Multi-Stochastic Variable Programming Methods And Their Application In Economic Operation Optimization Of Microgrid

Stochastic parameters are prevalent in many practical optimization problems,and stochastic programming methods have been widely applied to such optimization problems with random parameters.For stochastic programming problem with multiple random parameters,multi-random variables not only increase the complexity of solving the problem,but the correlation of random variables also need to be considered,however,lots of stochastic programming methods rarely consider the correlation of random variables at present.In this dissertation,the programming problem with multiple random variables is studied and applied to the economic operation optimization of microgrid containing wind power generation and photovoltaic generation.On the basis of the basic stochastic programming method,the definition of multi-stochastic variable programming problem is given,and the basic processing methods of multi-stochastic variable programming are proposed.Firstly,in order to deal with uncorrelated random variables,the LHS method is respectively combined with Cholesky decomposition and Gram-Schmidt orthogonalization based on the advantage of sampling uniformity of LHS method,and the CDLHS method and GSLHS method are proposed to establish the sample model of uncorrelated random variables,and the two methods are compared through simulation.Secondly,in order to deal with correlated random variables,the Nataf transformation is combined with the LHS method,and the NTLHS method is proposed to establish the sample model of correlated random variables.Finally,the performance of GA is improved from the aspects of initial population generation,stochastic constraints handling,evaluation function and evolutionary strategy,and the evaluation function based on reliability and evolutionary strategy based on heuristic idea are proposed,and the improved GA is combined with CDLHS method or NTLHS method to solve multi-stochastic variable programming problems.The results of the simulation indicate that the proposed multi-stochastic variable programming methods are effective.The economic operation optimization problem of microgrid containing wind power generation and photovoltaic generation is studied.Firstly,considering the correlations and randomness of wind speeds and solar radiation,the probability distribution models based on kernel density estimation method and correlated sample models based on NTLHS method of wind speeds and solar radiation are respectively presented,and the results of the simulation indicate that the proposed models are effective.Secondly,considering the operation cost and pollution emission cost of microgrid,a dynamic stochastic chance-constraint programming model(DSCCPM)of the economic operation optimization of microgrid containing wind power generation and photovoltaic generation is established.Finally,the proposed multi-stochastic variable programming methods are applied to solve the economic operation optimization model of microgrid,and the solving procedure of the continuous operation optimization of microgrid is given.For the typical microgrid system containing wind power generation and photovoltaic generation,in this dissertation,the simulation results are obtained under different sample sizes,different reliability levels,different correlations and continuous operation of microgrid.And the effects of the different sample sizes,different reliability levels and different correlations on economic operation optimization of microgrid are respectively analyzed and discussed,and the results of the simulation prove that the established DSCCPM of microgrid is reasonable and the proposed multi-stochastic variable programming methods are effective.