Research On The Diagnosis Method Of Rice Pests And Diseases Based On Knowledge Graph

As one of the main food crops,the planting area of rice has gradually been increased.However,production reduction and food security issues during the planting process often occurred due to rice diseases and insect pests.Not only leads to the economic depression of the whole industry,but also affects the stable and development of society.Solving the problem of rice pests has become one of the main directions of current scholars’ research.The government has also enhanced its system for dealing with crop pests and diseases on all fronts and promulgated the Regulations on the Control of Crop Pests and Diseases.President Xi Jinping once said: "The Chinese should fulfill our bowl with our own cereals”.At present,expert system is an effective means to solve rice pests and diseases.Rice diseases and pests data have the characteristics of multi-source,heterogeneous and diverse types,and traditional expert systems are still inadequate as they have difficulty in handling the correlation between complex data.Therefore,it is significant to study the relationship between complex data needs to be fully explored and diagnosed with high efficiency.As for such problems,based on the knowledge graph of the network structure as the bottom storage,this paper constructs the knowledge graph method for rice diseases and insect pests.Intelligent diagnosis model was further investigated by the constructed method.The main work of the thesis includes:(1)Construction of knowledge graph of rice diseases and pests.The knowledge graph schema layer and data layer are constructed based on the collection,representation,extraction and storage of data.The neural network model of Bi-directional Long Short-Term Memory-Conditional Random Fields was used to recognition rice diseases and pests entity to obtain knowledge of the data layer.The accuracy of the model in recognition rice diseases names and pests names and hazard symptoms were 91.67%,89.66%,and 87.02%,respectively.(2)Rice diseases and pests diagnosis model was constructed based on the knowledge graph.To solve the time-consuming problem of dynamically changing data processing,SKIP LIST algorithm was introduced by analysing and comparing the retrieval efficiency from contact link,community division and similar disease discovery.At the same time,a diagnosis algorithm was constructed based on Certainty Factor(CF)model and knowledge graph to diagnosis of rice pests and diseases by employing the symptoms of diseased rice plants.This model has increased the accuracy and interpretability of diagnosis and given full play to the practical value of data in this field.(3)An intelligent diagnosis system was independently developed for rice diseases and insect pests.The functions of displaying,searching,diagnosing and early warning of rice diseases and insect pests,and displaying entity relationships through ECharts visualization tools are realized.Also,the application of rice pests and diseases data in different scenarios is realized.The stress test and comprehensive evaluation of the system have obtained good results.