Summary
Earth pressure balance (EPB) shield tunneling machine has been widely used in underground construction. To avoid the catastrophic accidents caused by earth pressure imbalance, the earth pressure on excavation face must be controlled balance to that in chamber. To solve this problem better, a multi‐variable data‐driven optimal control method for shield machine based on dual‐heuristic programming (DHP) is proposed. The DHP controller is constructed with action network, model network, and critic network based on back‐propagation neural networks (BPNNs). Following Bellman's principle of optimality, a cost function of DHP controller for the chamber's earth pressure is presented, which simplifies a multi‐level optimization to a single‐level optimization. To minimize the cost function, the action network utilizes the critic network's error to achieve multi‐variable optimization, and the optimal control parameters for the tunneling process are obtained at last. The simulation results show that the method can effectively control the earth pressure balance. Even in case of disturbance, the system has strong anti‐interference ability and the control process is also quicker and steadier.