Application of physics-informed machine learning for excavator working resistance modeling

Abstract

Accurate measurement of the working resistance encountered during excavation plays a vital role in improving production efficiency, reducing energy consumption, and enabling intelligent capabilities in excavators. To address the challenges of the inaccurate measurement of working resistance, we present a novel method for measuring excavator working resistance based on physics-informed machine learning (PIML). Firstly, we construct a mechanism model, which incorporates the kinematic and dynamic models of the excavator working device. Next, by integrating the mechanism models with machine learning algorithms, a soft sensing model for excavator working resistance based on the PIML method is developed. Finally, experimental analysis is conducted to compare the predictive performance of three different methods: mechanism modeling, data-driven modeling, and PIML. The results indicate that the PIML method allows for a more precise estimation of the working resistance experienced during the excavation process of excavators, and it holds significant theoretical and practical implications.