摘要
针对传统的BP神经网络未对影响因子进行筛选和筛选过程的主观性等问题,提出了改进的BP神经网络模型。选取山西省阳泉市地下水饮用水水源地的10个水质监测指标,进行Pearson相关分析得到相关系数;运用信息指标评价法对模拟因子进行筛选,得到最优的模拟因子,在明确BP神经网络的结构后,把最优模拟因子作为BP神经网络的输入样本,被模拟因子(水质状况综合指数)作为输出样本,建立水质预测模型。结果表明,预测的水质状况综合指数与实际值平均相对误差为3.80%,水质指数平均相对误差为0,较传统的BP神经网络模型精度高。
In order to solve the problems of the screening and selection of influencing factors in conventional BP neural network, an improved BP neural network model is proposed. Firstly, ten water quality monitoring indicators of groundwater drinking water sources in Yangquan, Shanxi are selected and the correlation coefficient is obtained by Pearson correlation analysis. Then the information index evaluation method is used to filter the simulation factors and the optimal simulation factors are obtained. Finally, the water quality prediction model is established by taking the optimal simulation factors as the input of BP neural network after determining the structure of BP neural network and the simulated factor （ comprehensive index of water quality） as output sample. The simulation results show that the average relative error of predicted water quality comprehensive index is 3.80% to actual values, and the average relative error of water quality index is zero. The forecasting accuracy is higher than that of traditional BP neural network model.
出处
《水力发电》
北大核心
2017年第10期1-4,共4页
Water Power
基金
山西省自然科学基金资助项目(2015021169)
关键词
水质安全预测
Pearson相关分析
信息指标评价法
BP神经网络
water quality safety forecasting
Pearson correlation analysis
information index evaluation method
BP neural network
