Load Modeling Platform Database Based On SOM And Wavelet Transform Data Mining

Power system load modeling is an important foundation of the power system simulation. Simulation using the typical parameters of the conventional load model is inconsistent with the actual situation. The development of the power system load modeling platform based on the measurement method and statistical synthesis method is an important way to the load modeling project. The efficient management of the load characteristics data is important to ensure the effective operation of the load modeling platform. Based on a large number of loads modeling data, the paper carries initial excavation. And the load platform modeling database is constructed on this basis.Firstly, according to the substation load characteristics of variability and region, the paper uses the SOM neural network clustering data mining method to cluster the Shandong substations. According to the clustering results, seven substation categories are appropriate. At the same time, various types of substation load characteristics results are obtained by the cluster centers. Then the corresponding typical substations are selected. Besides that, the new substation category is judged without having to re-adjust the existing categories. To substation measurement data, this paper uses the wavelet transform method for data mining. As simulation examples, the Shandong University dynamic model test failure data and the substation measured dynamic data are analyzed by the wavelet transform. Using the wavelet transform modulus maxima principle, the dynamic disturbance detection and time positioning of the data is obtained, verifying the validity of the method.Secondly, this paper uses the Shandong power grid load characteristic data and substation disturbance measurements to build the load modeling platform database. The load modeling platform is based on the statistical synthesis method to carry on the industry and substation modeling, based on the measurement method to carry on the static, mechanism and non-mechanism modeling. According to the principle of the platform and taking into account the role of data mining in the load modeling, this paper using the SQL Server2000to design and build the databases and the database tables. As the result, eleven databases are built and the data is imported. As well as to achieve the effective management of the database, the paper uses Visual Basic6.0to develop the database management system. The system realizes the load modeling data query, traversal and other basic functions. And in the aspect of the data mining, the system can use the wavelet transform method to identify the disturbance of the substation measurement data in the measurement modeling. And the disturbance data automatically interception is obtained. Besides that, SOM clustering analysis of the substation load characteristics data in the statistical synthesis modeling method can be realized. Then the clustering results are obtained and the typical substations are selected. The mining load modeling data laid the foundation of the next load modeling process.In short, the load modeling platform database based on data mining is good at data processing. The database building for load modeling platform provides better decision support, reduces the modeling effort and increase the modeling effectiveness and efficiency.