Research On Risk Assessment Of Power Network Dispatching Based On SVM

With the continuous expansion of the scale of the power grid and the increasing dependence of human society on power,the security of grid dispatching operations has become an issue of increasing concern.In this paper,risk assessment is used to consider the safety level of grid dispatching operation.Support vector machines have good generalization performance.At this stage,it is a trend to apply the support vector machine method to grid risk assessment.To this end,this paper has carried out the following research to provide a theoretical basis for risk assessment in grid dispatching operations:1.Aiming at the shortcomings of the day-to-day scheduling process and the traditional method of power flow risk assessment using power flow calculation,a risk assessment algorithm for power grid dispatching based on support vector machine and principal component analysis is proposed.Firstly,the traditional power flow risk assessment method based on power flow calculation is introduced.Secondly,the process of determining the expected faults in the pre-scheduling phase is briefly described.Next,the theoretical model of support vector classification is given,and the principal component analysis method(PCA)is introduced to give the specific step algorithm of principal component analysis.On the basis of these theories,select the appropriate "training time",conduct risk assessment on the"training time" grid,evaluate the data as the original training sample,and then send it to the support vector classifier for training after the principal component analysis method.The grid operation data of the target time is collected as the input part of the original test sample.After being processed by the principal component analysis method,it is sent to the trained support vector classifier,and the grid risk grading result of the current target time is obtained according to the test output.Continuously select new day-to-day target time to complete the grid power dispatch risk assessment2.Considering the characteristics of real-time scheduling process,a real-time scheduling risk assessment algorithm for support vector machine grid combined with K-means clustering is proposed.First,the process of determining the expected fault in the real-time scheduling phase is briefly described.Secondly,the mathematical model and algorithm of K-means are introduced.Since the support vector regression is used in the real-time risk assessment algorithm,the principle of the support vector machine to solve the regression problem is introduced.On the basis of these theories,select the appropriate"training time",conduct risk assessment on the"training time"grid,evaluate the data as the original training sample,and send it to the regression support vector machine for training after K-means clustering.The real-time grid operation data is collected,sent to the trained regression support vector machine,and the real-time grid risk determination result is obtained according to the test output3.For the problems existing in the operation of a power grid in a southeastern coastal province and the actual needs of power companies,combined with the analysis and processing of heterogeneous information related to power system risk,the mechanism of grid uncertainty factors,grid risk assessment and dynamic control strategies,etc.The research results obtained in many aspects have developed the technical support system for the whole process risk tracking and dynamic regulation of the power grid dispatching plan,and the development purpose and significance of the system,the main functional modules,the system hardware and software architecture and the function realization of the system have been carried out.Description.At the same time,because the system includes the risk assessment and real-time risk assessment functions,the same work is done in Chapter 2 and Chapter 3 of this paper.However,when calculating the date and real-time risk assessment data in the background of the system,the traditional risk assessment method based on power flow calculation is adopted for the sake of reliability and maturity,and the evaluation result of one moment is calculated every time in the background.The front end needs to wait for a while,which is the driving force behind the author’s search for a faster risk assessment method.This led to the study of SVM-based assessment methods in Chapters 2 and 3.By comparison,it is found that at the same time,the system displays the results of the previous risk assessment consistent with the results given in the analysis of the example in Chapter 2.Therefore,the effectiveness of the proposed method in Chapter 2 is further illustrated.