The Research for the Evaluation of Cultivated Land Quality Based on Deep Belief Networks

Qiao-bing YUE, Xu-bing ZHANG


Traditional evaluation methods of cultivated land quality are mainly on the basis of empirical judgments in the process of weight calculation and membership determination. In this paper, taking Enshi city as an example, we attempt to employ Deep Belief Networks (DBN) to estimate the cultivated land quality. We selected 8 evaluation indexes as the input data and the grade of cultivated land as the output data to construct an evaluation model and finally realized the classification of cultivated land quality grade. Compared with the results of the Supplement and Improvement of Farmland's Classification in Enshi city 2013 in quantity and spatial distribution, our experimental results show that accuracy of grade 11, grade 12 and grade 13 in quantity reach as high as 99.06%, 90.04% and 96.07%, respectively. Besides, the spatial distribution is substantially in agreement with real distribution.


Cultivated land quality, Grading evaluation, Deep belief networks.Text


Full Text:



  • There are currently no refbacks.