Research On Equivalent Modeling Of Distributed Generation For The Comprehensive Load

Power system simulation is to protect the power system planning, construction and analysis, and accurate component models is the basis for reliable power system simulation. With the development of the adjustment of the structure of the world's energy caused by the energy crisis gradual and smart grid, various types of distributed power supply, with its advantages in the aspects of environmental protection, convenience, reliability has been rapid development in the traditional power grid. Distributed power high permeability grid access will undoubtedly change the original energy flow and information flow, and thus have a profound impact on the terms of relay protection, voltage stability, power scheduling. Due to the greater human and material consumption caused by actual physical experiments of power grid, computer simulation is an important way to study such issues, thus establishing the equivalent model of distributed power is necessitated problem in present. Based on this, relying on the National Natural Science Foundation-funded projects "Containing multiple micro-power load modeling and its impact on grid stability mechanism (51277055)"this paper expand research on external characteristics equivalent modeling of several typical distributed power to provide the necessary support model for simulation of complex power systems with distributed power.Firstly,base on the physical mechanism of photovoltaic power system and MATLAB/simulink A simulation system for the grid-connected photovoltaic (PV) power generation system is developed. The influence of the grid-connected PV power generation system on the comprehensive load characteristic of distribution network is studied. And several equivalent models available to the PV power generation system are analyzed and compared. On this basis, a new equivalent model based on transfer functions for the PV power generation system is proposed. The model has a simple structure, less parameters, and can effectively describe the output limit of PV power generation system. At the same time, it is easy to implement in the power system simulation software. A generalized comprehensive load model is developed by paralleling the conventional comprehensive load model with the new equivalent model to describe the characteristic of the comprehensive load containing PV power generation systems. Simulation results show that the generalized comprehensive load model has good generalization ability and stable identification parameters, while meeting the need of engineering simulation.Next, the SOFC (Solid Oxide Fuel Cell) and its grid-connected system is established base on Matlab/Simulink, and the external characteristics of SOFC power systems is researched; then a equivalent model of SOFC power generation system is proposed from the perspective of external characteristic equivalent fitting and ignoring the reactive power of the SOFC generation system. This equivalent model has a simple structure, less parameters, can be identified easily. It can fit the external characteristics of SOFC power systems effectively in the case of large disturbance and continuous small disturbance of power system, compared with other equivalent models, this model can simulate the limit of the fuel cell's output; the synthesis load model multiplied with equivalent model is used to fitting the characteristics of power synthesis load containing the SOFC power systems, examples show that this generalized load model can describe the load characteristic of distribution network with fuel cell power generation system, and identification result of it's parameters are stability.Secondly,This paper overviews the application of the various neural network model in power system load modeling, and compares their merits and defects. Feedback branch of Elman neural network for has been improved to make it meet the needs of unified equivalent modeling of distributed generation. A combination ideas between an improved Elman neural network and the power system analysis software package (PSASP) has been proposed. Simulation examples strongly show that the improved Elman neural network can effectively describe the external characteristics of a variety of distributed generation and meet the needs of engineering simulation.Finally, A complex load modeling simulation system containing micro power is constructed, this system combined system simulation, modeling components, synthesis load modeling into one product, it includes three kinds of identification algorithms-PSO, improved genetic algorithms and simulated annealing algorithm, and reserve the user-defined modules. It can not only show the results of this study, but also to meet the needs of power system simulation modeling and parameter identification.