基于支持向量机的城市湖泊水质评价研究

发布时间：2018-03-03 17:50

本文选题：支持向量机　切入点：小样本　出处：《湖南大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来，随着城市规模的不断扩大，人口的日益增长和经济的快速发展，水环境的污染问题日渐加剧，对区域经济发展、饮用水安全和生态环境构成了严重威胁，并且日益严重的水污染问题严重制约了我国经济社会的发展。根据《2010年中国环境状况公报》数据显示，由环保部组织监测营养状况的26个湖泊中，呈现富营养化的湖泊占42.3%。城市湖泊几乎全部处于重富营养化或异常营养化状态，绝大多数大中型湖泊已在富营养化的边缘或已经处于富营养化状态（中华人民共和国环境保护部，2010）。因此，科学有效地揭示湖泊水质健康状况以及合理的水质评价，不仅可了解和掌握水体的污染程度和发展趋势，而且能为湖泊管理部门制定有效的控制方案及政府部门的环保决策提供有力依据。本文基于以上情况，利用支持向量机模型良好的推广和泛化能力，以及在处理分类问题及小样本问题方面的优势，构建了湖泊水质评价模型，并运用此模型对白云湖的水质评价进行了研究。在两年多的研究时间内，对广州市白云区白云湖的水质进行了八次跟踪监测，获得了大批相关数据。依据国内外相关研究及湖泊的实际监测数据的分析，选取了溶解氧、化学需氧量、生化需氧量、氨氮、总磷、石油类在内的六项指标，并借鉴国家制定的环保标准对六项指标进行了分级，构建了由五个等级组成的以白云湖为研究对象的城市湖泊水质评价指标体系。最后，以评价指标为训练样本在支持向量机上进行训练，得到水质评价模型，并将该模型应用于白云湖水质评价中，进而获得白云湖的水质健康状况。结果显示，白云湖的水质2013年4月份比2011年1月份有明显好转，但其生态系统还十分脆弱，，净化水质能力有限；广和泵站引水口的水质等级提高时，其出水水质也有相应的提高。根据评价结果，提出了对白云湖要进行全面截污，提高进水水质，并增加挺水植物、沉水植物的数量以及对石油类污染物的处理措施等方面的建议来改善湖泊水质健康状况。此外，本研究中还将支持向量机的湖泊水质评价模型与传统的单因子评价法和模糊综合评价法进行了比较，评价结果显示，支持向量机水质评价模型较传统的评价方法而言，没有水质参评项目个数的限制，能够全面、客观地反映水资源各水质参评项目的综合污染程度，在小样本的情况下也能得到较好的分类结果，能够为决策者在制定决策时提供更加科学客观的依据，有助于促进环境友好型社会的一步发展，具有广阔的应用前景。
[Abstract]:In recent years, with the continuous expansion of urban scale, increasing population and rapid economic development, the pollution of water environment has become increasingly serious, which poses a serious threat to the development of regional economy, the safety of drinking water and the ecological environment. And the increasingly serious problem of water pollution has seriously restricted the economic and social development of our country. According to the data from the 2010 China Environmental status Bulletin, the Ministry of Environmental Protection organized and monitored the nutritional status of 26 lakes. The proportion of lakes with eutrophication was 42.3%. Almost all the urban lakes were in the condition of heavy eutrophication or abnormal nutrition. The vast majority of large and medium-sized lakes are already on the edge of eutrophication or in eutrophication (Ministry of Environmental Protection of the people's Republic of China). It can not only understand and master the pollution degree and development trend of water body, but also provide powerful basis for lake management department to make effective control plan and environmental protection decision of government department. Based on the above situation, a lake water quality evaluation model is constructed based on the good generalization and generalization ability of support vector machine model and its advantages in dealing with classification problems and small sample problems. The water quality evaluation of Baiyun Lake has been studied by using this model. In more than two years, the water quality of Baiyun Lake in Baiyun District of Guangzhou City has been tracked and monitored eight times. A large number of related data have been obtained. Based on the analysis of domestic and foreign related research and the actual monitoring data of lakes, six indexes, including dissolved oxygen, chemical oxygen demand, biochemical oxygen demand, ammonia nitrogen, total phosphorus and petroleum, are selected. And draw lessons from the national environmental protection standard to grade the six indexes, and construct the evaluation index system of urban lake water quality, which is composed of five grades, taking Baiyun Lake as the research object. The water quality evaluation model is obtained by training on support vector machine with evaluation index as training sample. The model is applied to the water quality evaluation of Baiyun Lake, and then the water quality health of Baiyun Lake is obtained. The results show that the water quality of Baiyun Lake is healthy. The water quality of Baiyun Lake in April 2013 was obviously better than that of January 2011, but its ecosystem was still very fragile and its ability to purify water was limited. According to the evaluation results, it is suggested that the Baiyun Lake should be cut off completely, the influent water quality should be improved, and the plants should be increased. The quantity of submerged plants and the treatment of petroleum pollutants are suggested to improve the water quality and health of lakes. In addition, the lake water quality evaluation model based on support vector machine is compared with the traditional single factor evaluation method and fuzzy comprehensive evaluation method. The evaluation results show that the support vector machine water quality evaluation model is more than the traditional evaluation method. Without limiting the number of water quality assessment items, it can comprehensively and objectively reflect the comprehensive pollution degree of each water quality assessment project, and can also obtain better classification results in the case of small samples. It can provide a more scientific and objective basis for decision-makers to make decisions. It is helpful to promote the one-step development of environment-friendly society and has a broad application prospect.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：X824

【相似文献】