基于太湖藻类演变规律的PSO-νSVR模型应用研究

Application of PSO-νSVR Model based on Evolution Pattern of Algae in Taihu Lake

基于2001年1月-2018年12月太湖各观测站点位数据,分析营养盐时空分异特征、藻类浓度演变规律;选择六个水质指标作为输入变量,叶绿素a浓度作为预测变量,使用改进粒子群支持向量回归PSO-νSVR进行水质预测。结果显示:自1996年氮磷达峰值后,太湖整体呈先恶化、后好转的波动变化趋势;主要污染区域位于太湖和北部水域,氮、磷指标变化趋势相同,但变化幅度氮要大于磷;空间异质性明显,西部和北部水域变化幅度大于其它湖区。整体趋势符合浓度发展预测,拟合预测的回归与收敛效果良好。

Based on the data from January 2001 to December 2018 at observation stations in Taihu Lake,the spatial and temporal variation characteristics of nutrients and the evolution pattern of algal concentrations were analyzed.Six water quality indicators were selected as input variables and chlorophyll a concentrations were used as predictor variables for water quality prediction using improved particle swarm support vector regression PSO-νSVR.The results showed that:since the peak of nitrogen and phosphorus in 1996,Taihu Lake as a whole has shown a fluctuating trend of deterioration followed by improvement.The main pollution areas are located in Taihu Lake and the northern waters,with the same trend of change in nitrogen and phosphorus indicators,but the magnitude of change is greater for nitrogen than phosphorus.Spatial heterogeneity is obvious,and the magnitude of change is greater in the western and northern waters than in other lake areas.The overall trend is consistent with the prediction of concentration development,and the regression and convergence of the fitted prediction works well.