Data-driven Adaptive Control Method And Application

In this thesis,based on the data-driven adaptive control method,a series of improved algorithm design strategies are proposed for the strong non-linearity,strong coupling and strong external disturbance encountered in the actual control process of the rotating system of the hydraulic anchored drill.The feasibility of the proposed method is studied and analyzed,and the strict mathematical derivation and simulation verification in the algorithm design link are given.The main innovations and structure of this thesis are summarized as follows:1.For the actual drilling rig hydraulic servo rotary system,the non-linear models of its key components are constructed in detail.Firstly,several modeling methods of hydraulic system in practical application are briefly introduced,and the drawbacks of existing linear modeling methods are pointed out.Secondly,by analyzing the mechanical structure and oil flow characteristics of each component of the valve-controlled hydraulic motor system,different from the general approximate linearization modeling method,the non-linear mathematical models such as the flow continuity equation,the inertial load moment balance equation and the continuous flow equation of the servo valve for the hydraulic motor system are constructed respectively in the practical modeling.The friction factors in the key components,the unmodeled dynamic,modeling uncertainty and the non-linear characteristics are fully considered.Furthermore,a new non-linear state space mathematical model of the valve-controlled hydraulic motor rotating system is constructed by combining the non-linear equations as mentioned above.2.An intelligent self-learning PID control method based on partial format dynamic linearization（PFDL）is proposed to solve the problem that the drilling speed of discrete non-linear hydraulic rotary system cannot be accurately controlled.First,PFDL method is used to linearize its discrete model into affine data model with non-linear uncertainties,in which only I/O measurements of the original controlled system are used.Secondly,in the data-driven framework,an intelligent self-learning PID control method based on PFDL is proposed by introducing additional error feedback information.In order to solve the problems of time-varying parameters and non-linear uncertainties in this method,pseudogradient parameter estimation algorithm and time-difference based adaptive control algorithm are designed respectively.Finally,theoretical analysis and simulation study fully prove the effectiveness of the proposed control method.3.A data-driven adaptive control method based on extended state observer is proposed to deal with the inherent non-linearity and structural uncertainties in valve-controlled electro-hydraulic servo rotary system.Firstly,all nonlinearities and uncertainties in the electro-hydraulic system are compressed into a nonlinearity term,and an extended state observer is designed to estimate this nonlinearity term on-line and compensate for the complex dynamics of the system.Then,a time-varying parameter updating algorithm and data-driven adaptive control law are proposed based on the optimal theory,furthermore,the convergence of tracking error is proved by mathematical analysis.Finally,the simulation results verify the speed tracking performance of the system under actual operating conditions.4.A data-driven adaptive sliding mode control method based on Radial Basis Function neural network disturbance observer is proposed for valve-controlled electro-hydraulic servo rotary system with strong external disturbance,strong coupling and strong nonlinearity.First,the PFDL method is used to linearize the controlled system equivalent to an incremental form that is only relevant to the input and output,and the unknown generalized disturbance is combined into an additional non-linear disturbance term.Secondly,a radial basis network disturbance observer is designed to estimate the non-linear disturbance on-line,and a control algorithm is designed and pseudo gradient parameters are estimated using the I/O data of the system.Thus,the control law combining data-driven adaptive control with sliding mode control and the parameter update law are given respectively.Finally,the simulation results shows that the designed control scheme can well compensate the unknown load of the system and its external disturbance.