Research On Hillslope Soil Erosion Evaluation Model And Method Based On Association Rule In North-west Of Hubei Province

Soil erosion concerns are both local and completely global, and it has attracted much attention and research. This paper addresses the issue of hillslope Soil erosion of Danjiangkou Reservoir region through artificial slope runoff plots of FangXian and YunXi county. In this paper, the author First introduces the research data Sources of Danjiangkou Reservoir region and the data uploading to the SQL server database. Based on the overview of oversea and domestic research generalization and main assessment methods, The paper presents its adopted three kands of model or research methods. The first adopted method is the Aprior algorithm based on association rule aiming at qualitative analysizing the pertinence relation between the slope Soil erosion and its influence factors. The second method is to build the Universal Soil Loss Equation and apply to some county in Danjiangkou Reservoir region. Setting up the support vector machines(SVM)modle for forecasting the soil erosion modulus.According to the current data of research area stored in SQLserver database, this paper first explains related concepts of da ta mining, and then select the Apriori algorithm for Association Rule analysizing among the hillslope soil erosion and its influence factors,such as farming methods factor, ratio of slope factor, forest vegetation coverage factor. By means of quantizing and achieving related analysizing these influence factors, we obtain the qualitative interdependence coefficients between the hillslope erosion and these factors.Universal Soil Loss Equation model(USLE) is one of the most influential evaluateion statistical modle. And until now it is still adopted by the soil and water conservation workers and relevant department researchers in wide use. Because of the model's special fitness for researching hillslope soil erosion, the paper first introduces the prototype model of USLE, and then talks in detail about the computing method for every influence factor of the USLE model. In the end the USLE modle applicable to the Danjiangkou Reservoir region is successfully derived from the prototype model and available research data. we calculate the amount of soil erosion on hillslopesin respective year from above equations. Ultimately the error analysis has conducted. Teh experiments show that this method is satisfactory, practical and effective.Machine Learning based on experimental data has become the focus of currently study in many industries. As two kinds of machine leardnings, back-propagation neural network(BPNN) and SVM algorithm shed light on some sets of observed data and discover some regularities which can't be obtained by principle analysis. And simultaneously these two models can use these regularities for forecasting in connection with the future or inaccessible data.This article first gives a brief introduction to the regression principle of BPNN and SVM model. And then the paper gets the predicting result of the amount of hillslope soil erosion and error analysis by using these models based on the experimental data of the research area. The results show that the two methods,especially the BPNN method, are very effective in connection with our research data. At the same time,there are several undesirable Predicting Outcomes. The reasons for this have to do with the limitations of existing data, experimental data precision and the parameter selection of the model.