基于粒子群算法的神经网络在水资源评价中的应用

Assessment on the Sustainable Utilization Level of Water Resources by Artificial Neural Network Based on Particle Swarm

为了改善传统的人工神经网络,在训练过程中容易陷入局部最小导致应用于水资源评价时存在对训练样本的拟合精度不高的缺点,采用粒子群算法优化人工神经网络的权值和阈值,然后将其应用于中国12个地区的水资源可持续利用系统评价实例中,并和传统的人工神经网络进行了对照。结果表明,基于粒子群算法的人工神经网络和传统的人工神经网络相比,能较好的提高对训练样本的拟合精度,表明基于粒子群算法的人工神经网络,用于水资源可持续利用系统评价是可行的。

In order to overcome the drawback of Artificial Neural Network based on back propagation algorithm （BP- ANN）, that is, easy entrapment in a local minimum, artificial neural network based on particle swarm optimization （PSO-ANN） was applied to assessing the sustainable utilization level of water resources for 12 areas in China, and its＇ training result was compared with that of BP-ANN. The results show that the PSO-ANN could simulate the training data better than BP-ANN. It is shown that the PSO-ANN was an alternative method for the assessment of water resources sustainable utilization level.