Preprocessing Method for Industrial Data Based on LSTM Model Prediction

In recent years, in the field of deep learning applications, LSTM have great advantages in the prediction of time-series data. But in the multi-dimensional industrial test data experiment, the effect is not very ideal. In this paper, the deredundancy compression method, the pooling processing and the noise removal method are proposed to perform the data preprocessing of the cube. By comparing the correlation coefficient and the correlation threshold value, the dataset with high correlation coefficient is selected for pooling and denoise. The result shows that the prediction accuracy is high after the data preprocessing in this paper. Meanwhile, the effectiveness of the preprocessing method is fully verified.