基于PCA-SVR的池塘DO预测模型

为解决传统水质预测模型泛化能力低、预测精度差等问题,提出了基于主成分分析和支持向量机相结合的养殖池塘溶解氧预测模型.该模型通过主成分分析筛选反映池塘水体溶解氧信息的关键指标,减少模型输入变量,采用支持向量机算法建立水质预测模型,并用于长沙市乔口镇与望城区池塘养殖溶解氧预测中.结果表明,该模型预测精度高,同时具有很强的泛化能力与适应数据变化的能力,可用于池塘溶解氧预测.

Forecasting model for DO of pond water quality based on PCA-SVR

In order to solve the problem of low prediction accuracy and poor generalization ability of the traditional forecasting methods in water quality, this paper proposed forecasting model for DO value of pond water quality based on PCA-SVR.The model picks key indicators which can reflect DO condition of pond water environment by the principal component analysis, reduce the model input variables, uses support vector machine algorithm for establishing water quality prediction model and adapts it to pond aquaculture water of Qiaokou town and Wang cheng district, Changsha.The application examples show that the model prediction has strong generalization ability and adaptability to change of data and functions,meanwhile has high prediction precision, it can be used to forecast aquaculture water dissolved oxygen quality.