Research And Design Of Environmental Monitoring System For Piggery Based On Internet Of Things And Convolutional Neural Network

China’s rapid growth of population and people’s pursuit of higher quality of life,the urgent need for more livestock breeding quantity and higher meat quality.As a big pork consumption country,China has a huge demand for pigs,and intensive and largescale pig breeding can effectively meet people’s pursuit.Traditional large-scale farming needs higher costs,including human costs,garbage disposal and so on.With the increase of the number of pigs,it is difficult to strictly control the feeding state of pigs in daily work.The use of sensors,cameras and other equipment can effectively solve the feeding cost.In the farm deployment sensor collection breeding environment,combined with the Internet of things technology and web technology can real-time control the parameters of pig breeding process,in the active and effective management conditions,can solve the problems of pig feeding before the problem occurs.And in the daily management process,the use of face recognition technology can accurately and easily carry out employee attendance,work status recognition(on-thejob or not),and personnel access management,in the system access to the public security system,can effectively carry out security early warning.In addition to the identification of people in the breeding process,the identification of pigs is also very necessary.A kind ofThe main research work of this paper is as follows:(1)This paper analyzes the demand of intensive pig house breeding detection system under the intensive production and breeding mode;selects carbon dioxide,ammonia,temperature and humidity indicators as the monitoring objects of the system;selects low-energy,low-cost,high-precision sensors and Zig Bee and other technical components wireless local area sensor network.(2)The algorithm and model of face detection and recognition are studied.The multi-level convolution neural network is used to detect the input image samples,and the improved RESNET network is used to extract the features.The application of face recognition technology in the monitoring system greatly facilitates the user’s login operation,and has considerable scalability.In the later function expansion,the replacement and increase of equipment sensors,the setting of early warning information,security and other aspects can deploy face recognition.(3)The traditional pig identity authentication is mostly realized through the invasive operation(electronic ear tag,etc.).The essence of image recognition is to extract the image features,and then match the features to get the recognition results.Through the exploration of face recognition,the optimal face recognition scheme is applied to the pig identification scheme.The non-invasive identification of pig samples can greatly facilitate pig identification in intensive and large-scale breeding process.(4)Based on the requirement analysis of piggery environmental monitoring system,My SQL is selected as the system,and the database design is completed.The system is designed and developed based on J2 EE framework,combined with HTML,JS,CSS front-end technology and so on.The system is divided into two parts: the management side and the employee side.It realizes the functions of account password login,face recognition login,pig identity authentication and environmental monitoring.The pig breeding monitoring system designed by this system has rich functions,high stability and comprehensive functions.It can carry out diversified management of the pig house environment.The way of separating the management authority of the employees can effectively control the employee information,and the employees can set the environmental parameters warning according to the needs.