Parameter Estimation And Servo Control Of Machine Tool Feed Drives

To improve the tracking performance and stability of control systems under different op-erating conditions,this thesis proposes the methods of online parameter estimation and servo control for machine tool feed drives with varying parameters and modeling errors.With the development of high-speed manufacturing,the closed-loop bandwidth of feed drive systems needs to be as high as possible to achieve high precision tracking at high speeds.However,if the closed-loop bandwidth is close to or higher than the natural frequency,the resonant mode can be easily excited,and the resulting structural vibrations can significantly reduce the tracking performance,and thereby reducing the manufacturing quality.On the other hand,the varying operating conditions of feed drive systems could lead to parameter variations.For instance,the varying table position determines the dynamic stiffness and the equivalent mass is varying in cutting and additive manufacturing.The parameter variations can deteriorate the performance of the controller designed for the nominal model,and even lead to an unstable closed-loop system.Besides,the influences of the modeling errors can not be ignored during high-speed manufac-turing.Therefore,it is necessary to consider the parameter variations and modeling errors,and propose control strategies for machine tool feed drives,to suppress structural vibrations and reject external disturbances while increasing the closed-loop bandwidth.For the parameter variations in feed drive systems,this dissertation proposes the control methods using the robust control and gain-scheduling control to improve the tracking perfor-mance and closed-loop stability.Since the table position of ball screw drives determines the dy-namic stiffness,the natural frequency of resonant modes is varying within the operating range,thereby deteriorating the performance and stability of the control system.To take into account the influence of table motions,an uncertain model is built for ball screw drives and it regards the varying frequency responses of the axial resonant mode as the frequency-domain uncertainty,which is represented by a weighting function.Based on the uncertain model,the?-synthesis technique is used to design the robust controller,which minimizes the structural singular value of the closed-loop system and achieves the robust stability and robust performance in the entire operating range.The reference model is further introduced into the standard control structure to tune the closed-loop performance in the time domain.A gain-scheduling robust controller design method is proposed in order to consider the table motion and workpiece mass simultaneously.The workpiece mass is regarded as the pa-rameter uncertainty for designing the?-synthesis robust sub-controllers that achieve the robust performance and robust stability against the uncertain workpiece mass.Since the table posi-tion can be measured by the encoder,it can be used as the scheduling variable to design the gain-scheduling robust controller,which further reduces the control conservatism caused by the uncertainties.The gain-scheduling robust controller is designed by linearly interpolating the transfer function coefficients of a set of?-synthesis sub-controllers,which are designed at different table positions.The stability is posteriorly verified by using the parameter-dependent Lyapunov function,which theoretically guarantees the closed-loop stability of the system under the interpolating gain-scheduling controller.In machine tool feed drives,it is usually difficult to measure the load mass in real-time.If the variation range of the load mass is large,to guarantee the closed-loop stability,the robust controller would be very conservative.Therefore,to explicitly deal with the variations in un-measurable parameters,an online parameter estimation method is proposed and the estimated parameter is used to schedule the controller gain.The perturbations caused by the parameter variations are extracted from the state-space model and regarded as the new states in an extended state-space model.By using the state observer for estimating the perturbations,the parameter variations can be obtained online.Accordingly,with the extended state observer,the state feed-back controller with double integrators is proposed,which eliminates the steady-state error for ramp inputs and improves the disturbance rejection.Since the conventional model-based control methods cannot avoid the modeling errors,the data-driven controller is developed for machine tool feed drives,which directly uses the fre-quency response data to design the optimal H_?controller.The two-degree-of-freedom control structure is adopted for the controller design,which consists of a feedback controller,a feed-forward controller,and a fixed controller.Through the linearization at the initial controller,the convex-concave optimization problem using frequency response data is reduced to the convex optimization problem and can be solved with matrix inequalities.The solutions to the opti-mization problem with different controller structures are discussed,and the proposed control algorithm is extended to the scenarios using the reference model and multiple data uncertainty.The aforementioned estimation and control methods are compared with conventional clas-sical control methods and existing advanced control methods,and both the analysis and exper-imental results demonstrate the advantages and effectiveness of the proposed methods.