A fault prediction method for small generators based on improved lstm-mlp

Abstract

The invention relates to an improved LSTM‑MLP-based fault prediction method for small generators, which belongs to the technical field of artificial intelligence and machine learning. The method includes two models based on the improved LSTM-MLP and multi-layer perceptron, and predicts the input by inputting the data collected by the DTU module of the small generating set. The improved LSTM network in the network is responsible for the small generating set as the front network. For state prediction, the multi-layer perceptron is responsible for the fault classification of small generator sets and outputs the results of the entire system as a post-network. The present invention improves the LSTM network structure and introduces the covariance of the previous timing state and the current timing state into the gating structure of the network as a weight coefficient, further improves the accuracy of LSTM network prediction, and increases the reliable use of small generators time, and reduce the labor cost required for the operation and maintenance of the generating set.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence and machine learning, and relates to a small generator fault prediction method based on an improved LSTM-MLP. Background technique [0002] Fault prediction is a very critical part of industrial equipment. In recent years, with the innovation of manufacturing models and the emergence of emerging technologies, "smart manufacturing diagnostic evaluation" has become one of the indispensable systems in the industry. Most of the traditional fault diagnosis and prediction technologies rely on expert systems or manual evaluation. Most of the small generator equipment operation and maintenance enterprises cannot pay high expert consultation fees. Equipment fault diagnosis and predictive maintenance, as a major part of the entire intelligent diagnosis system, are a hurdle that must be crossed in the transformation and upgrading of the manufacturing industry towards intelligence and digitali...

AI Technical Summary

Problems solved by technology

However, in the traditional LSTM network, as the input dimension of the network increases, the coupling between the dimensions becomes lower and lower, resulting in a decrease in the accuracy of the prediction.

Images