基于投影寻踪的湖泊富营养化程度评价模型

Lake eutrophication evaluation model based on projection pursuit method

针对湖泊富营养化评价指标的不相容性以及现有方法对等级的分辨率较粗略等问题,基于我国湖泊富营养化评价标准构建了投影指标函数,引入粒子群算法优化投影指标函数寻求最佳投影方向,应用最佳投影方向计算投影值,从而建立了评价湖泊富营养化等级的投影寻踪模型。同时引入混沌变量解决了采用随机生成样本系列的方法确定分段函数端点值存在的困难。实例应用结果表明,基于混沌映射的投影寻踪评价模型以投影值为单指标,采用分段函数的模型形式解决了湖泊富营养化的多指标综合评价问题,且该模型对湖泊富营养化等级的分辨率较高。

In connection with the problems of incompatibility of evaluation indexes and low resolution of the evaluation grade, and based on the assessment criteria for lake eutrophication in China, a projection index function was constructed. Particle Swarm Optimization （PSO） was introduced to search for the optimal projection direction, which was then used to calculate the projected value. The Projection Pursuit Model, which describes the relationship between the projected value and the grades of lake eutrophication, was developed based on these approaches. The difficulty in determining the end point values of the piecewise function was solved by introducing a chaotic variable when using a randomly generated sample series. The results from the practice application showed that based on the chaotic mapping the Projection Pursuit Model used projected value as a single index and adopted a piecewise function solved the problem in using a multi-index comprehensive assessment method. The model has a high resolution in grading lake eutrophication.