基于智能计算的影响河流水质的分类预测.docVIP

------------------------------------------------------------------------------------------------ —————————————————————————————————————— 基于智能计算的影响河流水质的分类预测 摘 要：为了有效地控制和治理河流水体污染，保障水资源的可持续利用，了解水资源现状及 发展变化趋势，增加水质预测系统具有十分重要的现实意义。基于影响水体水质的因素众多以 及BP网络存在的局限性，建立主成分分析法、BP神经网络、遗传算法结合的水质分类预测模 型。本文首先采用主成分分析法进行降维，提高BP网络训练速度。采用遗传算法优化BP网络 的参数，优化出全局最优的网络参数，提高水质预测分析的精度。该模型利用太湖某断面的水 质指标实测数据进行训练和仿真。研究结果表明：23个指标因素可压缩为15个综合指标，未污 染水质的仿真个数为124，预测正确为100，准确率为80.65%；污染水质的仿真个数为44，预 测正确为41，准确率为93.18%。总体预测率约为83%。 关键词：主成分分析；BP神经网络；遗传算法；水质预测 中图分号类： 文献标识码：A Based on Intelligent Computing the Classification of River Water Quality Forecasting Abstract: In order to effectively control and river water pollution control, protection and sustainable utilization of water resources, understanding of situation and development trend of water resources, increasing water quality forecast system has great practical significance. Based on the factors that affect the water quality of many existing limitations and BP network, the establishment of principal component analysis, BP neural network, genetic algorithm classification of water quality prediction model. Firstly, using principal component analysis to reduce the dimensionality and improve the training speed of BP network. BP network using genetic algorithm to optimize the parameters to optimize the global optimum of the network parameters to improve the accuracy of prediction of water quality. In this model, the water quality of Taihu Lake in a section of the measured data for training and simulation. The results show that: compressibility factor of 23 to 15 indicators of composite indicator, the number of uncontaminated water quality simulation 124, 100 prediction is correct, accuracy was 80.65%; pollution, water quality simulation number 44, correctly predicted 41 , accuracy was 93.18%. Overall prediction rate of about 83%. Keywords: Principal components analysis; BP neural network; Genetic a