Research On Optimal Dispatch Of Microgrid With Demand Response Based On Prediction Information Of Wind Power

The development of renewable energy can reduce the dependence on fossil fuels and reduce environmental pollution effectively.Wind power generation technology is the most mature renewable energy generation technology and the largest scale of development,but its low predictability and low schedulability has been becoming the bottleneck of large-scale wind power development and utilization gradually.Microgrid can promote the large-scale access of wind power and is an important technical to solve the local consumption of wind power.Accurate and reliable wind power prediction,reasonable optimal dispatching strategy and price strategy are the key to ensure the economic and reliable operation of microgrid.In this background,microgrid economic scheduling model based on deterministic wind power prediction information and robust economic scheduling model considering the uncertainty of source-load are build in this thesis from the perspective of improving wind power prediction performance,and a dynamic optimization TOU strategy is proposed.And then the effect of the uncertainty of source-load and TOU demand response on the microgrid operation optimization is studied.The main work of this thesis is as follows:Firstly,a combination model of wind power interval prediction based on different kernel machine learning methods is proposed.The point prediction model of wind power is established by kernel extreme learning machine and particle swarm optimization least square support vector machine,and the mixed kernel density estimation is used to estimate the relative error of point prediction and construct the prediction interval.At the same time,the entropy weight method is used to determine the weighted coefficients of the two different kernel machine learning methods,and the prediction intervals of are weighted.The simulation results show that the mixed kernel density estimation has a high error goodness of fit.The interval prediction combination method based on entropy weight method can take into account the characteristics of different kernel machine learning methods effectively,and improve the comprehensive performance of interval prediction.It can provide accurate and reliable prediction information for the optimization scheduling research of microgrid including wind power.Secondly,the optimal scheduling strategy of microgrid based on deterministic wind power prediction information is studied.An economic dispatching model of microgrid with the minimum day-ahead operating cost objective function is established in the grid-connected microgrid including wind turbine,gas generator,energy storage unit,flexible load and conventional load.On this basis,the uncertainty set of wind power and load power prediction are constructed based on the interval prediction,and the robust tunable parameter is introduced to establish the robust economic dispatching model of microgrid considering the uncertainty of source-load,which is solved by the column constraint generation algorithm.The deterministic economic scheduling model and the robust economic scheduling model with uncertainty are compared and analyzed through simulation experiments.It is verified that the robust economic scheduling strategy improves the ability to resist the risk of intraday real-time electricity price fluctuation and enhances the robustness of microgrid at the cost of daily operation cost.By changing the robust tunable parameters,the economy and conservatism of robust economic scheduling can be adjusted flexibly.Finally,an optimal period partitioning algorithm based on a moving boundary technique is proposed,and the dynamic optimization time-of-use price model is established based on the price elasticity theory of power demand.The dynamic optimization time-of-use price strategy is introduced into the deterministic economic scheduling model and the robust economic scheduling model with uncertainty.Simulation results show that dynamic optimization time-of-use price strategy can divide periods and formulate price strategy more reasonable.The overall operating cost of microgrid is reduced,and the economic optimization of microgrid dispatching operation is realized by the dynamic optimization time-of-use price strategy.