Study On Clustering Forecasting And Robust Dispatching Method For Wind Power Uncertainty

Developing and utilizing renewable energy is an important measure to promote the transition of energy structure and realize the sustainable development of energy industry in China.The wind power generation technology is the most mature in the renewable energy generation,and it has the largest development scale.But its low predictability and low schedulability has become the bottleneck of large-scale wind power development and utilization.In this paper,two aspects of solving the problem,increasing the accuracy of wind power forecasting and improving the power system scheduling method,are studied.The main work is as follows.In order to improve the accuracy of wind power forecasting,a neural network model based on continuous time clustering and wavelet packet decomposition methods is established.By means of wind speed similar day clustering and continuous time clustering,the whole year is divided into several continuous time segments with stronger regularity of wind speed.Then,the wind power data is decomposed based on wavelet packet method to obtain more regular wind speed subsequences with different frequency.Finally,model each subsequence with radial basis function neural network method,and the forecasted wind power is obtained after wind speed-power curve transformation.Clustering and wavelet packet decomposition methods increase the regularity of the wind speed subsequences,so the accuracy of wind power forecasting is improved.Simulation results show the effectiveness of the proposed method.In view of the fact that the forecasting error of wind power can only be reduced instead of being eliminated,the modeling method of wind power output uncertainty is put forward.It uses the risk cost to meaure the loss of wind curtailment and load shedding,and define the composite cost as the sum of risk cost and operation cost.Then it presents a method for the optimization of uncertain set's scale and uncertain set's boundary at different periods in order to achieve the optimal performance of the scheduling plan's economy and robustness.In order to improve the scheduling method of power system,a robust scheduling model with multi-type power sources is established.It proposes a two-stage(start/stop scheduling and output scheduling)algorithm for robust scheduling model.Start/stop scheduling based on short term wind power forecasting data to formulate a robust unit commitment,which is the centralized embodiment of the robustness.The methods of active start/stop variables identification and active constraints in extreme scenarios identification are used to promote the computational efficiency.More accurate unit fuel cost function and ultra short term wind power forecasting data are used to reduce the composite cost in output scheduling.The output scheduling model is a nonlinear programming problem without integer variables,so the primal dual interior point method is used to solve the problem.Considering two facts that the microgrid is an important way for the accommodation of renewable energy,and microgrid's wind power installed capacity ratio,equipment types and energy supply forms are different with large power network,it applies the robust scheduling method to the microgrid.The results show that the robust scheduling method has the advantages of both risk cost and composite cost.Also,it analyzes the influence of three kinds of factors on the robust scheduling method's economic benefit.At last,it proves that the reasonable scheduling of electric vehicles can help microgrid accommodate more wind power with less cost.