Construction Of Diagnosis Model Of Rice Diseases And Insect Pests And Implementation Of The Intelligent Diagnosis System

This thesis is one part of the subprogram titled as " Key technology Research On networked Real-time Diagnosis models and Interactive Platform of Remote Identification for Agricultural Pests " of the China National High-Tech Program (863 Program) (2002AA243041). In the thesis, diagnosis knowledge of rice insect pests and diseases was sorted systematically and studied in detail, and an intelligent decision-making system was designed and realized.Diagnosis model and inquired diagnosis system of rice insect pests and diseases are mainly composed of knowledge base, knowledge acquisition, integrated database and reasoning engine. Knowledge base is mainly used for storing rule of diagnosing rice insect pests and diseases; knowledge acquisition which facing to agricultural experts implementes the function to modify and expand the knowledge database; Integrated database is used for storing the basic information of rice, users get the inquired information from the database; According to the original information, Reasoning engine works on the base of diagnosis model. During the designing procedure, based on Delphi 6 Platform, the expert system was implemented by using Object-Oriented programming technology, ADO (ActiveX Data Objects) technology and Database technology. The expert system integrated with the database technology to accomplish the storing and management of the rules for reasoning.A diagnose method of diseases and insects pests based on character was presented and realized in this paper. Through the arrangement and collection of the symptoms of species of diseases on the different parts of rice (leaves, stem, panicle) and the individual character of species of insect pests, the knowledge base was implemented based on the character of rice pests and diseases. According to the characteristics of rice insect pests and diseases, several methods of knowledge expression were applied in the diagnosis model. One knowledge expression method adopted the relationship expression. In this method, multi-conditional composition approach was used to express knowledge and to store all the knowledge in the relational table. Correspondingly, the inferred engineer was designed by SQL (Structured Query Language) language. Another knowledge expression method adopted the way of character-rule table, through the collection and sorting the symptoms of plant diseases and insect pests. Using BP (BackPropagation) neural network theory, the multi-level neural complicated network architecture model was built, and a diagnose sub-system of diseases and insect pests on the basis of symptom was realized.The system efficiently combined the computer technology with knowledge of experts in plant protection, and integrated the storing, browsing and inquiring function of information of insect pests and plant disease into comprehensive body. The application of Geographic Information System and multi-media technology augmented its practicability and intuitiveness for users. It was important and meaningful for mitigating the lack of experts in plant protection and improving the capability of protection rice from rice diseases and insect pests, and hastening the transformation of achievements of research and development into application.