| With the gradual advance of the energy revolution,the amount of wind power access in the power grid continues to grow.While improving the environmental friendliness of the power system,it also further enhances the randomness of the power supply side.At the same time,the access of multiple types of new loads and the advancement of power market reforms have also injected significant uncertainties into the power system on the demand side and the market side.The widespread existence of multiple uncertain factors in the power system has become an inevitable trend,which has brought great challenges to the safe and economic operation of the power system.Therefore,it is of great significance to analyze the correlation characteristics between different uncertainty factors,establish the joint probability distribution model of uncertainty,and realize its application in different scenario scheduling models,so as to promote the elimination of new energy and improve the level of safe economic operation of power system.In view of this,this dissertation takes the uncertainty model as the main line,takes different subjects as scenarios,and conducts meaningful research on the uncertainty modeling method of power system and its application in multi scenario optimizationIn order to solve the problem of dynamic probability distribution modeling of uncertainty in power system,a method of power system uncertainty modeling considering dynamic correlation and heteroscedasticity effect is proposed.First,ARIMA model is introduced as the dynamic conditional mean model,GARCH model as the dynamic conditional variance model,and the dynamic marginal distribution model of prediction error is established to characterize the time-varying characteristics of prediction error marginal distribution.After that,the fluctuation of the predicted value is introduced as an auxiliary variable.Based on its high correlation level with the prediction error,the dynamic joint distribution model of the two is constructed based on the dynamic Copula theory,and the parameter estimation and evaluation method of the model is proposed by combining IFM and AIC method.Finally,based on the model,a conditional probability distribution model for forecasting errors considering heteroscedasticity effects is established.The simulation results show that,compared with the traditional static probability distribution modeling method,the proposed model gives a more accurate uncertainty confidence interval at a lower redundancy level.In order to cope with the uncertainty of source and load in the active distribution network and achieve the equilibrium of multi-stakeholders,an active distribution network dynamic game optimization operation method considering the dynamic correlation of source-charge prediction errors is proposed.First,based on the proposed forecast error conditional probability distribution model,the two-sided uncertainty models of source and load are constructed respectively,and the dynamic copula theory is introduced to construct the dynamic joint distribution of the two.After that,in order to deal with the equilibrium of multi-stakeholders of active distribution network,the dynamic game model and the response behavior model of each agent are constructed considering the uncertainty of source and load.Furthermore,a Latin hypercube sampling method considering the dynamic correlation of source and load is proposed,and the stochastic chance constraints involved in the dynamic game model of active distribution network are dealt with by this method.Finally,combining the proposed sampling method with particle swarm optimization,a dynamic game particle swarm optimization algorithm is proposed,and the Nash equilibrium of the problem is solved by the algorithm.The simulation results show that the joint distribution model proposed in this dissertation has higher accuracy than the traditional method.The game optimization operation mode proposed in this dissertation can fully mobilize the production enthusiasm of different stakeholders,reduce the reduction of overall benefits caused by single objective optimization,and improve the wind power consumption and overall benefits of the power system.In order to solve the problem of optimizing the scheduling of regional power grids with multi-wind farms,a random optimization scheduling strategy of regional power grids that consider the dynamic correlation of multi-wind farms is proposed.First,based on the proposed conditional probability distribution model and the dynamic Copula theory,a multi-dimensional dynamic joint distribution of source load in transmission network is established.Afterwards,in order to deal with the ‘dimension disaster’ problem caused by the increase of the dimensions of multivariate correlation random variables,a linear operation method of multivariate dynamic correlation random variables based on discrete convolution and recursion process was proposed to improve the solution efficiency.Finally,based on stochastic chance constraint programming,a regional power grid stochastic optimization scheduling model considering multivariate dynamic correlation randomness is established,and the proposed linear operation method is applied to determine the equivalent class conversion of stochastic chance constraint,and the model is solved efficiently.The simulation results show that the multivariate dynamic correlation random variable linear operation method proposed in this dissertation can reduce the complexity of solving the model while ensuring the accuracy of the model;compared with the traditional static marginal distribution and static correlation modeling methods,the dynamic model proposed in this dissertation can improve the economic and safe operation level of the system.In order to cope with the self-scheduling problem of renewable energy power generation companies under the influence of uncertainties,a day-to-day optimization operation strategy of renewable energy power generation companies considering the source-price dynamic correlation is proposed.Then,for the nonlinear operation problem of multivariate uncertainty involved in the objective function of the model the nonlinear operation method of multivariate dynamic correlation random variables is derived,and the operation results are represented in the form of high-dimensional discrete matrix.Finally,the original problem is transformed into MILP by combining the method with Va R theory.The results of practical examples show that,compared with the traditional static joint distribution model,the joint distribution model proposed in this dissertation can describe the probability distribution characteristics of risk and return of new energy power generation companies more accurately,and the operation strategy of new energy power generation companies can also help new energy power generation companies to carry out multi-stage market arbitrage and realize the improvement of return. |
