The electrical power system load modeling researchers have always hoped to establish the practical load model which fully reflects the load physical property.As the load characteristic measurement-recording instruments are gradually applied, massive field data is recorded to use in the dynamic characteristic research,which has provided the sufficient dynamic data for the establishment of accurate load model.In order to excavate the information of measured data, the thesis uses the synthesis optimization algorithm to identify model parameters with improved synthesis load model, and also has analyzed the load dynamic model parameters'statistics characteristic. On one hand, load classical statistical characteristics are analyzed. On the other hand, the thesis includes the analysis on Support Vectors which represents all sample data, make relations between SVs and load physical natures,and reveals why SVs can represent all samples.A new load modeling methoed is proposed,which is based on pattern recognition,data filting and clustering. As the method includes the effects of different voltage, it has good generalization capability.
|