Research On Load Modeling Generalization Ability In Power System

The digital simulation of power system is the core technology for power system planning and design, operational controlling, analysis and research. Load modeling has become a critical problem that is urgent to be solved in the digital simulation power system. It is of great importance to build an accurate load model for stability operation and safety administration of power system. In order to promote the study on the practicability of load modeling field, this paper studies some problems in the practical process of load modeling.It is a crucial first step to select an accurate model structure in load modeling field. This paper compares structure and classification of load model, including static load model, dynamic load model, comprehensive load model. Further introduces the principle and the advantages and disadvantages of the different load model. With the scale of power grid constantly expanding and the fluctuations frequent. It cannot accurately describe the load characteristics to use static model or dynamic model alone. The comprehensive load model has many advantages, such as load capacity of adaptive, data structure self-describing, low dimension. So the paper chooses comprehensive load model as simulation model structure.It can effectively improve the accuracy of load modeling to choose a proper modeling method. This paper carefully introduces commonly used modeling methods, including component based approach, measurement based approach, simulation based approach, hybrid approach. On the basis of analyzing the principle of the modeling methods, this paper compares the advantages and disadvantages of the modeling methods. Construction of a 39-node system model in Power System Analysis Software Package, it obtains a large amount of measured data as the basic data of load modeling with setting the different system faults. Though searching the model parameters in MATLAB to complete simulations and experiments, it chooses the best modeling method. Considering the applicable environment of measurement based approach and comprehensive measurement approach, this paper makes use of measurement based methodology to establish load model.It can greatly improve the efficiency of loading modeling to use a correct parameter identification method. Based on introducing model parameter identification, this paper analyses the principle and methods of identification in more detail, and studies the least square method and the particle swarm optimization algorithm. It chooses particle swarm optimization algorithm to identify model parameters by the comparison of two identification methods. At the same time, the simulation research validates the identifiability and dispersion of model parameters by means of in-depth analysis of the characteristics of the model parameters and the range of each model parameter.Load model generalization ability is the crucial dilemmas of the practical application and extension of load modeling. This paper studies on load model generalization based on previous research on the generalization. Then it proposes the research direction of load model generalization ability and concludes present research achievement of generalization ability. Finally, it focuses on the analysis of generalization ability of load model in different operation system. The basic data obtained by the simulation system is used for example simulation for generalization ability in different fault types and different load levels. The paper performs previous work that is the foundation of current study after the discussion and analysis. It has important significance to promote the practical process of the load model.