Objective Cantilever roadheaders represent roadway tunnelling equipment that integrates the deep coupling of mechanical, electrical, and hydraulic systems, and their control accuracy directly determines the forming quality of roadways.

Methods Cantilever roadheaders involve nonlinear mechanical motion, hydraulic control, and cutting load, rendering it challenging to establish accurate control models. Using the philosophy of digital twin modeling, this study developed a method for optimizing the digital twin model of the mechanical-electro-hydraulic system of cantilever roadheaders based on virtual-physical consistency evaluation. Specifically, a digital twin model of the mechanical-electro-hydraulic system was developed using multi-domain co-simulation between AMESim and Simulink. This model allows the coupling properties and nonlinear dynamic behavior of the mechanical-electro-hydraulic system to be accurately expressed in the digital space. By establishing an evaluation system for the consistency between virtual and physical control behaviors in roadway cutting, the evolutionary mechanisms of the deviations between the digital twin model and the physical control system were explored. Based on the evaluation results, the iterative calibration of key physical parameters in the digital twin model was conducted using the particle swarm optimization algorithm. By minimizing the deviations between virtual and physical control behaviors, the digital twin model’s ability to reproduce the physical cutting process of roadways was enhanced.

Results and Conclusions  The proposed optimization method for the digital twin model was verified using an experimental platform. The verification results indicate that after model optimization with cylinder displacement as a benchmark, the deviation between virtual and physical control behaviors was reduced by 84.5%, with the root mean square error (RMSE) and the mean absolute error (MAE) both decreased by up to 70% and the maximum absolute error decreased by 60%. After model optimization with cylinder pressures under normal and abnormal load operating conditions as benchmarks, the deviations decreased by 83.57% and 81.98%, respectively. These findings suggest that the model optimization can effectively reduce the deviation between virtual and physical behaviors under complex operating conditions. Overall, the optimization method proposed in this study provides an efficient and feasible solution for digital twin-based precise modeling and virtual-physical interactions of mining and tunneling equipment under complex underground operating conditions.