Intelligent Early Warning and Monitoring Decision Model for Abnormal Vibration States of Motors and an Empirical Study

In chemical reactors, motor failures like belt breakage can disrupt production. Traditional methods for detecting motor abnormalities, such as manual inspections and monitoring control room currents, lack effectiveness. This study introduces a T2-LSTM algorithm combined with motor vibration sensors to predict motor quality and establish an early warning system. Statistical Process Control (SPC) techniques determine warning thresholds, while the ISO 10816-1 standard categorizes motor states as Good, Satisfactory, Unsatisfactory, or Unacceptable. Tested with 150 data points in a real-world scenario, the system achieved a 98% prediction accuracy (147/150), with 146 predictions within the normal range and only one unsatisfactory point among the four that exceeded the threshold.