A Comparative Study Of Cultivated Land Quality Evaluation Based On Factor Method And BP Neural Network Model

Carrying out the evaluation of cultivated land quality is a necessary work to accurately grasp and comprehensively find out the quality level of cultivated land,and also lay a solid foundation for the efficient protection of cultivated land,scientific management of cultivated land,and orderly improvement of cultivated land quality.In the context of the new era,China is ushering in new opportunities for comprehensive economic and social development,and is also facing new challenges.The demand for cultivated land,especially the quality of cultivated land,is becoming higher and higher.Therefore,the evaluation of cultivated land quality also needs to be updated in time to keep up with the society Trend,meet the needs of the times.The traditional cultivated land quality evaluation methods are mostly qualitative,and the quantitative methods are relatively lacking.Innovative evaluation methods focus on the combination of qualitative and quantitative methods and enrich the improvement of cultivated land quality evaluation methods and means.The article is based on the use of two methods to evaluate the quality of cultivated land,comparative analysis of the evaluation results,sum up experience,and learn from each other.Comprehensively analyze the factors influencing the quality of cultivated land in Xiangyang urban area,and construct an evaluation index system that integrates the three main aspects of nature,economy,and location into climate,terrain,soil,etc.,a total of 20 indicators,using formula calculation,spatial difference Values,superposition analysis and other methods are used to obtain the index data and give quantitative processing.The factor method and the BP neural network method are used to evaluate the quality of the cultivated land.The Delphi method and the entropy weight method are used to comprehensively determine the index weight of the factor method and construct"Factor-Quality Score" table,summarize the comprehensive quality score;construct a 20:12:1 three-level BP neural network framework,confirm the activation function,learning rate,target error and other related parameters,train the model through sample data,input data for simulation exercises In order to obtain the score of cultivated land quality,the evaluation results of the two methods are compared and analyzed to explore the similarities and differences of the methods and the scope of application.The main research results of the article are as follows:(1)The quality of cultivated land in Xiangyang urban area is generally good,the quantity of cultivated land is distributed in the central urban area,and the surrounding suburbs are more.The overall quality of cultivated land shows a trend of decreasing quality from northeast to southwest.The factor method uses two methods to determine the weight,quantify and assign the indicators,and summarize and sum the scores to obtain the cultivated land quality score;BP neural network method is based on building a model,continuously debugging parameters,reducing errors,and training through sample data The model is then substituted into the data of each block to predict the comprehensive score.(2)The evaluation results of the two methods are good,which is consistent with the trend of the cultivated land and other results.The overall accuracy R of the BP neural network model reaches about 0.93.The comparison between the predicted quality score and the actual quality score shows that the relative And the absolute difference are relatively small;the overall difference in the evaluation results of the two methods is not large,only a small difference in the quantity of cultivated quality and the spatial distribution of cultivated land quality,therefore,it can be preliminarily considered that both methods can be used in the evaluation of cultivated land quality Use,and the evaluation effect is better.(3)Compare and analyze the evaluation process and evaluation characteristics of the two methods,and draw conclusions that each method has its own suitability.When performing specific evaluation work,it should be selected according to local conditions to meet the needs of the evaluation area and meet the characteristics of the evaluation area.More suitable evaluation method.The factor method is relatively independent and operable,and can be used in areas where the previous work and data are incomplete and lacking;the BP neural network method is convenient,fast,and objective,but it has more requirements on the previous work and data and is suitable for Use in areas with good pre-foundation and complete data.