A brief assessment of outliers and malfunctions detecting techniques with an application on lubricant condition monitoring

Abstract

Condition monitoring and machine status classification are of great practical importance in the manufacturing industry as it provides online updates on the state of the machine, avoiding loss of production and minimizing the probability of generating catastrophic damage to the machine. In this paper, the classification of conditions is based on the processing of information using wavelets based on the results of the monitoring and the data collected during such an action, measuring the characteristics of the lubricating oil over some time sufficient to produce a time series of results. In this paper, the classification system is tested and validated using observation sequences based on the maximum wavelet distribution obtained from the collected signals, monitoring the state of the lubricating oil, to define and diagnose singularities in time series.