盾构土压平衡动态神经网络逆控制技术研究

盾构密封舱的土压平衡作为地表沉降控制的关键因素对盾构安全施工具有重要保障。为映射影响土压平衡掘进参数之间的非线性耦合关系，增强非线性控制模型的动态性能，提高土压平衡的控制精度，根据盾构施工中影响密封舱内土压平衡的掘进参数调控的难易程度，依托现场监测数据建立基于动态神经网络逆控制前馈作用下的螺旋输送机转速控制模型。对控制模型的性能与效果进行分析验证，结果表明： 动态神经网络输出的前馈螺旋输送机转速能够对推进速度、刀盘转矩的变化响应灵敏； 在给定掘进条件下与通过人工调节螺旋输送机转速控制土舱压力的方法相比，动态神经网络逆控制前馈作用下密封舱土压的最大波动误差由9.8%降为5.3%。

Inverse Control Technology of Dynamic Neural Network of EPB Shield

As a key factor for ground settlement control, the earth pressure balance of seal chamber is an important guarantee for safe shield construction. In order to map the nonlinear coupling relationship between the excavation parameters affecting the earth pressure balance, enhance the dynamic performance of the nonlinear control model and improve the control precision of the earth pressure balance in seal chamber, a screw conveyor speed control model based on feedforward structure of inverse dynamic neural network is established according to the monitoring data and the control difficulty of the excavation parameters affecting the earth pressure balance. The performance and effect of the control model are analyzed and verified. The results show that: (1) the feedforward speed by the dynamic neural network can respond quickly to the variations of propulsion speed and cutter torque; (2) compared with the method of manually controlling the speed of the screw conveyor under the condition of excavation, the maximum fluctuation error of the earth pressure of the seal chamber under the feedforward control of the dynamic neural network is reduced from 9.8% to 5.3%