引用本文:肖金球, 周翔, 潘杨, 冯威, 陈多观.GA-BP优化TS模糊神经网络水质监测与评价系统预测模型的应用——以太湖为例[J].西南大学学报(自然科学版),2019,41(12):110~119
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 93次   下载 42 本文二维码信息
码上扫一扫!
分享到: 微信 更多
GA-BP优化TS模糊神经网络水质监测与评价系统预测模型的应用——以太湖为例
肖金球, 周翔, 潘杨, 冯威, 陈多观1,2,3,4
1. 苏州科技大学 电子与信息工程学院, 江苏 苏州 215009;2. 江苏省环境科学与工程重点实验室, 江苏 苏州 215009;3. 江苏省工程技术中心, 江苏 苏州 215009;4. 苏州科技大学 环境生物科技研究所, 江苏 苏州 215009
摘要:
针对水质监测与评价系统在太湖应用过程中水质数据和水质等级评价不准确的问题,建立了一种多隐含层改进型GA-BP神经网络来辨识复杂的水质模型,以均方误差MSE作为个体适应度,并在权值调整过程中加入动量因子来加快收敛速度,获取最优权阈值,提高其拟合程度和泛化能力.根据校准后水质的pH、溶解氧、浊度和氨氮数据,利用TS模糊神经网络建立了适用于当地水质评价的模型.仿真测试结果充分说明改进型GA-BP优化TS模糊神经网络对复杂水质模型的拟合程度更高,水质数据的均方误差、绝对误差更小,绝对误差保持在1.5%以内,水质等级预测精度提高14.28%.
关键词:  水质评价  多参数  遗传算法  BP神经网络  TS模糊神经网络
DOI:10.13718/j.cnki.xdzk.2019.12.015
分类号:TP391
基金项目:国家科技部重大水专项(2017ZX07205003);江苏省产学研前瞻性联合基金项目(BY2011132);江苏省研究生创新与教改项目(091580001);苏州科技大学研究生创新工程基金项目(SKCX17_025).
Application of a GA-BP Optimized TSFNN Water Quality Monitoring and Evaluation System Prediction Model ——A Case Study of Taihu Lake
XIAO Jin-qiu, ZHOU Xiang, PAN Yang, FENG Wei, CHEN Duo-guan1,2,3,4
1. Department of Electronic and Engineering, Suzhou University of Science and Technology, Suzhou Jiangsu 215009, China;2. Jiangsu Key Laboratory of Environment Science and Engineering, Suzhou Jiangsu 215009, China;3. Jiangsu Engineering Technology Center., Suzhou Jiangsu 215009, China;4. Environment Biotechnology Research Institute, Suzhou University of Science and Technology, Suzhou Jiangsu 215009, China
Abstract:
The present Water Quality Monitoring and Evaluation System which is applied in Taihu Lake has the disadvantage of inaccurate water quality data and water quality rating. To solve this problem, a multi-hidden layer improved GA-BP neural network is proposed in this paper for identifying the complex water quality model. MSE (mean square error) is used as the individual fitness, and a momentum factor is added to the weight adjustment process to speed up the convergence for obtaining the optimal weight and threshold and improving its goodness of fit and generalization ability. Based on the calibrated data of pH, dissolved oxygen, turbidity and ammonia nitrogen of the water, and using the TS fuzzy neural network, a model suitable for evaluating the local water quality is established. The results of a simulation test show that the improved GA-BP optimized TS fuzzy neural network has a higher degree of fitting for the complex water quality model, and the mean square error and the absolute error of the water quality data are smaller. In addition, the absolute error percentage is kept within 1.5%, and the water quality level prediction accuracy is improved by 14.28%.
Key words:  water quality evaluation  multi-parameter  genetic algorithm  BP neural network  TS fuzzy neural network
手机扫一扫看