Research On Reactive Power Optimization With Distribution Network Contained Wind Power

Wind power generation ,one of the renewable energy generation technologies, is becoming one of the most commercial development distributed generation technology ,because of its wide distribution and no pollution. But because of its unstable power output, it will affect reactive power of the power system. While reactive power optimization is the important means to improve the economic, safety and power quality of running grid. The research of more effective reactive power optimization with distribution network contained wind farms becomes an urgent problem needed to solve.Firstly, this paper introduces the principle of wind power and makes comprehensive analysis to the doubly fed induction generator (hereinafter referred to as DFIG) wind which is the most potential in the variable-speed constant-frequency wind turbine. The steady state model of doubly-fed induction wind generator is established and its power characteristics is also studied, and its ability of adjusting reactive power is applied for the participation in the dynamic reactive power optimization of distribution network as a continuous power source.Furthermore, this paper makes the forecast to the load and output of wind farm, since the main uncertainties which affected reactive power optimization in the distribution network contained wind farms are the load and output power of wind farm. Least squares support vector machine (hereinafter referred to as LS-SVM) method has advantage in simple structure, global optimal and generalization ability and is applied for pattern recognition, regression problem and other fields. This paper uses the LS-SVM method combining cluster analysis to forecast the load and output power of wind farms. The samples are analyzed using Fuzzy C-means method, and the forecasting samples with similar characteristics are selected as training samples to make forecasting model for load and output power of wind farms.Finally, this paper establishes the dynamic reactive power optimization model with the objective function as active power loss minimum and uses adaptive particle swarm optimization to solve the model. Based on the static reactive power optimization with each time using particle swarm optimization, the control equipment values are obtained. The pre-action table is formatted through the different between the same control equipment at the adjacent time, and is adjusted dynamically. Then the dynamic reactive power optimization strategy is completely formed. The experiment results show that the method employed in this paper can obtain the effective control programs of the equipments in one day. In the meanwhile the DFIG win farms used for reactive power source can save the cost of additional reactive compensation installation, and the method can overcome the difficulties of traditional reactive power optimization.