A data Processing Method for Fault Prediction of Industrial Pipeline Time Series Data

This paper solves the problem of data processing in fault prediction in the manufacturing process of industrial pipeline based on time series data. A data processing method is provided, which can segment and reorganize the pipeline data, and clean and preprocess the reconstructed data. At the same time, data training and evaluation prediction models are proposed to evaluate the processed data. The implementation of the data processing method is modular and scalable, supporting the underlying production process and collecting data changes. Based on the recurrent neural network, a prediction model was proposed and the data collected by the thin film transistor liquid crystal displayer production line was used to evaluate the method.