Crop nutrition diagnosis expert system based on artificial neural networks

Aimed at the low self-learning ability drawbacks of traditional expert system, artificial neural networks(ANN) were applied in crop nutrition diagnosis system. Wheat nutrition disorder symptoms were collected from the following five aspects: macroscopical phenomenon, stalk (root) symptoms, leaf symptoms, fruit symptoms and crop nosogenesis. After field diagnosis from these five aspects by experts, the confidences and the corresponding conclusions which they inputted were as the input neurons and output neurons of the ANN. Running on a personal computer, the ANN study results were obtained, and then were saved in flash memory and regarded as the system knowledge base. Using MCS-51C language, the diagnosis function was realized on the single chip computer. The analysis of the field validation test results indicates that this system adequately simulates the expert diagnosis process, and greatly improves the diagnosis efficiency.