| Agriculture Internet of Things is one of the important development directions of Internet of Things. Taking the technology of Internet of Things in the field of agriculture can bring immeasurable impetus and immeasurable prospects for the development of agriculture. In the Agriculture Internet of Things, how to process, analysis and display the data information which collected in the production, and make it better served for the agriculture is the critical issues now needed to solve. The data processing of Agriculture Internet of Things is located in the information application layer of Agriculture Internet of Things. After the data processing, integration and application, it can make scientific management decisions, and achieve control the agricultural production process. Now the study of data information processing of Agriculture Internet of Things is still in the research and development stage. It faces challenges in three areas:(1) it lacks agriculture forecasting model which considerate the dynamic characteristics of agricultural data and the impact of historical data makes agricultural production data dynamics and historic are ignored;(2) the knowledge processing and decision-making models of the existing expert system are almost all developed based on a particular crop, some crops or for a few aspects. Knowledge processing and decision-making model to process large quantities and many types of agricultural knowledge has not yet appeared;(3) in the area of processing customized, multi-class agricultural production control lacks appropriate solutions.In order to solve the above problems, we need to systematically study the forecasting models of Agriculture Internet of Things, knowledge processing and decision-making models, and so on. Therefore, this thesis studies the key technologies in three aspects:the forecasting models of Agricultural Internet of Things, knowledge processing and decision-making models, and the access control of the production process. Firstly, we introduce the exiting agriculture data processing system and the key technology of data processing in Agriculture Internet of Things. Secondly, in the area of forecasting model, for considering the dynamic of process information and reflecting the influence of historical data, we propose an improved output-input feedback mechanism Elman neural network (IOIF-Elman neural network) model. In the area of knowledge processing and decision-making model, through the dynamic inheritance and establishing hierarchical experience libraries, we design a knowledge processing and decision-making model based class, frames and production rules. In the area of production control, to solve the access control problem in the agricultural production process information processing system which can be customized and processes multiple categories, we propose a workflow access control model based on attributes and tasks, and elaborate their access processes in details.The main innovations generated in this study are:(1) For solve the problem of BP neural network forecasting model which is widely used in agriculture forecasting area, BP neural network can not meet the dynamicly processing information and loss the effects of historical data, we propose an improved output-input feedback mechanism Elman neural network (IOIF-Elman neural network) model The new model takes the dynamic neural network model——Elman neural network as a base, and adds the output layer feedback and the output layer feedback at the last time on it. By increasing the two feedback layers, it can reflect the impact of historical data to prediction, and to get a more accurate prediction results. Simulation results show that, IOIF-Elman neural network has a better predictability and stability than the Elman neural network and BP neural network. Adding the feedback network enhances IOIF-Elman network the capacity of processing dynamic information, and is more meeting the practical applications.(Chapter III)(2) Almost all of the existing agricultural expert system at home and abroad developed based on a certain crops or several similar crops. However the species category of involved in the agricultural production is various. In order to avoid repeated work burden of development, developing information system which has the versatility, and can handle a variety of agriculture product at the same time, is the one of the main agricultural expert system research direction. The biggest problem to develope this system lies in how to deal with large quantity and variety of agricultural knowledge. This paper proposes a knowledge processing and decision-making model based class, frames and production rules. The characteristic of the model is taking the key control points in the process of production as the main line to divide the class, and through establishing hierarchical case storehouse and introducing base-case to increase the speed of reasoning and decision-making. The model applys in the agricultural production record data processing and service system. The actual application shows that knowledge processing and decision-making model based class, frames and production rules can effectively solve the general agricultural system faced problem that large quantity and variety of agricultural knowledge processing and decision-making. Compared with the existing knowledge processing model, it is more concise, complete in representation of knowledge, and has strong ability of semantic comprehension and automatically inherit. Layered thought and the introduction of the base-case can be more accurate and rapid to reason-decision.(Chapter IV)(3) In order to optimize the control ability of agricultural production, against how to design a workflow system model for customized agricultural data processing system, we proposed a workflow access control model based on attributes and tasks. The customized system can process data for all kinds of farm product, but this make the existing workflow model is not suitable for the customized data processing systems to control and assign privilege. By describing user and task with attributes, the workflow access control model based on attributes and tasks make attributes linking with tasks, tasks linking with privileges, and achieve dynamic management of privileges by the task. Compared with other models, it is more fine-grained and better flexibility, and reduces the administrative costs. The model has applied in the agricultural products biographical data processing and services system to verify its feasibility.(Chapter V)The research content of this thesis, as the academic achievements of National Key project of Scientific and Technical Supporting Programs "Agriculture Internet of Things and the research of food quality and safety control system"(No.:2011AA100706), has been applied in the "agricultural products biographical data processing and service system", and provided effective solutions and engineering practice guidance to help them achieving the data collection and processing services in agricultural production process. |
PDF Full Text Request