Study On Optimal Dispatching Strategy Of Active Distribution Network Based On Risk Reserve

In order to cope with the trend of high proportion of distributed power supply and diverse energy types,improve the power grid's ability to accept new energy,better integrate diverse energy structures and realize more intelligent and comprehensive operation and control of the power grid,the concept of active distribution network was came up.Active power distribution network is composed of distributed generation(DG),electrical energy storage(EES),controllable load(CL)and others.It is an intelligent power distribution system which realizes active control and management of power grid.Social growth and economic development have driven the growth of users' electricity consumption,and emerging loads such as electric vehicles have made the electricity environment more complex.In addition,the distributed energy output accessed by the active distribution network has obvious fluctuation,so accurate load forecasting is crucial to the balanced and efficient dispatching of the active distribution network.On the other hand,the uncertainty of distributed energy output requires more spare capacity to ensure the stable operation of the system,and the backup service,which is an important auxiliary service of the active distribution network,faces the problem of high reserve cost and uncertainty of usage.Therefore,the risk reserve is an evaluation index that can balance the reliability and economy of spare capacity.It is of great significance for the setting of spare capacity and the development of active distribution network.In this paper,the following related researches are carried out on the active distribution network scheduling strategy and the load forecasting and spare capacity affecting the operation of the active distribution network:(1)In view of the diversified and complicated power environment with load forecasting objects,this paper proposes a NE-GMDH(Elastic net-group Method of Data Handling)algorithm based on Elastic net regression for the problem that the classical GMDH(Group Method of Data Handling)algorithm is prone to over fitting and unable to handle multicollinearity in the complex system.At same time,this paper proposes a variable weight weighting algorithm based on the degree of historical data difference to improve the accuracy of the model.Then,10 characteristic variables such as meteorological factor and holiday factor were used as input variables to establish the short-term load prediction model,and the performance of the model was tested and analyzed by time series cross-validation in the analysis of calculation examples.(2)According to the requirements of reliability and economic evaluation of reservecapacity,this paper first introduces the concept and significance of risk assessment,and sorts out the main methods of risk assessment;Then,in view of the limitation of the traditional solution method of risk standby,combined with the time series characteristics of historical data of power system and the limited sample size,the risk standby solution method of Bootstrap was proposed,and the results of the solution of risk standby were tested and analyzed with Kupiec test in a treatment case.(3)Finally,this paper takes the optimization of active distribution network scheduling strategy as the objective,gives the constraint conditions of risk reserve in active distribution network scheduling,considers the difference of multiple reserve capacity costs,and gives the backup service selection results based on different confidence levels of risk reserve in the scheduling results.Then,aiming at the lowest operation cost of the network,this paper puts forward the scheduling strategy solution method based on Hopfield neural network,introduces the principle of energy function,mapping method and solving steps in detail,and analyzes the model performance in the example analysis.