Speed reducer fault prediction method and system based on multi-sensor fusion

Abstract

The invention relates to the technical field of data processing, in particular to a speed reducer fault prediction method and system based on multi-sensor fusion, and the method comprises the steps: obtaining the basic features of a plurality of sensors, and constructing an enhanced feature vector for the sensors, the enhanced feature vector comprises a current basic feature of a corresponding sensor, a historical basic feature in a set window length, and current basic features of other sensors physically associated with the corresponding sensor; and training a normal behavior model based on the enhanced feature vector in the healthy state. According to the method, the enhanced feature vector containing the time sequence information and the cross-sensor associated information is constructed for each sensor, and the exclusive TCE-iForest model is used for evaluation, so that the reliability of the sensor is accurately evaluated, various sensor abnormal modes including slow drifting and transient peaks can be accurately identified, and the reliability of the sensor is improved. The defect that a traditional method neglects time sequence and relevance is overcome.

Images