An Immune Colonal Hybrid Algorithm For Solving Stochastic Chance-constrained Programming And Its Application

With the multiple applications of stochastic optimization theory, the diversification of the stochastic programming model and the complexity of the model solution, the stochastic theory and its applications are in need of a method to solve the model. In this paper a hybrid algorithm for solving Stochastic Chance-Constrained Programming Model (SCCP) is proposed and applied to the solution of reservoir optimal scheduling strategy. Details are as follows:Firstly, a modified Immune Clonal Selection Algorithm (ICSA) is proposed. The dual-cloning strategy is proposed. The introduction of the concentration adjustment mechanism suppresses the similar antibodies to ensure the diversity of the population and the use of affinity vector mechanism to ensure the performance of the population. Two-mutation strategy is also proposed. The introduction of the estimation of distribution algorithm mutation operator improves the ability of global optimization and the use of Gauss mutation operator improves the ability of local optimization. The benchmark function results show that compared with existing algorithms, convergence speed and accuracy has been greatly improved.Secondly, a hybrid algorithm based on ICSA, Monte Carlo algorithm and neural network is proposed in this paper. Monte Carlo algorithm is used to produce training random variables sample matrix for Back Propagation neural network (BPNN) to approximate the stochastic function. Solution feasible is checked via Monte Carlo algorithm and fitness value is calculated via the trained BPNN in ICSA until it can get the solution. The simulation results show that satisfactory result has achieved before100generation, moreover the precision in the single objective optimization problem is improved by1.3%and in multiobjective optimization problems is increased by65%compared with the other existing algorithm.Finally, because SCCP can more accurately describe the randomness of runoff and risk of the implementation of strategies, SCCP of reservoir operation in accordance with the reservoir environment characteristics and runoff characteristics of Dahekou and Xishanwan cascade reservoirs in Inner Mongolia is proposed and solved by the intelligent optimal algorithm in this paper. The simulation results show the correctness and effectiveness of the model and the algorithm. So it can provide an effective and feasible solution to solve complex reservoir long-term scheduling problem.