Research On Key Techniques In Situation Awareness For Active Distribution Systems With Multi-microgrids Operation

With the advancement of the Energy Internet construction and the wide penetration of advanced information and communication technology,the main body of the power grid will tend to be more complex and diversified,and the information type will tend to be heterogeneous and multi-source.Situational awareness technology is of vital importance to improve the "perception and prediction ability" and "decision support ability" of power grid.However,the existing power grid situation awareness system cannot meet the development of smart distribution network,including measurement information accuracy assessment,uncertainty factor prediction,flexible resource potential mining,multi-users' power interchange capabilities evaluation and so on.At the same time,the formulation of the "three types and two networks" strategy of the State Grid Corporation of China will be accompanied with the vigorous promotion of the construction of the ubiquitous power Internet of things.The improvement of situation awareness system for medium and low voltage distribution network is of great theoretical and practical significance for realizing high-precision data perception,deep mining of potential commercial value from energy and environmental data,and improving the optimization and coordination performance at the edge of power grid.In order to ensure situation awareness and optimal dispatching for active distribution network with multi-microgrids,this paper carries out relevant research from the perspectives of situation comprehensive,situation prediction and situation orientation.Firstly,the random matrix theory(RMT)based method is used to effectively identify anomaly measurements and to accurately perceive the operation status of the system.Secondly,a probability prediction method based on the Gauss mixture model(GMM)and parallel hidden Markov model(PHMM)is proposed to obtain the probability distribution interval and variation law of renewable energy output under multiple weather conditions for achieving situation prediction.From the standpoint of operator,model predictive control theory(MPC)and optimal dispatching method are used to ensure the security and economy of interaction and cooperation between multi-microgrids and active distribution network.This method can reduce the impact of uncertainties and realize situation orientation reliably.Specific research contents are as follows:(1)A situation comprehensive technology based on data-and model-driven methods is proposed.Considering bad data caused by communication anomalies or human factors,this paper uses RMT to analyze the correlation between measurements and detect abnormal measurements through real-time monitoring of correlation changes based on a data monitoring platform composed of smart meters and advanced metering infrastructure.Secondly,edge computing based anomaly measurement identification algorithm considering network partition is proposed to effectively reduce the real-time data transmission capacity and computational burden,and facilitate the distributed processing of data.The anomaly measurement identification results are used by the state estimation method to realize the accurate perception of the current operating state of the system.Finally,on the basis of results of accurate perception,the RMT based high-dimensional statistical indicator is calculated for evaluating the security and stability of power grid.An example shows the advantages of the proposed methods in the measurement perception accuracy and security operation evaluation.(2)A multi-time scale probabilistic prediction method is proposed for various type of renewable energy.Considering the fact that the uncertain distribution of forecasting data,this chapter utilizes the GMM to fit the probability distribution of the observation data firstly.On this basis,the time series of the forecast object in weather-sensitive regions are extracted by multi-stage pretreatment,which consists of the information entropy correlation analysis and the time series synthesis.Finally,the PHMM is used to train the observation sequence,and Bayesian formula and data distribution characteristics are used to realize multi-time scale probability prediction.The fluctuation law of photovoltaic output under different weather conditions was analyzed by numerical examples.(3)A multi-time scale demand response resource optimal dispatching method is proposed,which enable microgrid participates in the short-term electricity market through real-time sensing of response resources of microgrid.A day-ahead optimal economic dispatching model is established based on the operation cost of microgrid and the compensation benefits of demand response.In order to correct the predictive deviation of renewable energy and load,a rolling optimization method based on MPC is proposed to minimize the tie-line power difference and SOC(State Of Charge)deviation of energy storage.In the dispatching model,the adjustable capacity ratio factor is introduced to ensure the certain adjustable capability of microgrid while consuming renewable energy sources.The influence of adjustable capacity ratio factor on power tracking effect and demand response capability of tie-line in microgrid is discussed.(4)An active distribution network situation orientation strategy based on power interchange capability is proposed,which ensures economic operation and improves the power supply capability of the distribution network for dealing with uncertain abnormal events.The proposed MPC based predictor-corrector model is used to perceive the power exchange capacity between the multiple microgrid and the distribution network dynamically.On this basis,the optimal bidding strategies of each microgrid in each period are determined through the multiple microgrids bidding strategy model.The bidding results are conducive to the realization of situation orientation in the interaction process between active distribution network and multiple microgrids,and achieve win-win situation for both sides.In addition,the mixed method of point estimation and second-order cone programming is used to solve this proposed model,which ensures the accuracy of the optimization model while considering the uncertainties in the model.The flexibility of the interaction strategy is verified by the test and analysis of an example.This strategy can provide optimization guidance for the efficient application of demand response resources in the multiple microgrids.Considering the research hotspots and development trends in this field,this paper proposed a situation awareness and situation orientation method for active distribution networks with multi-microgrids using data-and model-driven methods.Through a large number of simulation and real-world examples,it is proved that the researches of this paper can help improve the ability of situation awareness and optimal scheduling for complex systems,and provide theoretical reference for resource awareness and edge calculation of future distribution networks and Energy Internet.