Research on Application of Support Vector Machine Method in Ningbo Marine Ecological Environment Security Prediction

Using the data related to the marine ecological environment of Ningbo City, China from 2003 to 2013, an index system for the prediction of Ningbo marine ecological security was established from three aspects, pressure, state and response. Support vector machine (SVM) method was adopted to establish the prediction model for Ningbo marine ecological environment security. Then phase space reconstruction was performed on multivariate time series, and the evolution trend of Ningbo marine ecological environment security from 2014 to 2016 was predicted. The result showed the SVM prediction model had a satisfying simulation accuracy. It was predicted that the percentage of Ningbo sea areas with water quality of worse than Grade IV from 2014 to 2016 was about 50%, indicating an unsound marine environmental security in Ningbo.