基于人工智能对地表水的水质预测与评价研究

发布时间：2018-04-21 22:29

本文选题：动态可修正灰色预测 + 动态时变指数平滑预测　；参考：《东北电力大学》2017年硕士论文

【摘要】：现阶段对水质预测与评价的研究已有一定成果,具有代表性的单项预测与评价模型被广泛应用于实际。但是各种单项预测模型分别从不同角度对样本水质进行分析,会造成某些重要信息的遗缺。此外,水环境系统是一个动态的复杂系统,其相关水质参数一直处于动态变化之中。当前的水质预测与评价无法准确的反映水质的总体情况,故深入研究水质预测与评价是十分必要与迫切的。针对所研究背景环境的复杂性,为适应水质变化的动态特性,提高预测精度,结合人工智能算法在水质智能化建模方面的较好应用,于是提出动态可修正灰色预测模型与动态时变指数平滑预测模型作为研究水质的单项预测模型。并将这两种预测模型进行组合,建立基于单项预测模型预测有效度的组合预测模型。该组合模型可以充分利用各单项模型的优势,通过一个适当的权重进行组合,以单项模型的动态更新来适应水质动态变化的特点。为验证所建立模型的有效性,以吉林省某河段真实监测的水质数据为基础,对溶解氧、高锰酸盐指数、氨氮、总磷、总氮五项水质参数进行水质预测。实验结果表明,该组合模型与单项预测模型相比,其预测效果更为理想,样本水质的发展态势与模型预测结果曲线的拟合性更好,在水质预测方面具有较好的实用价值。在以上工作的基础上,应用支持向量机对相应的水质进行水质评价。介绍了支持向量机由二分类构建多分类的方法,以及使用粒子群算法对支持向量机的相关参数进行寻优。实验结果表明,基于支持向量机的多分类应用在水质评价方面具有较高的分类精度。评价结果准确、可靠,符合客观实际。
[Abstract]:At present, the research on water quality prediction and evaluation has made some achievements, and the representative single prediction and evaluation model has been widely used in practice. However, a variety of single prediction models are used to analyze the sample water quality from different angles, which will lead to the absence of some important information. In addition, the water environment system is a dynamic complex system, and its water quality parameters are always changing dynamically. The current water quality prediction and evaluation can not accurately reflect the overall situation of water quality, so it is very necessary and urgent to study water quality prediction and evaluation in depth. In view of the complexity of the background environment, in order to adapt to the dynamic characteristics of water quality change and improve the prediction accuracy, combining with artificial intelligence algorithm in water quality intelligent modeling, The dynamic modifiable grey prediction model and the dynamic time-varying exponential smoothing prediction model are proposed as the single prediction models for the study of water quality. The two forecasting models are combined to establish a combined prediction model based on the prediction validity of single prediction model. The combined model can make full use of the advantages of each single model and can be combined with a proper weight to adapt to the characteristics of the dynamic change of water quality by the dynamic updating of the single model. In order to verify the validity of the established model, the water quality parameters of dissolved oxygen, permanganate index, ammonia nitrogen, total phosphorus and total nitrogen were forecasted on the basis of the water quality data of a river reach in Jilin Province. The experimental results show that the combined model is more effective than the single prediction model, and the development trend of the sample water quality is better than the curve of the model prediction results, and it has better practical value in water quality prediction. Based on the above work, support vector machine is used to evaluate the water quality. This paper introduces the method of constructing multi-classification by two-classification in support vector machine, and uses particle swarm optimization algorithm to optimize the related parameters of support vector machine. The experimental results show that the multi-classification based on support vector machine has higher classification accuracy in water quality assessment. The evaluation results are accurate, reliable and in line with the objective reality.
【学位授予单位】：东北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X824;TP18

【参考文献】