论文部分内容阅读
水质预测是实现非线性水系统的柔性管理、防治水污染的前提工作.机理性水质预测模型的构建往往较复杂并且需要大量运算与数据,预测效果有时不够精确,其进一步推广应用也受到限制.文中以淮河复杂水环境非机理性水质预测为目的,构建改进的量子遗传算法优化BP神经网络模型,采用动态改进策略和灾变策略作为进化操作准则来优化BP模型的权值和阈值,用历史观测数据作为学习范例训练模型.对比实验结果发现,模型改进以后,进化代数、收敛速度和预测结果的准确率有较大提高.该模型用于水质预测的黑箱问题是可行的,拓展水环境管理的思路.
Prediction of water quality is a precondition for the flexible management of non-linear water systems and prevention and control of water pollution.The construction of a mechanism-based water quality prediction model is often complicated and requires a large amount of computation and data, the prediction effect is sometimes not accurate enough, and its further application is also limited. In order to improve the non-structural water quality of complex water environment in Huaihe River, BP neural network model based on improved quantum genetic algorithm (GA) optimization is constructed. Dynamic improvement strategy and catastrophe strategy are used as evolutionary operation criteria to optimize BP model weights and thresholds. Data as a training example.Experimental results show that after the model is improved, the evolutionary algebra, the convergence rate and the accuracy of the prediction results are greatly improved.It is feasible to use this model to solve the black box problem of water quality prediction, and to expand the water environment management Ideas.