| 引用本文: | 徐 睿,张晓斌,薛鹏松.基于改进的GRNN-Markov水质预测模型研究及应用[J].灌溉排水学报,2022,41(S1):104-110. |
| XU Rui,ZHANG Xiaobin,XUE Pengsong.基于改进的GRNN-Markov水质预测模型研究及应用[J].灌溉排水学报,2022,41(S1):104-110. |
|
| 摘要: |
| 【目的】针对水环境系统的复杂性，结合GRNN网络预测模型和马尔科夫理论，构建了改进GRNN-Markov水质预测模型，为模拟不确定性、复杂多变的河流水质变化趋势提供可靠的方法依据。【方法】以汾河入黄口主要污染物CODcr为研究指标，采用 准则对水质监测资料进行预处理，利用灰色关联分析（GRA）确定输入节点，解决GRNN网络对关键参数不能自动寻优的缺点，同时采用改进GRNN模型对水质数据进行模拟预测，针对水质预测数据的随机波动性，通过Markov模型修正误差残值，以达到更好的预测结果，为水环境保护与治理提供新思路、新方法。【结果】改进GRNN-Markov水质预测模型，可以提高水质预测结果的精度，使相对误差从-38.27%~-15.71%提高到-25.77%~-5.25%，修正结果更加接近实测值。【结论】验证了组合模型在小样本水质预测中的可行性，为水环境管理提供了科学依据。 |
| 关键词: 灰色关联分析（GRA）；改进GRNN-Markov模型；汾河入黄口；水质预测 |
| DOI：10.13522/j.cnki.ggps.2022152 |
| 分类号: |
| 基金项目: |
|
| Research and Application on Improved GRNN-Markov Water Quality Prediction Model |
|
XU Rui, ZHANG Xiaobin, XUE Pengsong
|
|
1. Yuncheng Water Conservancy Survey and Design Research Institute Co., Ltd., Yuncheng 044000, China;
2. Yuncheng University, Department of Applied Chemistry, Yuncheng 044000, China;
3. Shaanxi Jinke Environmental Protection Engineering Co., Ltd., Xi’an 710119, China
|
| Abstract: |
| 【Objective】In view of the complexity of the water environment system, combined with the GRNN network prediction model and Markov theory, an improved GRNN-Markov water quality prediction model was constructed. This model, which provides a reliable method for simulating uncertain, complex and variable river water quality trends.【Method】 According to actual situation of water quality in Fen River’s estuary to Yellow River, taking the main pollutant CODcr as the research index, the water quality monitoring data were preprocessed using the Laida criteria. Using Grey Relational Analysis to determine input nodes of GRNN Network which solved the problem that the GRNN network unable to auto select and optimize input nodes. At the same time, the improved GRNN model was used to simulate and predict the water quality data. In view of the random fluctuation of the water quality prediction data, the Markov model was used to correct the error residual value to achieve better prediction results and provide new ideas and methods for water environmental protection and governance. 【Result】The combined model provides new ideas and methods for water environmental protection and governance. The research show that the improved GRNN-Markov water quality prediction model can improve the accuracy of water quality prediction results. Relative error was -38.27%~-15.71% based on GRA-GRNN model, Relative error was -25.77%~-5.25% based on improved GRNN-Markov water quality prediction model, the correction result was close to the measured value. 【Conclusion】 This combination model can be used in water quality prediction based on small sample data. This study provides a scientific basis for water environment management. |
| Key words: grey relational analysis (GRA); improved GRNN-Markov model; Fen River’s estuary to Yellow River; water quality prediction |
