Research on Equipment Fault Prediction Method Based on Industrial Big Data

Abstract

— Aiming at the problems of low accuracy and long prediction time of industrial equipment fault prediction, a prediction method based on main feature extraction and back propagation neural network (MFE-BPNN) was proposed. This method firstly preprocesses the missing, abnormal and high-noised industrial equipment data, then uses the method of recursive feature elimination combined with cross validation to extract the main feature variables, then designs the numbers of hidden layers and neurons, and weights of training and learning rates. This method improves the accuracy of industrial equipment fault prediction by preprocessing industrial data and establishing a prediction model based on a neural network. The prediction time is reduced by extracting the main characteristic variables. The experimental results of fan blade icing fault prediction in the field of power generation verify the effectiveness of this method.