Crop nutrition diagnosis expert system based on artificial neural networks

This research aims at designing an intelligent and carry-home diagnosis expert system (ES) to help inexpert farmers detect crop nutrition disorders in time. To ensure the reasoning veracity of the system, artificial neural networks (ANN) were proposed, and a single chip computer was applied for spot using possibility. Two subsystems and their corresponding ANN clusters were created according to the location where the nutrition disorders first took place. The symptoms of six crops were collected. The confidences and conclusions of symptoms diagnosis by field experts were used as input and output neurons of ANN. Study results were saved as a knowledge base in a flash memory. Using MCS-51C language, single chip computer diagnosis was realized. Field validation indicated that diagnosis errors were less than 8%. Moreover, the combination of ANN and ES can make up traditional expert system defects and improve the systems intelligence and diagnosis efficiency.