基于机器学习算法的水质预测及相关算法比较研究

Water Quality Prediction Based on Machine Learning Algorithm and Comparative Study of Related Algorithms

近年来,水环境受到严重污染,各国开始运用各种技术来预测未来水质变化趋势。为了评估小流域水环境质量,提前采取行动减缓污染速度,实现改善水质的目的,本文分析了赤水河流域2019年每月水质等级时空分布情况,并建立了基于机器学习算法的水质预测模型。结果表明,从空间上来看,流域中下游污染比上游更加严重;从时间上来看,水质在夏天较差,在冬天相对较好;几种水质预测模型中,支持向量机效果最佳,平均精度达到0.9,这和支持向量机偏向于低样本量并可进行整体最佳拟合等特性有关,说明通过构建合适的水质预测模型可实现流域内水质等级的模拟,根据预测结果能够提前采取措施避免水质进一步恶化。

In recent years,water environment has been seriously polluted,and many countries have begun to use various technologies to predict future trends in water quality.In order to evaluate the water environment quality of small watersheds,take action in advance to slow down the rate of pollution,and achieve the goal of improving water quality,this paper analyzes the temporal and spatial distribution of water quality grades in Chishui River Basin in each month of 2019,and establishes a water quality prediction model based on machine learning algorithm.The results show that,spatially,pollution is more serious in the middle and lower reaches of the basin than in the upper reaches;temporally,water quality is worse in summer and relatively better in winter.Among several water quality prediction models,support vector machine performs the best with an average accuracy of 0.9,which is related to the characteristics of support vector machine that it is biased towards low sample sizes and can achieve the overall best fit.This indicates that simulating the water quality grade in the basin can be achieved by constructing a suitable water quality prediction model,and measures can be taken in advance based on the prediction results to avoid further deterioration of water quality.