Coordinated Tuning of Slurry Shield Control Parameters based on Reinforcement Learning

Excavation face stability and controllable shield attitude are two core issues in slurry shield tunneling. Given the complex interactions between various control systems and their environment, achieving both face stability and controllable shield attitude simultaneously presents a significant challenge. Current methods, which rely entirely on human experience, pose potential risks to construction safety and quality. For this reason, we put forward the idea of coordinating the tuning shield control parameters based on reinforcement learning. After an in-depth analysis of the control system, we design a coordinated control algorithm based on Soft Actor-Critic reinforcement learning and establish a comprehensive evaluation index system for shield construction, which realizes the coordinated control of the tunneling working face pressure and the tunneling speed. Experimental results demonstrate that this model can offer practical and effective guidance for slurry shield tunneling construction, fulfilling the objective of coordinated control over settlement and controllable attitude parameters.