Researches Of Power System Short-term Load Forecasting And Dynamic Reactive Power Optimization

Short-term load forecasting is an important factor of security dispatching and economically running, and it is the basis of dynamic reactive optimization which is important way to improve the security and economically. With the development of power grid, more precision short-term load forecasting and more effective dynamic reactive optimization is in active demand. The two problems are studied in the paper.The Locally Linear Embedding (LLE) algorithm and Support Vector Machine (SVM) is introduced in the paper and the load character is analyzed firstly and then a model for short-term load forecasting is represented. The LLE algorithm miming data and reduce the dimension of load sample. Lastly the SVM is used for regression of load which has the merits of non-linear fitting and generalization. Take 1-day load of some area in ZheJiang as the sample, the effective and proper of the method is proved.Dynamic reactive optimization is more complicated. The paper studies the static reactive optimization firstly and proposes a dynamic reactive optimization based on Improved Genetic Algorithm (IGA). Use IGA to obtain the discrete control equipment value in each time-interval and the D-value of adjacent time-interval through static reactive power optimization. There is correlation between different control equipments in the practical operational power system, by using this correlation and load coefficient, combining the D-value of adjacent time-interval to renew the dispatch schedule. All of this makes the one day's reactive power optimization problem in whole. The simulation result shows the new method not only is simple in calculation, but also can get the control equipment one day's dispatch schedule effectively.