Load Dispatching Optimization Based On Large-scale System Theory

Modeling and optimization of unit commitment is an important part of the power system dispatching management,plays a key role in reducing units operating costs and effective use of resources.Since the output of units represented by continuous variables and operating status of units represented by discrete variables coupling relationship,the issue has become a strong NP-hard mathematical problem.As the number of combinations increase exponentially with the number of units increasing,it is difficult to find the optimal solution of the problem in a fixed time,because the iterative time is long.Aiming at the above problems,this paper establishes the ARIMA model for the historical load data by using the time series theory and predicts the load range in one day.In the Java program,the R code is called by Rserve to simulate the forecasting method.The autocorrelation function and the partial autocorrelation function of the load are analyzed,and the forecasting model of the maximum and minimum load are obtained by combining the model selection criteria.The predicted five sets of data passed the white noise of the residual test.According to the load range forecasted,calculate the number of units that need to be startup at most and at least.All combinations of units that do not meet the requirements are efficiently deleted,which greatly reduces the selection range of the unit portfolio.The simulation results show that the proposed algorithm can reduce the residual number of units by less than 10% under the precondition of guaranteeing precision,and achieves good results.A large-scale system decomposition and coordination method is used to decompose the load distribution problem into three-layer structure.The middle layer and the bottom layer coordinate their inputs and outputs by Lagrange multipliers,and the upper layer iterates the combinations of units that meet the conditions,and finally obtains the optimal unit combination and the output power of each unit.Through the simulation analysis and comparison,the calculation speed is fast and the results are consistent.In this paper,after analyzed the causes of "dimension disaster" and "dual gap" deeply,combining time series with large-scale system theory,a new solution is proposed to make the unit commitment problem greatly simplified.And the feasibility of the algorithm is validated by simulation analysis.