Research On The Optimization Of Cultivated Land Natural Quality Evaluation Based On GA-BP Neural Network

Cultivated land is an important carrier of agricultural production,and it is essential to ensure national food security,meet the needs of industrialization and urbanization for land use,and build an ecological environment.In recent years,problems such as the decline in the quality,fragmentation,abandonment,and ecological degradation of cultivated land have brought major challenges to national food security.In this regard,the state proposes to implement the "three-in-one" protection of the quantity,quality and ecology of cultivated land resources.Based on this goal,the third national land survey also made specific requirements for the classification method system of cultivated land resource quality in terms of ecological environment conditions under the control of natural geographic patterns.However,the current cultivated land quality evaluation index system is mostly considered from the level of soil elements,and there are few studies on the influence of regional factors on the natural quality of cultivated land.Therefore,how to construct a more comprehensive evaluation index system from different spatial dimensions,scientifically quantify the natural quality of cultivated land,objectively understand its spatial distribution law,deeply explore the main influencing factors,and explore new ways to manage and protect cultivated land has become a realistic demand for current cultivated land research..Shanxi Province has complex terrain,obvious differentiation of land resource endowment factors,serious soil erosion,fragile food production conditions,and generally weakened awareness of food security.Taking Shanxi Province as the research area,this study is based on the clarification of the connotation of the natural quality of cultivated land,based on theories of agricultural location,human-land relationship and soil fertility,and conducts the evaluation index system from the four spatial dimensions of soil,plot,ecological environment and climate in accordance with the idea of "element-region" hierarchical control,In this paper,the factor correction method is used to quantify the natural quality of cultivated land.Then taking the evaluation result of the factor correction method as a reference,and the BP neural network—GA-BP neural network is used in turn to optimize the index weights,and the most effective evaluation method is selected by comparing the accuracy of the evaluation results;Finally,the geographic detector model is used to reveal the main factors affecting the spatial differentiation of cultivated land natural quality,in order to provide a basis for the government to differentiate management and to protect cultivated land.The main research contents and conclusions of this paper are as follows:(1)The construction of an evaluation index system that follows the "element-region" overall planning approach.On the basis of defining the connotation of the natural quality of cultivated land,referring to the existing research and combining with the actual conditions of Shanxi Province,this paper selected 15 measurement indicators in four dimensions from the micro-element scale to the macro-regional scale: soil conditions,site conditions,ecological environment conditions,and climate conditions to construct an evaluation system suitable for the provincial scale.(2)The optimization of the weight of the natural quality evaluation index of cultivated land.Taking the natural quality results of cultivated land evaluated by the factor correction method combining subjective and objective weighting as a reference,the BP and GA-BP neural network technologies are successively introduced to optimize the evaluation index weights,and the accuracy of the evaluation results is compared.The results show that the absolute error,relative error and average absolute percentage error of BP neural network prediction output are larger than those of GA-BP neural network.BP,GA-BP and the factor correction method are basically the same as the trend of the area ratio of each grade,but the predicted area of the fourth grade is BP neural network>GA-BP neural network>factor correction method,the fifth grade is BP neural network<GA-BP Neural network <factor correction method.Overall,BP and GA-BP neural networks have strong learning capabilities,but the latter has better predictive effects.Analysis of the possible reason is that the BP neural network has fallen into the local optimum,so that the weight of the network training is not the optimal solution,so the genetic algorithm is introduced to optimize the BP weight through the global search,and the evaluation accuracy is improved.(3)The factor correction method is used to obtain the cultivated land natural quality score,which is divided into five grades by the natural breakpoint classification method,and the main influencing factors of the cultivated land natural quality spatial differentiation are analyzed with the help of the geo-detector model.The results show that the natural quality of cultivated land in Shanxi Province is relatively poor,with few high-quality lands and wide distribution of middle and low lands.It presents a spatial pattern of high in the south and low in the north,high in the basin and low in the mountain.The overall distribution is consistent with the distribution of topography and accumulated temperature.The spatial distribution of the natural quality of cultivated land is the result of the comprehensive effects of multiple factors.In terms of single factor impact analysis,accumulated temperature ≥10 ℃,elevation,effective soil thickness,rainfall and irrigation guarantee rate are the main influencing factors for the formation of spatial differentiation;from the interaction of influence factors,it can be found that the explanatory power of the interaction of each factor is stronger than that of the single factor,which is mainly manifested as nonlinear enhancement and dual-factor enhancement.Therefore,to protect the natural quality of cultivated land,measures can be formulated in terms of improving infrastructure construction,increasing land reclamation,and improving ecological environmental conditions.This study can provide a basis for strengthening the protection of cultivated land.This study can provide a basis for strengthening the protection of cultivated land.