Soft sensing of seawater chlorophyll-a based on support vector machine algorithm

Recently with the rapid developing of economy, the problem of seawater eutrophication due to pollution of marine environment becomes more serious. Seawater eutrophication will lead to more reproduction of algae in seawater, the reproduction state of algae can be obtained by monitoring the concentration of seawater chlorophyll-a. A soft sensing method of measuring the concentration of seawater chlorophyll-a based on Support Vector Machines(SVM) has been proposed in this paper. Cross-validation method was used for searching the optimal parameters of SVM model. The result of soft sensing based on SVM has been compared with the way of T-S fuzzy neural network, the testing result indicates that this soft sensing method based on SVM can effectively estimate the concentration of seawater chlorophyll-a, and then perform the real time state monitoring of seawater eutrophication.