计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (4): 251-255.DOI: 10.3778/j.issn.1002-8331.1506-0282

• 工程与应用 • 上一篇    下一篇

IMNMSSPC算法在盾构机土压平衡控制中的应用

宋英莉,刘宣宇,张凯举,施惠元,王  倩   

  1. 辽宁石油化工大学 信息与控制工程学院,辽宁 抚顺 113001
  • 出版日期:2017-02-15 发布日期:2017-05-11

Algorithm of IMNMSSPC applied in Earth Pressure Balance(EPB) control of shield machine

SONG Yingli, LIU Xuanyu, ZHANG Kaiju, SHI Huiyuan, WANG Qian   

  1. School of Information and Control Engineering, Liaoning Shihua University, Fushun, Liaoning 113001, China
  • Online:2017-02-15 Published:2017-05-11

摘要: 针对盾构掘进过程中密封舱内土压平衡控制的精确度低、效果差的问题,提出了一种改进的多变量非最小相位状态空间模型预测控制(IMNMSSPC)应用于密封舱多点土压平衡控制。该方法利用阶跃响应数据建立传递函数模型,将模型转换成状态空间的形式,在目标函数中引入了可调因子,反馈校正中对预测误差补偿进行校正,有效减小了模型失配时产生的误差;建立了以密封舱多监测点土压力为输出,以盾构机推进速度、螺旋输送机转速为输入的盾构土压平衡过程模型,通过多步预测直接递推出过程预测输出,并对控制量进行优化求解,从而实现土压平衡的自动控制。应用结果表明该算法具有较高的控制精度、良好的跟踪性能和强鲁棒性。

关键词: 密封舱, 土压平衡, 多变量系统, 状态空间, 模型预测控制

Abstract: As control accuracy of Earth Pressure Balance(EPB) of pressure chamber lowly and its effect badly during tunneling process, an improved multivariable non-minimal state space model predictive control(IMNMSSPC) algorithm that is applied to the multi-point earth pressure balance control of head chamber is proposed. A transfer function matrix model is built through the step-response dates, and the construction of state space can be transformed into this model. This method introduces an adjustable factor to the cost function, and puts forward a new error compensation method in feedback correction, which can effectively reduce some errors produced through model mismatch. Predictive outputs are deduced with the theory of multi-step method. The earth pressure balance control model of shield machine takes the driving speed, the screw conveyor’s speed as inputs, and takes the pressure of head chamber as output. The algorithm can optimize the control variable. The application results show the proposed algorithm has more excellent control performance and a good tracking, disturbance rejection and robustness.

Key words:  head chamber, earth pressure balance, multivariable system, state space, model predictive control