| With the vigorous development of computer technology,the informationization of modern forestry has become a necessary condition.However,the development level of Chinese agriculture and forestry information is seriously insufficient at present.It has not achieved the real-time collection and monitoring of agricultural and forestry crop growth environment information and the occurrence of diseases and insect pests,nor can it achieve the effective collection and sharing of historical data.The decisions of front-line producers and agricultural and forestry management departments are often based on experience and difficult to be data-driven.In view of the current situation of agroforestry information service in China,based on the technologies of Internet of things design,automation design,server software design,mobile APP software design,Web development,database and image processing,this paper studies and realizes a remote monitoring system of agricultural and forestry diseases and pests and meteorological information,which can automatically collect,visualize management,identify and warn diseases and pests to make up for the shortcomings of the existing agricultural and forestry information construction.The main research contents of this paper are as follows:1.The system information collection device is designed and realized.STM32,Raspberry Pi and its peripheral devices are selected to build the hardware platform.A new image acquisition device for disease spores based on tape is designed.And an embedded software is developed to realize the automatic collection,transmission and backup of meteorological information,disease and insect pest information in the growing environment of agricultural and forestry crops.2.The system server software is designed and implemented.The dynamic SQL is explored and designed to realize the dynamic expansion of the collection hardware.To meet the need of strong real-time connection,Socket service is designed and implemented by C++.The service receives data transmitted by multiple acquisition terminals concurrently,saves it to the database,and automatically detects the algorithm and sends out early warning information.Aiming at the requirement of stateless connection,the HTTP service is designed and implemented by Java to meet the connection needs of clients under different systems and realize the separate development of front and rear ends.According to the implementation of related business logic,design related data structure and IOCP event resolution mechanism to improve the performance of server software.3.The user client of the system is designed and realized,and the Android mobile APP and Web page are developed to realize the visual display of the collected data,the remote control of the acquisition terminal devices,the reception of the early warning information and the management of the system operation.And the problem that the user can not access the system anytime and anywhere and receive the warning in time in the traditional monitoring software is solved.4.The pest detection and identification algorithm is designed,optimized and realized,and related algorithms were investigated.Pest detection and identification is realized based on Faster-RCNN and PCA feature dimension reduction,and self-owned data sets are used for training.Finally,according to the actual deployment of engineering optimization,the accuracy and availability of the algorithm are balanced,and the precision rate is 86.27%,and the recall rate is 85.21%.5.The function and performance of the system are tested as a whole,the field test is carried out,and the relevant performance parameters are given when the system is in normal use. |
