Research On Intelligent Early Warning System For Scale Breeding Based On Improved Neural Network

A strong country must first strengthen agriculture,and only by strengthening agriculture can the country be strong.With the development of China’s animal husbandry,safety issues are becoming increasingly prominent and must be taken seriously.The prominent issues in modern aquaculture include the quality and safety of products and the production safety of aquaculture.In the process of breeding,manual labor is cumbersome and vague,which cannot accurately grasp the environmental parameters,affecting the quality and safety of products;In addition,fire prevention can not be ignored.Fire incidents in aquaculture are also common.Once the fire spreads,aquaculture users will suffer irreparable losses.Therefore,whether early detection of fire can be achieved is the most critical link to reduce fire losses.In order to detect the occurrence of fires as soon as possible and accurately,and to provide timely warning of fires;This paper proposes an intelligent early warning system for large-scale aquaculture based on an improved neural network to accurately grasp the environmental conditions parameters during the aquaculture process and provide early warning control for excessive parameters.The main research work is as follows.(1)In terms of fire warning algorithm design,the structure and basic principles of BP neural network algorithm and genetic algorithm were discussed,and the BP neural network algorithm optimized by BP neural network algorithm and genetic algorithm was used for fire warning.Due to the shortcomings of traditional BP neural networks such as difficulty in convergence and tendency to fall into local minima,a neural network improvement algorithm based on a combination of backpropagation algorithm and genetic algorithm is proposed to improve the structure and algorithm of the neural network.This algorithm introduces genetic algorithm into traditional BP neural networks,and improves the training speed and generalization ability of the network by optimizing the initialization and selection of weights and thresholds,thereby achieving more accurate fire warning.(2)In terms of fire monitoring and early warning design,smoke sensors,temperature sensors,and CO sensors are selected,and then fire detection nodes,fire recognition integrated machines,and fire alarm linkage devices are designed from both hardware and software aspects.Finally,software design is carried out for the upper computer.(3)In terms of environmental monitoring and early warning design,this part is based on Internet of Things technology and can complete functions such as temperature and humidity monitoring,carbon dioxide monitoring,intelligent ventilation,intelligent cooling,intelligent light supplementation,and timed flushing,achieving precise mastery of environmental parameters and early warning control.In terms of perception layer,select temperature and humidity sensors,carbon dioxide sensors,and photosensitive sensors,and design hardware and software for STM32 controllers and alarm linkage devices;In terms of network layer,Wi-Fi network is used to transmit data through the local area network to Alibaba Cloud servers for data analysis and processing,achieving remote monitoring and management of breeding environment conditions;In terms of application layer,the front-end visualization interface is designed using Java language to display temperature and humidity,carbon dioxide content,lighting intensity,and status information of lighting,stepper motors,fans,and water pumps.Finally,test this section.In summary,the intelligent early warning system for large-scale aquaculture based on improved neural networks has the characteristics of real-time,accurate,and reliable,which can effectively reduce the risk of fire and product quality and safety issues during the aquaculture process,improve aquaculture efficiency and product quality,and promote the high-quality development of China’s aquaculture industry.