Active Frequency Response Control Of Bulk Power System Following A Major Power Deficit

To solve the frequency stability problems of power grid caused by the high-power loss of ultra-high voltage faults and the high penetration levels of renewable energy,the gradually enriching frequency response regulation means are needed to be better coordinated.Thus,some certain limitations of the current frequency response regulation method are presented.Since the advanced communication technologies and the perfect control systems have existed in the security and stability control framework of power system in China,new frequency response regulation method can be proposed without adding more hardware facilities besides the communication devices and channels,etc.On the basis of the time-space distribution characteristics of the frequency in bulk power system,the control idea and control framework of the active frequency response(AFR)are proposed.The frequency response control is transformed from the decentralized proportional feedback control which is regulated based on the local frequency deviation to the centralized event/parameter feedforward control which is regulated based on the parameter of the disturbed bus.By the optimization and coordination of various frequency response regulation means in the "source-grid-load-storage" ends,the control capability of the existing equipments is fully utilized.Thus,the frequency stability control capability of power system is improved and the tension of system frequency stability following a major power deficit is improved.The proposed AFR control framework is consist of off-line analysis and on-line application.Among which,the typical scenario generation,frequency security level grading,and coherency identification of generators are the major concerns of off-line analysis.In off-line analysis,the typical scenarios are generated based on the historical operating data of power grid,the frequency security levels are graded for the typical scenarios,and the coherent generators are identified for the typical scenarios with lower frequency security levels.Thus,the workload of off-line analysis is reduced and the rapidity demand of frequency response control is satisfied.In on-line application,the AFR based on model predictive control(MPC)is the research focus.By matching the current operation scenario to the typical scenario obtained in off-line analysis,various frequency response regulation means in power system are centrally coordinated and controlled.Thus,the whole frequency response capability of power system is fully utilized.The research on each content is as follows.By using the frequency nadirs as the clustering inputs of operation scenarios,the effect of each system parameter is comprehensively considered.With the improved fuzzy C-means clustering algorithm,the operation scenarios are clustered and selected based on the cluster validity indexes and the frequency security principle.Large number of operation scenarios and multiple factors involved in the control strategies faced by AFR control are solved.Thus,the control efficiency is improved on the premise of ensuring the control accuracy.By demonstrating that the ratio between the frequency nadir and the quasi-stable state frequency of power system is nearly a constant after disturbance,the frequency nadirs of power system are estimated based on the quasi-stable state frequency with the linear regression method of historical data.The frequency security level grading standard is proposed and the frequency response reserve of generator units and the initial operation frequency of power system,etc.are considered.Thus,the frequency security level and the actual frequency response capability of power system can be real-time evaluated based on the frequency nadir.The theoretical basis can be provided for the selection of the frequency response control modes and the dispatch of the frequency response regulation means.By using the frequency as the coherency identification basis of generators,the influence of the model parameters,operation modes,and disturbance locations,etc.of power system operation on the clustering results can be comprehensively considered.With the support vector clustering algorithm,the time series of generator frequency deviation are mapped from the initial space to the high-dimensional feature space.The radius of the minimal enclosing sphere which can envelop all data samples is obtained with the sequence minimum optimization algorithm.Thus,the coherent generators are identified based on the evaluating index of cluster compactness and separation.With the AFR clustering control,not only the frequency nadir of power system is little affected,but also the transient maximum frequency of generators connected to non-disturbed bus can be controlled within the security constraints of system frequency.New frequency security and stability problems of power system can be avoided.By using MPC in AFR control,not only the control hysteresis problems caused by the existing frequency response delay are overcome,but also the regulation characteristics of various frequency response regulation means and the operation constraints of power system are comprehensively considered.Moreover,on the premise of ensuring the operation security of power system,the output power of each frequency response regulation means is centrally coordinated and controlled.The whole frequency response capability of power system is fully utilized and thus the frequency stability resistance ability of power system following a major power deficit is improved.The analysis of the electrical distance between the controlled frequency response regulation means and the disturbed bus shows that the expected effect can be achieved in a certain distance from the disturbed bus.The theoretical basis and the implementation foundation required in the above researches have existed in the security and stability control framework of power system in China.No new hardware facilities besides the communication devices and channels,stability control systems,etc.are needed in this paper.The simulation and analysis results show that the proposed method is feasible and effective.