With the rapid development of China's economy, the rapid increase in the number of chemical enterprises, which poses a serious threat to the security of surrounding water environment. Therefore, in the form of China's economy into the new normal, the connotation and extension of risk have changed, which makes the water environment risk become more and more complicated. Lake water resources play an important role in the sustainable development of national economy and society. It is an important way to implement the optimal control strategy for the sustainable development of lake water resources in china. However, there are serious chemical pollution problems in the exploitation of lake water resources in our country, such as pH, dissolved oxygen, permanganate and ammonia exceed the standard value, which cause a serious impact on people's production and life. Therefore, it is an important link to evaluate the water quality of Lake. The traditional model is based on the mass and energy of matter and the principle of momentum conservation. Although the physical concept is clear, it is limited to the amount of calculation and parameter identification. In practice, the model structure is simplified and assumed, which affects the accuracy and practicability of the model. For this reason, this paper establishes an input response relationship between water quality index and the corresponding pollution source, and then introduces the improved support vector regression method to classify and evaluate the water quality of Taihu Lake. The experimental results show that the method has the advantages of high precision, strong practicability, simple calculation and so on. It can give a reasonable water quality classification for the lake water samples, which is worthy of popularization and application.