Research On Comprehensive Assessment And Prediction Of Poyang Lake Water Quality Based On KH-SVM And GM(1,1)

Poyang Lake is the largest fresh water lake in China,with huge ecological,economic and social values.However,with the rapid development of industrial and agricultural production and urbanization,the water pollution problem in the lake area is increasingly prominent.In addition,the water level of Poyang Lake changes dramatically seasonally,which makes the water quality in the lake area show obvious dynamic change characteristics,and aggravates the complexity and difficulty of water quality assessment and trend prediction of Poyang Lake.Therefore,this paper takes Poyang Lake as the research object,based on the measured water quality data of the lake area from 2013 to 2018,and uses the support vector machine(SVM)model optimized by krill herd(KH)algorithm,namely KH-SVM model,to comprehensively evaluate and analyze the water quality of Poyang Lake,The GM(1,1)prediction model modified based on residual sequence and Markov theory is used to predict the future water quality change trend of Poyang Lake,which provides a scientific basis for decision-makers to understand the water quality status and change trend of Poyang Lake,analyze the problems and cruxes of the current water environment,and formulate relevant protection measures.The main research content and conclusions of this article are as follows:(1)An analysis was conducted on the trend of changes in various water quality indicators in the lake area from 2013 to 2018.The results showed that dissolved oxygen and p H values showed a trend of first increasing and then decreasing year by year;The concentrations of phosphate and ammonia nitrogen showed small fluctuations and a slight upward trend;Nitrite showed a fluctuating upward trend,with its content decreasing in 2017 and then significantly increasing;The nitrate nitrogen showed a trend of first increasing,then decreasing,and then increasing year by year from 2016 to 2018;Chemical oxygen demand,total nitrogen and total phosphorus are relatively stable,and there is no obvious change trend.Based on the above analysis,combined with the significance of each indicator,dissolved oxygen,total nitrogen,ammonia nitrogen,chemical oxygen demand,and total phosphorus are selected as representative water quality evaluation indicators for the subsequent Poyang Lake water quality evaluation study.(2)The KH-SVM model was applied to classify multiple sets of data randomly interpolated based on surface water quality standards,and compared with the original SVM model.The results showed that the accuracy of the former classification was as high as 100%,higher than the latter’s 89.2%,and the time efficiency difference was not significant.The established KH-SVM model compensated for the blind parameter selection of the original SVM model,and the accuracy of classification was also significantly improved.(3)Based on the measured water quality data of Poyang Lake,the water quality assessment results of KH-SVM model and traditional water quality assessment methods(single factor,comprehensive index(WQI),SVM model)are compared and analyzed;The WQI evaluation is overly optimistic,with slightly subjective evaluation principles and overly focused evaluation results;The water quality evaluation results based on SVM do not match the actual water quality characteristics during certain special periods,and the accuracy is insufficient;The KH-SVM model is more objective,accurate,and reliable,and has good applicability in water quality evaluation.(4)The water quality assessment results based on the KH-SVM model show that the water quality of Poyang Lake from 2013 to 2018 showed a deterioration trend to a certain extent in time,and gradually improved from south to north in space.The water quality in the South Lake area fluctuated greatly,and the high flow period was better than the low flow period,and the water quality in all periods reached the Class V level by 2018;The water quality in Beihu District is relatively stable,with dry season being better than wet season being better than normal season.The water quality is mainly classified as Class III-IV,but it also shows a trend of concentrated deterioration towards Class IV,especially during dry season.(5)The water quality prediction and evaluation results based on the modified GM(1,1)model and KH-SVM model show that the total nitrogen concentration in Poyang Lake will continue to rise slightly in the next five years,and the other indicators will continue to decline;In the zoning prediction,the total phosphorus in the South Lake area and the total nitrogen and ammonia nitrogen in the North Lake area showed an upward trend,while the other indicators showed a significant downward trend;In the future,the water quality categories in the lake area will mainly be Class III-IV,and the overall water quality situation is expected to improve.However,the water quality categories in the southern and northern lake areas have not significantly improved.