The Use Of Cluster Analysis In Load Classification

With the development of the power system, numerical simulation has become the most important tool in power system planning and analysis. As an element of the simulation, load model affects the result greatly. However, it is widely known that the load modeling is a quite difficult problem due to its special properties such as time-variant characteristics of the loads which has been the limiting factor for power system simulation. In order to build practical load models, cluster analysis and multi-curve identification are used to classify and build the load models. Because the selection of the eigenvectors influences the results of the cluster analysis greatly, two main kinds of eigenvectors, the identified model parameters and the voltage responses of the load model, are also compared. In order to find the laws hidden in the load, further studies are carried on the classification results. With the purpose of enhancing the model generalization capability, a simple aggregation method is used after classifying the loads. Finally, a simple method for load classification is proposed. It also reveals that there are some laws in the loads which can be used in load modeling.