Research on Sensor Interference Recognition of Three-Phase Asynchronous Motor Based on Deep Learning

Three-phase asynchronous motor is widely used in petrochemical industry, but they are generally installed in harsh environments. Traditional maintenance and fault diagnosis are difficult to detect the potential faults of three-phase asynchronous motors effectively. It is a common method to detect the state of three-phase asynchronous motor based on temperature and humidity. Considering the characteristics of temperature and humidity data, this paper introduces a CM-RSL model based on deep learning. The model comprises the Random Forest (RF) algorithm for detecting data anomalies resulting from electromagnetic interference, a Support Vector Machine (SVC) classifier for identifying anomalies caused by sensor failure, and Long Short-Term Memory (LSTM). The CM-RSL model is used to analyze whether there is abnormal interference in the temperature and humidity data of the three-phase asynchronous motor, and then estimate the working condition of the connected motor. In the concluding phase, the proposed model underwent evaluation using long-term real-time temperature and humidity data from 32 three-phase asynchronous motors, collected through temperature and humidity sensors. The results demonstrate that the CM-RSL model effectively identifies the sources of disturbances leading to anomalies in the temperature and humidity data of three-phase asynchronous motors, achieving an accuracy rate exceeding 98%.