The Realization Of Load Model Parameter Library Based On Synthesis And Classification

With the increase of the field-based data, it becomes harder and harder to deal with them effectively. In order to solve the problem, this thesis gives detailed ideas on the synthesis and classification of dynamic loads. On this basis, through the Pattern Recognition with Neural Network, a load model classifier is realized. With the classifier, field-based data are classified to different classes to construct corresponding class libraries. And then a synthesis-based load model can be computed by the data in corresponding classes.First of all, the distance between different models' parameter is regarded as the index of classification, and then the percent of false classification is defined as the index of evaluating the model's structure. Moreover, in order to improve the precision of classification and link loads' working condition with model classification, considering Neural Networks' ability to solve fuzzy, uncertain,non-linear problems, through the Pattern Recognition with Neural Networks, a BP Network classifier is realized. With this classifier, real field data can be classified to different class libraries.