| A whole new concept as precision agriculture has been formed and developed rapidly in the world with the process of global agriculture informationization .Under the background variable-rate fertilization decision making in precision agriculture is researched in this paper .The main contents of this paper are as follows:(1) Traditional statistical method is applied to analyze the nutrient of field ,the conclusion is gained as follows:variation coefficient of N ,P ,K are 0.47,0.31,0.18. The result shows that the soil nurients have obvious variation ,So variable-rate fertilization has extensive objective foundations.(2) Brassica chinensis is used as experimental object ,by experiment in the field ,the quantitative mathematical model between yield and the application rate of N ,P , K concentration is established .It is proved that the relationship of quadratic function exists obviously between yield and feitilizer factors. The function provides the evidence for feitilizer decision making .(3) The variable -rate fertilization decision making model based on Artificial Neural Network is built and it can implement most economic benefit. "4-4-10-3"BP neural network structure, the inputs of the model are soil nutrient(N, P and K) and yield goal ,the outputs are application rate of fertilizer (N, P and K). The model reflects the nonlinear relationships among soil nutrients, application rate of fertilizer and yield . Fertilizer prescription integrated soil nutrient distribution maps with variable -rate fertilization decision making model is built .It provides evidence for variable -rate fertilization decision making .(4) The system of variable -rate fertilization decision making is programmed by Visual Basic 6.0 and implements variable -rate fertilization decision making . |