应用反向传播(BP)神经网络模型综合评价游泳场所水质
Water Quality Evaluation of Swimming Pools Based on Back Propagation Neural Network
-
摘要:
目的 应用反向传播(back-propagation,BP)神经网络模型,构建游泳场所水质综合评价模型,为加强游泳场所水质的卫生管理和保障游泳者的身体健康提供相关依据。 方法 运用专家评价的游泳场所水质分级标准并根据随机数产生样本,使用BP 神经网络进行训练与建模。将建立好的BP 神经网络模型用于上海市长宁区游泳场所快速水质等级判断。 结果 构建的基于BP 神经网络的游泳场所水质综合评价模型对训练数据的预测准确率达到95.2%。2009 年上海市长宁区游泳场所水样中,无重度污染水样。水质一般的水样所占比重最大,为54.44%。小区会所内设游泳场所水样中,轻度污染水样所占比重高于体育系统和星级宾馆内设的游泳场所。体育系统内设游泳场所水样中,水质优良水样所占比重高于星级宾馆和小区会所内设游泳场所的比重。 结论 构建的基于BP 神经网络的游泳场所水质综合评价模型的预测值相对误差较小,精度较高。2009 年上海市长宁区游泳场所水质一般,小区会所内设游泳场所水质较差,体育系统内设游泳场所水质较好。 Abstract:Objective To build a water quality evaluation model based on back propagation (BP) neural network for better water hygiene management of the swimming pools in Shanghai. Methods Based on the experts' proposed standards of water quality grading for swimming pools, random samples were selected. A BP neural network was adopted for training and modeling. The established model was then applied to rapid evaluation of water quality grade of swimming pools in Changning District, Shanghai. Results With the help of BP neural network, the prediction accuracy of the water quality evaluation model was up to 95.2% for the training data. Among the water samples from swimming pools in Changning District in 2009, no severe water pollution was found and more than half (54.44%) of the water samples were at general grade. There were more lightly polluted samples from the community clubs than from the sports institutes and the star-rated hotels. The quality of samples taken from sports institutes was higher than those from the star-rated hotels and the community clubs. Conclusion The accuracy of the BP neural network model is high for water quality evaluation. Except for the water sampled from the community clubs' swimming pools, the quality of water in swimming pools of Shanghai Changning District is at a general acceptable level.
下载:
