Improved Differential Evolution Algorithms And Their Application In Water Supply Scheduling Optimization

Global optimization problems are widely found in processes of scientificresearch and engineering application. Therefore, it is or great significance andpractical value to improve and apply the global optimization methods. Differentialevolution algorithm is a kind of evolutionary algorithm based on simulating thenatural evolution and the survival of the fittest, which can be used to effectivelysolve the problem of water scheduling optimization.This thesis, focusing on the optimal scheduling of water supply system ofHuangshan City and employing the improved differential evolution algorithms,raises corresponding improvement strategies on account of two state-of-the-artvariants, applies the improved algorithms to the optimal scheduling of water supplysystem of the city and significantly cuts the cost for the enterprise.The crossover rate in differential evolution exerts great influence to thealgorithm. This thesis takes in consideration that JADE algorithm has an implicitbias towards small values during self-adaptation process, and proposes thecrossover rate improvement method by basing on power mean operation so as tobalance this bias. In this way, the crossover rate is assigned with greater value asfar as possible while maintaining its adaptive capacity, and the convergence ofJADE is speeded.The mutation operator in differential evolution algorithm has many kinds ofmutation strategies that show different performance characteristics with each otherand are suitable to solve different problems. The existing adaptive method for themutation strategy is mainly coarse-grained adaptation, which is the choice betweendifferent mutation strategies. Based on the analysis of control parameter p inSaJADE, this thesis achieves a fine-grained adaptation for the mutation strategythrough the transformation of p. What’s more, the author to this thesis finds that theimproved algorithm in the experiments demonstrates higher reliability.According the actual problems in the water supply system of Huangshan City,A two-level hierarchical macroscopic model has been established by basing on thehistorical operational data of the water supply system and the status quo of the No.2Water Plant of the City. With the assistance of the model, the improveddifferential evolution algorithms are applied to solve the scheduling problems andsuccessfully save the water supplying cost.