MCPSO优化BP-NN在日用水量预测中的应用

The application of BP-NN optimized by MCPSO for forecasting of daily water consumption

论文使用MCPSO(多群合作粒子群算法)的全局随机优化能力修正了传统BP神经网络的收敛速度过慢,容易出现局部最小值的缺点,并结合长沙第八水厂的供水量数据,使用优化后的BP网络进行日用水量的预测,结果表明优化后的BP算法在减少迭代次数和预测准确性方面都有非常大的提升。

In the paper, the stochastic global optimization technique of the MCPSO （multi-swarm cooperative particle swarm optimizer） is used to amend the traditional BP-NN due to the disadvantage of slow congruence and local minimization, and then the paper make a forecasting of urban daily water consumption with the data from the eighth water-factory in Changsha, using the BP-NN optimized, the new method is proved to improve the precision of forecasting and reduce the iterative times very much.