Prediction And Analysis Of Total Transfer Capability Based On Machine Learning

The traditional section fine rule formulation uses multiple linear regression methods to transform the nonlinear power system into a linear model to solve the problem,and the calculation process is complicated.In recent years,machine learning has been widely used in various fields of power systems.Therefore,it is considered to use machine learning in the prediction and analysis of the total transfer capability of key sections of the power system.The actual operation data of the power system is abstracted into feature attributes.After dimensionality reduction,some attributes are selected as the key features of the system section.The selected feature attributes form a key feature attribute library,which is used as the input layer vector of the neural network.The artificial neural network is repeatedly trained to construct the nonlinear correspondence relationship between the key features of the system and the maximum transmission capacity of the section.Realizing the application of machine learning in the prediction of cross-section total transfer capability will be of great significance to the intelligent development of power grid control links.In order to obtain a large number of data samples needed for artificial neural network training,this paper uses the open function of the power system analysis software PSASP to construct a PSASP-MATLAB interconnection simulation interface,through which data is transmitted in real time,to help PSASP quickly complete the batch trend Calculate and save the calculated data to the mysql database in PSASP.For the problem of too much data in the power grid and too much redundancy between the data,The clustering algorithm is used to partition and reduce the dimensionality of the system data.After K-means clustering analysis,the initial feature set of the system is divided into different sub-sets,and the correlation coefficient between the sub-set data and the correlation coefficient between the data and the target set are obtained by defining the evaluation criteria.From the many initial feature attributes The attributes that can describe the total transfer capability of the critical section of the system are selected to form a key feature set,and finally the key feature selection verification is carried out through the IEEE 9-node system.Finally,the selected key feature set and a large number of data samples generated by interconnection simulation are used for artificial neural network training,and the non-linear mapping relationship between the feature attributes and the total transfer capability of the system section is obtained through the analysis of the 7-machine 36-node system of the PSASP system.The analysis of calculation examples shows that the prediction method of critical section total transfer capability based on artificial neural network not only meets the timeliness requirement,but also can effectively improve the accuracy of section total transfer capability prediction.