Fault Prediction and Diagnosis of Electromechanical Equipment based on LSTM Neural Network

Mechanical and electrical equipment as the core of production, its failure will lead to production enterprises can not operate normally, resulting in economic losses of enterprises. In order to effectively predict and accurately diagnose faults of mechanical and electrical equipment, this paper constructs a fault diagnosis system, whose functional modules are mainly divided into front-end monitoring module, server module and back-end management module. LSTM module is included in the server module, which is the main technology of fault prediction and diagnosis of electromechanical equipment in this paper. The fault diagnosis model of electromechanical equipment is constructed by LSTM neural network. The gap between the predicted current spectrum and the actual current spectrum is compared by this model, and the gap is regarded as fault data to determine the fault type. The experiment shows that the accuracy, recall rate and $F 1$ value of the model in this paper are the highest, which are 97.85%, 97.77% and 97.81% respectively. The training time and prediction time of the predicted model in this paper are the shortest, which are 95.81ms and 0.02ms respectively. Therefore, the model in this paper can be used in the fault prediction and diagnosis of electromechanical equipment with the best effect. The research in this paper has practical significance for the normal operation of electromechanical equipment in production.