Abstract: The eutrophication process of rivers and lakes is affected by the non-linear superposition of many factors such as water pollution, habitat destruction, and dam control, which limits the simulation accuracy of conventional aquatic ecological mechanism models to a certain extent. Non-parametric models have been widely used in the diagnosis and prediction of river and lake water ecological problems with their powerful data analysis capabilities. This paper systematically summarizes the relevant research with the visual analysis of big data in related literature based on WoS and CNKI databases, and comprehensively clarified the applicability and limitations of mainstream non-parametric models, such as, structural equation model (SEM), Bayesian network (BN), support vector machine (SVM), artificial neural network (ANN), random forest (RF), gradient propulsion machine (GBM), and generalized additive model (GAM) in the study of eutrophication of rivers and lakes. Comparative analysis of models with similar characteristics and prospects is put forward. It is expected to provide scientific and effective method support and application progress summary for research related to water ecological simulation. The results show that the application of non-parametric models in the field of river and lake eutrophication research is increasing exponentially. Among them, SEM, BN, RF, GBM and GAM models are suitable for the diagnosis of river and lake eutrophication problems and the identification of driving factors. BN, ANN, SVM, RF, GBM and GAM have good nonlinear fitting and prediction capabilities. Non-parametric models will become the key technical means for the development of aquatic ecological big data analysis and diagnosis and prediction control in the future. Comprehensive consideration of the regional heterogeneity and the response of multiple environmental factors on different time and space scales and the risk of river water ecological degradation under the interference of strong human activities, the use of ecological mechanism model and non-parametric model coupling solution and optimization algorithm introduction, accurately identify the environmental pressure threshold of water ecological health degradation, and carry out the prediction and early warning of water ecological degradation risk under changing environment, which will be the general direction of the future application of non-parametric models in the eutrophication of rivers and lakes.