Abstract: From the artificial intelligence point of view, based on the characteristics and feature interpretation of rice pest insects and diseases, a new knowledge expression method based on weight self-adjustment artificial neural networks was presented. In this paper 15 main rice diseases were selected as examples, through classifying, abstracting and coding on the characteristics and feature. The Back Propagation neural network was implemented using neural network and uncertain technology, samples were trained and the weight was regarded as knowledge base to diagnose rice diseases, and in this way rice pest insects and diseases diagnosis expert system was constructed.