and fractal propertiesof the gear transmission system to obtain the recurrence of the system. The characteristicsof different gear faults can be observed through the visualization of recurrence. Quantitativeparameters can be then calculated from the recurrence plots. Experimental data acquired from agearbox under variable working conditions was used to evaluate the proposed method. The analysisresults demonstrate that the proposed IQRA method is able to effectively quantify differentthe gear faults.