Ultra-Short Time Reactive Load Forecasting Algorithm for Dynamic Reactive Power Compensation in Steel Rolling Workshop

摘要

The loads in steel rolling workshop are impact loads with no definite period, which seriously affect the power quality in the company power supply system and the working life of relevant equipment. Considering the special characteristics of loads in steel rolling workshop, some improvements for Sage-Husa adaptive filtering algorithm with estimator of time-variant variation noise is proposed in this paper. This new algorithm can avoid the shortcomings of Sage-Husa adaptive filtering algorithm, and effectively improve forecasting accuracy. Meanwhile, the improved algorithm inherits the advantages of the original algorithm, whose computing speed is rapid and storage capacity is small. The new algorithm is perfectly suitable for online applications. The reactive power forecasting results of some substation buses in an iron and steel enterprise show that the accuracy and computing rate of the algorithm is practical for dynamic reactive compensation. The application of this algorithm in improving the power quality and saving energy is encouraging.