Study On Nonlinear Modeling,Identification And Control Optimization Of Pumped Turbine Regulation System

In recent years,the proportion of renewable energy power generation in China's energy structure has gradually increased,and the scale of clean energy power generation,such as hydropower,wind power,photoelectricity,has been increasing.Wind power and photoelectricity are intermittent energy with strong fluctuation,which seriously affects the stable operation of power grid.As the only power generation mode with both power generation and energy storage functions in hydropower,pumped storage power station can not only play the role of peak load regulation and frequency modulation,emergency standby,but also absorb the impact of intermittent energy on power grid.Therefore,in recent years,the construction of new pumped storage power station projects has been continuously started,and the single unit capacity of pumped storage units has been increasing,and the water diversion system has become increasingly complex.The traditional control method is difficult to meet the actual control requirements of pumped storage units,so it is urgent to study the control theory of pumped storage units.The regulation system of pumped storage unit is a complex system with strong nonlinear characteristics.There are great difficulties in precise modeling,identification and controlling optimization due to the strong nonlinear characteristics.In view of the above difficulties,the nonlinear characteristics of each main part of the regulation system are analyzed and applied to model by starting with the precise modeling of the regulation system of the pumped storage unit.On this basis,the problems of parameter identification,model identification and controlling optimization of the regulation system of the pumped storage unit are studied,and a new method is proposed in this paper.The new method is as follows:(1)The nonlinear characteristics of the pump turbine are mainly analyzed and the models of each part of the regulation system are established by taking each part of the regulation system as the study object.The improved Suter transform is used to preprocess the complete characteristic curve of the pump turbine,and then the least square support vector machine(LSSVM)is used to combine the preprocessed data to build the nonlinear model of the pump turbine.The hyperbolic tangent function is used in the model of the water diversion system.The dead zone,limit and other nonlinear links are considered in the executive structural mathematical model.And then,according to different research needs,the linear model,nonlinear model and numerical calculation model of the regulation system are established respectively,which provides theoretical basis for the identification and control optimization of the subsequent regulation system.(2)In this paper,the problem of parameter identification of the regulation system with known structure and unknown parameters is transformed into the problem of optimizing the nominal value of parameters.An improved whale optimization algorithm(MSWOA)with strong optimization ability is proposed.The linear model and nonlinear model of the regulation system of the pumped storage units are identified respectively by the algorithm.The identification results show that there is high accuracy identification of the linear model and nonlinear model of the regulation system by using the MSWOA algorithm.(3)The simulation platform of the pumped storage units is established and the nonlinear model of the pump turbine is the core.The unit speed output signal is obtained by the simulation platform under the band-limited white noise signal.The single input-output nonlinear autoregressive model(NARX model)is constructed according to the above signals and the training samples for identification are obtained.The identification model of Bi LSTM neural network is constructed and the optimal selection method of identification parameters of Bi LSTM neural network is studied by using the training samples.The identification results show that there is high accuracy identification of the Bi LSTM neural network identification model on the pumped storage units.(4)The corresponding controlling strategy is formulated for the model obtained from different identification.The linear model of the regulation system obtained by the parameter identification method is transformed into the uncertain singular time-delay system due to the hyperbolic tangent function used in the diversion system.The influence of parameters T_rand hw on the stability region of the system,and the change of performance index in the transition process under different PID parameters and different operating points are studied.Based on the above study,the H?controller is designed and transformed into an optimization problem with minimum attenuation?.The effectiveness of the design method is verified by experiments.PID controller is designed for the model of pumped storage units identified by Bi LSTM and the PID parameters are optimized by MSWOA algorithm.The PID controller is applied to the simulation platform of the pumped storage unit,and the transient process is calculated under three adjacent water head.The simulation results show that the time-domain stability performance indexes of the transient process under three adjacent water heads meet the requirements,and there is good applicability of the identification model.(5)In order to ensure the safety of the pumped storage units under the condition of pump outage and 100%load rejection,and effectively restrain the speed and pressure surge during the transition process,the impacts of different guide vane closing modes on the hydraulic characteristics of each hydraulic unit and the unit speed in the regulation system are studied by taking the pumping and storage power station with single pipe and single machine structure as the object of study.Based on the numerical calculation model of single pipe and single machine structure,an optimization model of guide vane closing law is established,and the improved multi-objective Gray Wolf algorithm is used to solve the optimization law of guide vane closing.The constraints of each link in the regulation system are considered in the model,and speed rise rate and the appreciation of water hammer pressure are selected as multi-objective to optimize objective function.The closing law of one-stage,two-stage and three-stage guide vanes are respectively optimized under load rejection and pump outage conditions.The effectiveness of the model is proved.