LSTM based hydraulic excavator angular velocity prediction model

Abstract

This paper proposes a long short-term memory (LSTM) model for predicting the angular velocity of an excavator. An excavator’s movement command appears at the speed of its hydraulic cylinder, which then appears as the angular velocity of its joint. Therefore, if the angular velocity of the joint, which changes as a function of the operating command, can be predicted, the excavator can be controlled. However, since the cylinder is a nonlinear system, it is difficult to create a system model. To solve this problem, we propose a model having long short-term memory (LSTM) based angular velocity prediction. We constructed an experimental environment for a hydraulic RC excavator, collected excavator data, and analyzed the prediction accuracy of our LSTM model. In addition, we applied the LSTM-based angular velocity prediction model to a PID control algorithm to compare the general PID control algorithm with the proposed control performance.