Research On Scheduling Optimization Of Wind Power Integrated System

In the new century, China’s wind power industry has reached high-speed development. However, the randomness, volatility and the anti-peaking characteristics of wind power bring forward new challenges to the power system analysis, scheduling and control. To solve this problem, power system dispatching management based on wind power prediction technique is the most economic and effective way. The major work of this paper includes:(1)Proposes an advanced BP neural network model to predict wind power which selects the training data set based on meteorological data similarity. Considering the shortcomings of the existing BP neural network model in selecting the training data set, the proposed model chooses the training data set according to meteorological data similarity respectively for each forecast period. Prediction outcome shows that, the advanced BP neural network model is superior to traditional BP neural network model.(2)Proposes an advanced multi-objective particle swarm optimization algorithm to solve wind power system scheduling optimization problem, the objective functions include the system total generation cost, total pollution gas emissions and operational risks caused by short-term fluctuation of wind power. When dealing with unit commitment problem, the algorithm firstly gets the initial unit commitment by the designed heuristic rules, which can guarantee the diversity and rationality of the initial unit commitment. And then the algorithm introduces genetic operators to improve the capacity of multi-objective particle swarm optimization searching unit commitment. When dealing with load distribution problem, the algorithm improves the capacity of multi-objective particle swarm optimization optimizing load distribution through the designed the external file maintenance strategy and the global optimum location selecting strategy. Finally, using fuzzy decision-making to find out the most appropriate scheduling scheme from the non-inferior solutions resulting from the algorithm. The Weight of the three objects can be defined by decision-maker based on his preference, the load level and wind power prediction condition. Simulation results show that wind power can significantly reduce total system cost and pollution gas emissions; meanwhile, using heuristic rules and genetic operators searching for unit commitment is superior to priority list method, the algorithm has some practical value.