Evolutionary Game Analysis of Farmers' Conservation Tillage Behavior in Black Soil Areas Guided by Deep Learning

To better protect the rights and interests of farmers, the evolutionary game theory and deep learning (DL) technology are used to analyze the conservation tillage behavior of farmers in black soil areas. Firstly, the basic hypotheses are put forward and an evolutionary game model is constructed. Secondly, the evolutionary game model between farmers and the government is analyzed, and dynamic equations are built. Finally, the model is deduced and studied, and a Convolutional Neural Network (CNN) with a double-sized convolution kernel is constructed to classify the land of remote sensing images. The experimental results manifest that after the dynamic evolutionary game, the net income generated by farmers adopting the conservation tillage strategy without government regulation is positive. The net income of government regulation is positive, and the game equilibrium point is (1, 0). After a dynamic evolutionary game, the game is balanced when the government does not regulate, the net income generated by farmers adopting conservation tillage strategies and the net income generated by government regulation and farmers adopting conservation tillage strategies are negative, the net income from government regulation is positive, and the game equilibrium point is (0, 1). The constructed CNN can achieve 91.32% overall accuracy for black soil classification, and the proposed scheme provides some references for the application of CNN in evolutionary games.