Study On The Identification Of Power System Transient Stability Rules Based On Margin Index

At present,the online operation security analysis of large power grid has become an important part of the operation of power dispatching,and the analysis of online security and stability has been integrated into the core business process of the dispatching operation.With the access of large-scale renewable energy and the expansion of the scale of power grid interconnection,the security and stability analysis of power system and the control of dispatching operation are facing more acid tests.The traditional production mode of "artificial experience judgment and then online calculation assistant decision-making" has been unable to meet the complex and changeable dispatching operation needs of the current power grid.Therefore,it is necessary to excavate the monitoring data of the power grid deeply and evaluate the transient stability of the power system without relying on the model.Through the current running state and historical data of the power grid,the stability level and the trend of the system are learned to help the operators to make the relevant measures to improve the stability of the power grid in time.The existing data driven power grid stability evaluation methods just judge whether the system is stable or not.Most of them do not give a quantitative description of the stability degree of the power grid.While the wide range of running states and the lack of stability samples are ubiquity in power grid data which affecting the efficiency and accuracy of the stability evaluation.In this paper,a power system stability rule identification method based on multiple attribute decision tree and power grid transient stability margin index is proposed.Mutual information is used to select the key running characteristics of the power grid.Then,the multi-attribute decision tree is constructed by feature transformation and stability margin discretization.The principle of obtaining the stable association rules is proposed,and the measures to improve the stability of the system are determined through the back-stepping analysis of the decision tree.Finally,the power grid operation time series data are analyzed,and the prediction method for the trend of power grid stability is put forward.The specific work is as follows:(1)To obtain the key operation features of power grid based on mutual information.Mutual information is used to select the key operation feature sets of power grid.The key electrical features which have greater correlation with the stability of the system are selected by calculating the mutual information between the features and the stability margin of the system getting from the simulation samples.By calculating the mutual information between multiple electrical feature combinations and stability margin,the correlation degree between the combination feature and the stability margin is obtained.A large number of simulation samples are obtained by using the IEEE39 node system,and the correctness and effectiveness of the above methods are analyzed and verified.(2)A method for identification of power system stability association rules based on multiple attribute decision tree is presented.The LDA transformation of the key feature set of the power grid is carried out.The multi attribute decision tree which can be used to stabilize the rule identification is generated for the specific fault based on the discretization of the stability margin,and the principles to obtain the relevant rules are put forward.According to the multi-attribute decision tree and the rules,a strategy of improving the stability of power grid based on decision tree back-stepping is proposed.Simulation of a specific fault in the IEEE39 node system verifies the feasibility and effectiveness of the proposed method?(3)The prediction method of power grid stability and change trend is presented based on the time series analysis method.Based on the K-means algorithm,the stability margin time series data are clustered and the stability margin trend of different samples is defined.The correlation between the change of tidal current section and the stability of the system is analyzed based on the time series variation data of power flow operation state.The reference sequence of each category is determined based on the separation index,and the time series trend category of each sample's stability margin is judged.The method of similarity matching is used toevaluate the change of power grid stability margin and determine its category,and an accelerated search method is proposed to improve the matching efficiency of similar series.Finally,the effectiveness of the proposed method is verified by simulation.