The Research On Power Load Modeling

The power load is an important constituent of the electrical power system, and its modeling work has always being a difficult problem of the power system, acquiring the domestic and foreign scholars' widespread attention. The load model has quite important influence to the electrical power system analysis and computation, thus further affects the electrical power system plan, the design and the control. The traditional static load model has small voltage change adaptation scope. At the same time, the dynamic load model takes load's variation by time and non-linear into account insufficiency. Therefore, in view of the insufficiency of the traditional load model, has conducted the research to the static load model and the dynamic load model establishment.Firstly, in view of traditional static load model limitation, static load model is established based on piecewise cubic Hermite interpolating polynomial. The modeling example based on the measured data approves that the static load model of piecewise cubic Hermite interpolating polynomial has a higher precision. Simulation to wide change of voltage shows the new model's stronger serviceability.Secondly, considering characteristics of the power load, as time denaturizing and non-linearity, and so on, dynamic load model will be established based on the method of time-process oriented and stacked neural network. In view of the characteristic of load's variation by time, according to the level of voltage and power on the generatrix, the courses are carved up for the first time. Then based on the responses to standard voltage excitation, the courses being divided firstly could be further thin delimited or merged. Finally, dynamic load model is established to the perturbation data in a course, using the stacked neural network, which has higher precision and strong non-linear mapping ability.Finally, the model structure is determined through modeling to the measured data. The active power and the reactive power coupling and uncoupling modeling indicated that the uncoupling model has a higher precision compared to the coupling model. Modeling example, based on curving courses up by time and responses to the standard voltage excitation and using stacked neural network, to...