The gradual learning static load modelling method based on real-time fault recorder data
Guoping Shi1, 2, Jun Liang1
COMPUTER MODELLING & NEW TECHNOLOGIES 18(5) 297-302
1School of Electrical Engineering, Shandong University, Jinan City, China, 250061
2School of Information and Electrical Engineering, Shandong Jianzhu University, Jinan City, China, 250101
Setting a real-time load model is an effective way to overcome time-variation of power load in course of power load modelling. On the basis of load data sorting, this paper proposes a gradual learning static load modelling method based on power fault recorder data. Firstly, power fault recorder collects and stores valid load data. Secondly, all valid load data will be classified by the time, static load model can be built corresponds to each classification. Thirdly, model parameters of each sort are identified by gradual learning method, for the goal of global fitting optimal for the measured active power and calculated active power, the load model parameters are optimized by using curve fitting method. The identified model parameters can be applied to power system calculation directly without preserving all load data, essential feature of all load data is reserved and modelling operational efficiency is improved greatly. Simulation results show that the gradual learning method is right and effective, which is easier to realize and is of higher precision compared with least squares method, therefore the method has widely applicable value and is prospective in power system on-line static load modelling.