Ecological Security Evaluation of the Cultivated Land in Yuxi Based on GA-BP Neural Network

PurposeTo resolve the limitations which occurred in land ecological security evaluation, such as slow convergence rate and getting into local minimum, genetic algorithm (GA) had been imported to improve BP neural network.

MethodAn evaluation index system composed of ecological pressure and ecological support, two sub-systems and 18 indices, had been built up according to cultivated land status in Yuxi City. Improved BP neural network based on GA was established to be applied in ecological security evaluation of cultivated land from 2001 to 2015. Simultaneously, BP neural network and GA-BP neural network had been compared in their performance and errors and both were compared with comprehensive index method by the three assessment results.

Results1) Evaluation results showed that ecological security index declined steadily from 0.772 7 in 2001 to 0.280 2 in 2015 and the security grade fell from safer (II) to unsafe (IV); 2) Compared with the traditional BP neural network, GA-BP neural network had the advantages that included small errors in training and prediction, fast convergence and higher accuracy in assessment results.

ConclusionsGA-BP neural network can be applied to evaluate the ecological safety of cultivated land and had great value in application for the improvement of network performance.