Optimal Model Predictive And Disturbance Rejection Control With Applications In Parabolic Trough Solar Fields

It is challenging to control the outlet temperature of parabolic trough solar fields due to strong nonlinearities,large inertia,variable constraints and multiple field disturbances such as the fluctuations of direct normal irradiation and heat transfer fluid.To this end,this thesis intends to investigate the method on optimal model predictive and disturbance rejection control in view of the deficiencies in the existing research,which is of great significance to improve the efficiency and ensure the safety of parabolic trough concentrating solar power plants.The main contributions and novelties are summarized as follows:(1)An optimal model predictive based disturbance rejection controller is proposed in this thesis,where the multiple field disturbances are regarded as mismatched lumped disturbances and a generalized extended state observer is designed to estimate the lumped disturbances of each state channel.More importantly,a new equilibrium real-time calculation method based on steady-state target is proposed to transform the disturbance rejection and setpoint tracking problem into a new equilibrium tracking problem,which can compensate the lumped disturbances feedforward while maintaining the optimality of the solution.Bounded inputbounded output(BIBO)stability analysis of the closed-loop system has been presented.Simulation results indicate that the field outlet temperature can track given references with the proposed controller in the presence of various disturbances.(2)An optimal model based disturbance predictive and rejection control method with disturbance prediction is proposed,in which various field disturbances are first reconstructed into a single lumped one through model transformation and an extended state observer is designed to estimate the lumped disturbance.Based on the current and past estimated lumped disturbance values,an incremental least-square kernel regression disturbance predictor is constructed to predict the future dynamics of the lumped disturbance online.The fusion of the predicted disturbance future dynamics and optimal model predictive control is solved by realtime correction of steady-state target sequence,which can achieve the dual goals of the predicted disturbance dynamics feedforward compensation and solution optimality maintenance under the asymptotic stability of the closed-loop system.The research results show that the proposed method can accurately predict the lumped disturbance and has better disturbance rejection and setpoint tracking performance than that without considering disturbance prediction.(3)A constraint optimal model predictive and disturbance rejection controller is proposed,in which a dynamic optimization method of steady-state target value/ steady-state target sequence is proposed to consider the multiple properties of constraints satisfying,disturbance rejection and setpoint tracking in both cases without/ with disturbance prediction,and then a maximum controlled admissible set with disturbance/disturbance prediction compensation is constructed to convert infinite constraints into finite number constraints,an equivalent quadratic programming constrained optimization problem is solved by using dual-mode control strategy.In addition,based on the Nyquist criterion and the theory of stability domain,the influence of controller parameters on the stability margin of the closed-loop system is analyzed,and it is proved that the proposed controller has a two-degree-of-freedom control structure.The simulation results show that the proposed method can achieve setpoint tracking and disturbance rejection while satisfying the given constraints,and the comprehensive control performance of the controller is better when disturbance prediction is considered.