Evaluation Of Water Quality Of Urban Lakes System Based On Support Vector Machine (SVM)

In recent years, with the expansion of the city, the population increasing andrapid development of economy, increasing pollution of the water environment notonly poses a serious threat for the regional economic development, drinking watersafety and the ecological environment, but also hampers the development of China’seconomy and society. According to the State of The Environment In China’s Bulletinin2010, lakes with eutrophication status accounted for42.3%in twenty-six lakes.Meanwhile, almost lakes were heavy-or abnormal-nutritional eutrophication state.Thus, scientific effectively revealing the health state of lake water quality and areasonable quality assessment is very important for government to develop effectivecontrol programs and policy.Support vector machine (SVM) is easy to be spread and generalized and has anadvantage in dealing with classification and small sample size problems. An evaluatemodel of lake water quality was proposed based on SVM and was used to evaluate thewater quality of Baiyun Lake in Guangzhou. According to the data of water quality inBaiyun Lake in Guangzhou during two years, six indicators involving dissolvedoxygen, chemical oxygen demand, biochemical oxygen demand, ammonia nitrogen,total phosphorus, oil were chosen to assess the health state of Baiyun lake waterquality. Meanwhile, these indexes were classified into five categories based on theenvironmental standards and previous research. The results of SVM model thatapplied into Baiyun Lake water quality assessment were obtained by training thesamples of indictors. It was showed that the water quality of Baiyun Lake in April2013has significantly improved over January2011, but still very fragile, the ability topurify water was limited. With the diversion of water quality i mproving, its outletwater quality is also a corresponding increase. According to the results, to improvewater quality system health of Baiyun Lake, three methods were suggested includingincreasing the submerged plants and floating plants, improving inpu t water quality,intercepting pollution sources and processing petroleum contaminants.In addition, we compared the support vector machine (SVM) of lake waterquality evaluation model with traditional single factor evaluation and fuzzycomprehensive evaluation. The results showed that SVM model can comprehensivelyand objectively reflect the pollution levels owing to the unlimited of the number of water project. The SVM model can receive a better classification results even undersmall sampling condition, thus can provide more scientific and object basis to makedecisions. SVM model can also promote the development of environment-friendlysociety and have broad application prospects.