Study On Model Predictive Control Of The Regulating System Of Hydropower Unit

Hydropower unit is a widely-used power production equipment which plays a key role in energy supply,peak load and frequency regulation and emergency standby to maintain system stability in modern power grids.It is a complex nonlinear system with hydraulic-mechanical-electrical coupling characteristics.Its operational safety,stability and efficiency attach great significance in power quality improvement and system stability maintenance.In recent years,the hydropower unit is showing a trend of large capacity and structural complexity with extensive utilization of large-scale hydropower units and pumped-storage units.For the sake of improving control performances and enhancing the stability of the unit under complicated operating conditions,the research on advanced control theory and method for the regulating system of hydropower unit is urgently demanded.Traditional control schemes of hydropower unit such as PID have simple structures and are easy to be understood.But they also have a single control objective,take little inner-state information into consideration and lack the capability of handling the intricate nonlinear constraints.For this reason,the control performances of these control schemes are often deteriorated in the increasingly complicated operating environment.Therefore,modern control theory and intelligent control theory have been widely utilized in design of the regulating system of hydropower unit and become hotspots of regulating system control of hydropower unit.Model predictive control(MPC)is a computer control method which is widely used in industrial control systems.Its algorithm structure is not constrained to system model types.The strong robustness and explicit treatment of state constraints make it applicable to the regulating system of hydropower unit with complex model structure and nonlinear elements such as saturation and dead band.Taking the demand for high quality control of frequency,voltage and power in the operation of hydropower unit as the entry point and concentrating on the key scientific problems in MPC of hydropower unit regulating system,this thesis studies MPC theories and methods for the governing system of pumped storage unit,the excitation system of synchronous generator and the load frequency control system of multi-area hydro-thermal interconnected power grid on the basis of precise modeling of the regulating system of hydropower unit on different research scales.In addition,the stability design and analysis schemes of MPC are further explored to develop the research frame of MPC applications in the regulating system of hydropower unit.Some theoretical and applicable results have been achieved.The main work and innovative results of this thesis are stated as follows.(1)Based on the analysis of dynamic characteristics of the regulation system from different perspectives,the nonlinear distributed parameter model for the governing system of pumped storage unit,the affine nonlinear model for the excitation system of synchronous generator and the state-space model for the load frequency control(LFC)system of multi-area interconnected power grid are established.Besides,the structural differences between regulating system models at different scales and treatment of nonlinear constraints in the system are discussed.The work lay a foundation for the study of MPC for the control systems of hydropower unit under different control objectives.(2)Focused on the governing control of a special type of hydropower unit,i.e.,pumped storage unit,a generalized predictive control(GPC)method for the governing system of pumped storage unit that is adaptive to different operating conditions in power generation direction is proposed.The time delay and the order of the instantaneous linearized predictive model is determined by a novel time series model estimation and parameter reduction strategy combining mechanism modelling with data-driven methods.According to the different operating conditions and control objectives,design of the frequency regulation mode for start-up and no-load conditions and the guide vane opening regulation mode for grid-connection and running with load conditions in the proposed GPC are presented,respectively.The effectiveness of the proposed control scheme has been validated with the dynamic simulations in different operation conditions using practical parameters of the governing system of a pumped-storage unit with the capacity of 300 MW in a pumped storage power plant in Central China.(3)A tree-seed algorithm(TSA)based stair-like intelligent MPC method for the excitation system of synchronous generator is proposed.The stair-like control strategy and intelligent optimization algorithm are introduced to the rolling optimization of MPC to reduce the dimension of nonlinear programming solution and improve the feasibility of control algorithm effectively.In order to guarantee the closed-loop stability of control system,the concept of “quasi-infinite horizon optimization” is applied in the design of MPC of synchronous generator excitation system.The terminal cost term and terminal inequality region are added to rolling optimization and the “fictitous” local feedback controller beyond the prediction horizon is designed.The simulation experiment results in both single-machine and multi-machine power system have demonstrated that the proposed MPC can improve the terminal voltage stability under disturbances and faults as well as shorten the damping transients of the system.(4)The secondary frequency modulation of the automatic generation control(AGC)when hydropower units are connected to the power grid has been intensively studied and a distributed model predictive control based load frequency control(MPC-LFC)using Laguerre functions is proposed.The discrete-time Laguerre functions are utilized to approximate the control trajectory of the distributed MPC within prediction horizon.This mechanism fixes the dimension of the online optimization in the controller to the order of Laguerre function,thus to overcome the deficiency that the dimension of the decision variable increases with the increase of prediction horizon in traditional MPC methods,improve the online computation efficiency of the rolling optimization with long prediction horizon and perfect the frequency regulation performances of the interconnected system with complex state constraints.(5)The MPC algorithms in this thesis are divided into three types,i.e.,the Laguerre function-based distributed MPC for constrained linear system,the stair-like intelligent MPC for affine-nonlinear system and the GPC for complicated nonlinear system.For MPC in the constrained linear system and affine-nonlinear system,the closed-loop stability of the control system is proved by Lyapunov stability criterion via the terminal equality constraint and a “quasi-infinite horizon optimization” design,respectively.In addition,the MPC stability issues in application of LFC and generator excitation control are investigated.For GPC in the complex nonlinear system,the difficulty in stability design has been analyzed and the operation stability of the controlled system when GPC is applied to pumped storage governing system has been investigated by nonlinear dynamic methods.