【作者】 刘建；

【导师】 龚国芳；

【作者基本信息】 浙江大学 ， 机械工程， 2018， 硕士

【摘要】 泥水盾构是集开挖、注浆、出渣、衬砌、测量于一体,适用于软弱砂质土层,可实现长距离复杂地质施工的一次成型的大型隧道开挖装备。它主要由推进系统、刀盘系统、泥水环流和分离处理系统等组成。其中推进系统是保障泥水盾构沿预定轴线掘进,减小隧道开挖误差的关键子系统。目前国内外对盾构机的研究主要集中在土压盾构中,对于泥水盾构的研究较少,对于姿态控制技术研究集中在轨迹规划及建模上,对外部负载考虑较少,难以应用于实际施工中。且国内工程中姿态调整依靠人工调整,相对比发达国家已应用模糊控制理论逐步实现自动化调姿的现状,研究智能化、高精度的泥水盾构姿态调整技术显得尤为突出。本课题是在省重大科技专项的资助下,设计了直径2.5m的泥水盾构试验台,利用历史施工数据进行鲁棒学习,设计了姿态调整智能决策系统,提出相应的自适应控制策略,采用仿真试验验证了系统的有效性。本论文的主要研究工作如下:1.分析了实际施工中盾体的负载力、驱动力,建立了泥水盾构掘进过程的拉格朗日动力学模型并,并分析了推进液压系统工作原理,建立了电比例三通减压阀控单作用液压缸的数学模型;2.通过分析基于人工操作的泥水盾构姿态调整技术,描述了姿态调整的越纠越偏、纠偏过度、纠偏过慢等不同情况下的数据分布,提出了优秀姿态调整数据的清洗原则,并通过利用核K均值聚类法对样本数据进行聚类及清洗;3.介绍了带有损失的最小二乘学习法,并结合一次范数和二次范数损失,设计了基于Huber损失最小化的鲁棒学习法,选用高斯核函数和8折交叉验证法训练样本集,得到了水平和垂直姿态调整学习模型的确定系数分别为约76.5%和71.5%左右,选择确定系数较大的模型学习参数建立了姿态智能调整模型;4.通过构建Diophantine方程,确定了推进系统模型参考自适应控制器的参数调节律和控制律;分别对无负载干扰和有负载干扰两种情况进行仿真分析,结果表明自适应控制器对负载干扰有一定的鲁棒性;5.通过综合鲁棒学习模型、自适应控制器、推进系统模型及动力学模型,建立了完整的姿态调整智能决策系统,对其进行了仿真试验,结果表明智能决策系统对水平方向的姿态调整有良好效果,对垂直方向姿态调整有少许滞后,这是由盾构大惯量特性导致的;6.针对泥水盾构在实际施工中的运动特性,设计了 5自由度的姿态模拟试验平台,通过带有虎克铰的推进系统和支撑随动机构,建立了姿态解算方程;并设计了随动支撑液压系统,采用模糊PID控制器对支撑缸压力进行控制,仿真结果表明支撑缸具有良好的随动性,能动态补偿盾体自重,实现5自由度的姿态模拟。更多还原

【Abstract】 Slurry pressure balance shield(SPB)is a large-scale tunnel excavation equipment that integrates excavation,grouting,slag discharge,lining and measurement.SPB is suitable for soft sandy soil layer so that it can realize long-distance complicated geological construction.SPB is mainly composed of cutter system,mud circulation and separation processing system,propulsion system,in which the latter is the key subsystem that guarantees SPB tunnels along the predetermined axis and reduces error.At present,the research on the shield machine is mainly focused on the earth press balance shield(EPB).Also,the attitude control is focused on trajectory planning and modeling,which is difficult to apply in actual tunneling because of ignoring external load.Moreover,the attitude adjustment in domestic engineering relies on manual adjustment.Compared with the status quo that developed countries have employed fuzzy control theory to gradually realize automatic attitude adjustment,intelligent and high-precision attitude adjustment technology of SPB is particularly prominent.A test platform of SPB with a diameter of 2.5m is designed with the fund of a major science and technology project of Henan province.The historical tunneling data is used for regression learning.An intelligent decision system of attitude adjustment is designed.A model referenced adaptive control strategy is proposed.The result of simulation test verifies the effectiveness of this system.The main work of this thesis is as follows:1.The load and propulsion of SPB in the actual tunneling was analyzed,and the Lagrangian dynamic model of SPB in tunneling process is set up;the principle of the propulsion hydraulic system is analyzed,and mathematical model of electro-hydraulic proportional 3-way pressure reducing valve that controls single-acting hydraulic cylinder is established;2.The distribution of data in different attitude adjustment situations is described by analyzing the manual operation-based attitude adjustment technology.A principle of data cleaning to select excellent attitude adjustment is proposed.The kernel K means clustering method is used to Cluster and clean sample data;3.The least-squares learning method with loss is introduced and by combining with the loss of first-order and second-order norm,a robust learning method based on Huber loss minimization is designed.By using Gaussian kernel function and 8-fold cross-validation,sample data the horizontal and vertical attitude adjustment is training,and the R-squared score of learning models are about 76.5%and 71.5%,respectively.An attitude adjustment model is established by selecting learning parameters with largest R-squared score;4.Through establishment of Diophantine equation,parameter tuning law and control law of model referenced adaptive controller of propulsion system is determined.Controlling processes with load disturbance and without load disturbance are respectively simulated and analyzed.The results showed that the adaptive controller is sensitive to load disturbance and has a certain degree of robustness;5.Through integrating robust learning model,adaptive controller,propulsion system model and dynamic model,a complete attitude adjustment intelligent decision system is established and simulated.The results show that the intelligent decision system have a good performance on attitude adjustment of horizontal direction,and there is a slight lag in attitude adjustments in the vertical for large weight and inertial of SPB;6.Aiming at the motion characteristics of SPB in real tunneling,a 5-DOF attitude simulation test platform is designed.The attitude solution equation is established through the propulsion system with Hooke hinge and the follow-up supporting mechanism.The Supporting hydraulic system is designed and simulated via a fuzzy PID controller.The simulation results show that the support cylinder has a good following function and can dynamically compensate the weight of shield’s body to achieve attitude simulation with 5-DOF.更多还原