Operation State Awareness Method Of Power Systems Based On Wide-area Measurement Data

With the proposal of "double carbon" goal,a higher proportion of renewable energy will be penetrated to power systems in the future,and its intermittent and uncertain output would bring severe challenges to power systems.Therefore,further research on power system operation state awareness are of great significance to the future secure and stable operation of power grids.At the same time,with the decrease of cost and the development of communication technology,the wide area measurement system(WAMS)has been widely deployed in power systems.Compared with the supervisory control and data acquisition(SCADA)system,WAMS’s high-precision and synchronous measurement data can further improve the state awareness ability of power system operators.This dissertation studies identification methods of power system false data injection attacks(FDIAs),abnormal event detection method of power systems,event location estimation and identification methods of power systems,online coherency identification method of power systems,and robust controlled islanding model of power systems,respectively;and tries to build the power system operation state awareness system based on the wide-area measurement data.The main research of this dissertation are as follows.(1)In view of the problem of FDIA in power systems,FDIA identification methods corresponding to two different scenarios with high and low data reporting rates are proposed respectively.For the synchrophasor measurement devices(SMDs)with high data reporting rates,the ensemble empirical mode decomposition(EEMD)algorithm is proposed to extract the intrinsic modal functions,which can reflect the multi-time scale characteristics of frequency data.The FDIA identification method based on fast Fourier transform(FFT)and feedforward neural network(FNN)are proposed as well.For widely deployed SMDs with low data reporting rates,a time-series feature extraction algorithm based on 1-dimensional convolutional neural network(1D-CNN)is proposed.Furthermore,a FDIA identification method for power systems based on gated recurrent unit(GRU)network is proposed.Case studies on the real-time measurement data in FNET/Grid Eye show that the proposed two methods can both achieve high identification accuracy in a short time window.(2)In view of the problem that the existing WAMS has a large communication burden and is not accurate enough to detect power abnormal events,a power system event detection method based on data compression and local outlier factor(LOF)is proposed.First,the measurement data is compressed in the WAMS sub-station and reconstructed in the WAMS master station using the unequal compression algorithm,so as to reduce the burden of the communication system.Then,the similarity search method based on principal component analysis(PCA)is used to determine the similarity between each pair of buses in power systems.Finally,the bus that differs greatly from the overall operating state of the system is found and abnormal events in power systems can be detected based on the LOF.Case studies on the WECC 179-bus system and actual power systems show that the proposed method can achieve accurate event detection and location results.(3)In view of the problem that the existing power system event identification methods fail to consider the time and location of event occurrence and the imbalance of training samples,an integrated power system event detection,location and identification method suitable for unbalanced samples is proposed.First,the rate of change of frequency(Ro Co F)is used to detect abnormal events in power systems and determine their occurrence time.Then,the abnormal events of power systems are located according to the time difference of arrival of disturbance wave.Finally,considering the determined time and location,an identification method of power system abnormal events is proposed based on 2-dimensional orthogonal locality preserving projection and random undersampling boosted trees.Case studies on the Eastern Interconnection(EI)system of United States show that the proposed method can achieve high accuracy for event location and event identification.(4)In view of the problem that most of the existing power system identification methods are based on the static physical model and cannot account for the real-time operating state of the system well,an online coherent generator identification method based on the fuzzy equivalent relationship(FER)clustering is proposed.First,several coherency indexes representing the generator trajectory similarity are proposed based on the real-time data measured by WAMS.Then,a method to determine the weights and decision matrix of generator coherency indexes based on the combined weight method and technique for order preference by similarity to an ideal solution(TOPSIS)is proposed.Finally,a method to divide the coherent generator group in power systems based on FER clustering and a method to determine the optimal coherent generator group based on F-statistics are proposed.Case studies on the 16-machine 68-bus system and actual power systems show that the proposed method can divide the coherent generator reasonably.(5)In view of the problem that the existing controlled islanding methods of power systems do not consider coherency constraints and the fluctuation of renewable energy,a controlled islanding model considering online coherency constraints and adjustable robust programming is constructed.First,a soft partition method based on Fuzzy Cmeans(FCM)clustering and its membership degree is proposed.Then,the coherency constraints of power system islanding control based on online coherency identification and fictitious power flow are constructed.Finally,an adjustable robust controlled islanding model is proposed considering the dual uncertainties of source and load.Case studies on the IEEE 39-bus system and WECC 179-bus system show that the proposed method can minimize the load shedding under extreme situations.The FDIA identification method,power system event detection,location and identification method,online coherency identification method,and controlled islanding method of power systems proposed in this dissertation enrich the operation state awareness theory of power systems,and they can help to provide corresponding supports for power system dispatchers and operators.