Studies On Application Of Model Predictive Control To Wind Farm Frequency And Voltage Regulation

With the development of economy and society,energy demands change towards clean and low-carbon way gradually,and the production activities and daily life of people depend more and more on electrical power.In the past two decades,as the power system scale expands and renewable energy integration such as wind power grows,the risk of massive blackout accidents always exists,which severely threatens the life,property and society stability.The fast power system restoration can effectively reduce the loss of the blackout accident,and the large-scale wind power integration brings new challenges to restoration.Black-start(BS)is the primary step of power system restoration,utilizing wind farm(WF)as BS source can effectively increase the restoration capability of local power grid.Therefore,comprehensive studies on the feasibility of utilizing WF as BS source and the key technical problems have important theoretical significance and strong farsightedness.The main challenges of using WF as BS source are the lack of self-starting ability,its weak ability of handling frequency and voltage disturbances,and the fluctuation and intermittence of WF output power.Energy storage system(ESS)can make WF capable of providing BS capability.Wind turbine generators(WTGs),ESS and static var generator(SVG)are coordinated by advanced control techniques to deal with wind power uncertainty and to enhance the ability for disturbance handling.Based on the current researches,exploiting Model Predictive Control(MPC),multi-objective optimization,Copula theory and Blade Element Momentum(BEM)theory etc.,the feasibility of WF providing BS capability,key technique problems such as frequency and voltage control are studied in this dissertation.The main contributions and innovations of this dissertation are summarized as follows:(1)For enabling WF to be a reliable BS source,an ESS sizing method based on copula theory is proposed,aiming at providing WF self-starting power,dealing with frequency deviation when starting up ancillary machines,and smoothing WF output power during the process of starting up generator unit.With the objective of minimizing capital cost,the power sizing and capacity sizing of ESS are considered as two random variables and the correlation between them is analyzed to show its asymmetric characteristics.To model this complicated dependence structure,joint probability distribution is built by asymmetric copula function.The marginal distributions are obtained by kernel density estimation.The most suitable copula function is determined by maximum likelihood estimation and a goodness of fit test.The case study with real WF data demonstrates that compared with the symmetric copula function,the proposed asymmetric copula function is more accurate and ESS sizing cost is lower consequently.(2)In high wind power penetration power system,using wind farm equipped with energy storage system(WF-ESS)as BS source needs to maintain system frequency stability when starting up ancillary machines.A hierarchical model predictive control(HMPC)strategy of WF-ESS for frequency regulation during BS is proposed in this paper,which consists of two control modes:frequency regulation mode and reserve recovery mode.For the first control mode,a PI controller is used to generate reference power for WF-ESS and to recover system frequency to 50Hz.The proposed HMPC is divided into WF level and WTG level to accurately track the reference power and guarantee secure operation of each WTG.Considering WTGs'different operation conditions,WF level MPC distributes reference power to WTGs according to the proportion of single WTG reserve to total WF reserve.The ESS power is utilized as supplementary when WF output power is insufficient.The rotor speed and pitch angle constraints are included in the WTG level MPC while providing frequency response.After frequency deviation is eliminated,WF and ESS are coordinated by the second control mode to avoid a second frequency drop meanwhile reserve a certain proportion of WF available power for managing potential frequency disturbances in the future.The case study shows that the hierarchical control structure can reduce the computational complexity and control error,and the secure operation of WTGs is guaranteed.(3)A coordinated voltage control method based on MPC is proposed in this paper for WF as BS source to start up a thermal generator unit.The reactive power regulation devices with different dynamic response characteristics including WTGs,ESS and SVG are coordinated by the proposed MPC to handle voltage and frequency disturbances during ancillary machine start.The voltage model of MPC is built based on voltage sensitivity coefficients for improving computational efficiency.When the bus voltages are within their constraints,the reactive power distribution between WTGs is optimized to maximize the dynamic reactive power reserve.When the bus voltage constraints are violated under disturbances,the dynamic reactive power of ESS and SVG are fully exploited to accelerate voltage recovery and prevent WTGs from tripping incidents.The impact of active power on bus voltage variations is considered due to low X/R ratio.The reactive and active power of WTGs and ESS are coordinately controlled for handling both voltage and frequency disturbances simultaneously.A WF with 33 WTGs rated 1.5MW is used in case study to demonstrate the enhanced disturbance handling capability of the proposed control during BS progress.(4)A MPC strategy for WTGs is proposed to improve the security operation of WT by reducing WT load on the blades.The blade load model is based on the BEM theory.The generator speed and pitch angle are simultaneously regulated to realize the control objectives.A two-stage optimization is designed in order to reduce the computational complexity.The objectives of the first stage are minimizing the ramping rate and maximizing the power generation.A trade-off is made between the two contradictory objectives by setting weight coefficients.The second stage reduces the WTG load and meanwhile guarantees the power reference from the first stage is tracked.Feedback is designed based on neural network prediction to compensate the error of the prediction model.Case studies show that the proposed control can significantly reduce the mechanical load on WTG blades under extreme weather conditions and thus the safe operation of WTGs is improved.