改进PSO算法在多水源供水系统优化调度中的应用

Application of Modified Particle Swarm Optimization Algorithm to Optimal Scheduling of Multi-source Water Supply Systems

针对大城市多水源供水系统的特征,以测压点压力的BP神经网络模型代替管网水力平衡方程组,建立了多水源供水系统的一级优化调度模型。采用混合罚函数法处理约束条件,将约束优化问题转化为无约束问题,然后采用粒子群算法(PSO)对问题进行求解。为避免算法陷入局部最优,提高算法精度,将混沌搜索引入到标准粒子群算法中形成了混沌粒子群算法,能够充分发挥二者的优势。最后,将上述模型应用于天津市供水系统,优化后的供水方案每日可使泵站电耗降低4 087 kW·h,年节省电量约149.2×104kW·h,总费用减少2.13%,经济效益明显,证明了模型的适用性和算法的可行性。

According to the features of multi-source water supply systems in big cities, the hydrau- lic balance equations were replaced by a BP neural network model of nodal pressures, and a primary optimal scheduling model of the multi-source water supply system was built. The constrained optimization problem was changed to an unconstrained problem after the mixed penalty method was used to treat con- straint conditions, and the particle swarm optimization （PSO） algorithm was used to solve the problem. To avoid the algorithm getting into a local optimum and improve the precision of the algorithm, the chaos search was brought into the standard PSO algorithm to form the chaotic PSO algorithm. The model was applied to the Tianjin water supply system. The optimized water supply scheme could reduce power con- sumption of pumping stations by 4 087 kW - h daily and 1. 492 million kW - h annually, and the total cost was reduced by 2.13%. This clear economic benefit indicated the applicability of the model and the feasibility of the algorithm.