Evaluation System of Curved Conveyor Belt Deviation State Based on the ARIMA–LSTM Combined Prediction Model

Abstract

The deviation of conveyor belts is a key factor that restricts material conveying efficiency of horizontal curves. An evaluation system of the deviation state of curve conveyor belt based on the ARIMA–LSTM combined prediction model is proposed in this study to investigate the deviation state of the conveyor belt. This system has been used for detection, prediction, deviation correction, and early warning of the conveyor belt deviation state. First, the experiment system of the conveyor belt deviation was built. The conveyor belt deviation images were collected using the machine vision method, and the conveyor belt deviation state data set was established. Second, the mechanical model of curve conveyor belt deviation is presented and solved. The correctable deviation range of the idler frame under different elevations and trough angles was obtained by solving the problem. Third, the ARIMA–LSTM combined prediction model of conveyor belt deviation based on series-parallel weighing method was put forward. The analysis results showed that the ARIMA–LSTM combined prediction model is suitable for the prediction of conveyor belt deviation in terms of accuracy, fitting degree, time, and performance. Finally, the deviation state evaluation system was established to realize the visual fusion of the ARIMA–LSTM combined model in the range of correctable deviation of the idler frame. The OCSVM algorithm was used to detect the abnormal deviation of the conveyor belt. The experiment shows that the evaluation system can predict and send out early warning signals according to the detection results, provide corresponding suggestions for adjusting the deviation correction angle, and realize an efficient and intelligent solution for the evaluation of the curved section belt deviation state.