Research And Application Of Sow Intelligent Feeding System Based On Internet Of Things And Expert System

As a major pig breeding country,China has always been struggled with low production efficiency and high breeding costs.Nowadays,digital agriculture is injecting new vigor and vitality into rural revitalization.Advanced technologies such as the internet of things and artificial intelligence have spawned a new type of breeding industry,and also called for the arrival of an era of industrial innovation.Reproductive sows are the core productivity of pig farms.The feeding and breeding level is directly related to the health status and reproductive performance of sows,which in turn affects the breeding efficiency of the pig farm.However,most domestic small and medium-sized pig farms still use the traditional mode of restricted corrals and manual feeding.The traditional mode not only increases the operating cost,but also makes it difficult to accurately realize the difference in feeding level between different sows,which is not consistent with the concept of precision feeding.In response to the above-mentioned drawbacks in sow feeding and breeding,the internet of things and artificial intelligence technologies have been introduced to research,and a sow intelligent feeding system based on the internet of things and expert systems has been implemented.For sow feeding amount and feed grade issues,it can quickly provide expert solutions,and control the on-site equipment in real time on the remote server platform.Therefore,the pig farm can truly achieve expert management and reduce the cost and increase the efficiency of feeding and breeding of sows.The specific work content and research results are as follows:(1)Build a three-tier architecture for the sow feeding internet of things system,including information collection and control equipment of the perception layer,aggregation nodes of the network layer,and remote servers and Android clients of the application layer.It not only provides a platform for breeders to remotely monitor the sow feeding situation and environmental information,but also provides real-time and accurate data for the expert system to improve the reliability of its decision-making.(2)In response to the problems of rough feeding,limited feed type and poor welfare of sows under the traditional feeding mode in restricted corrals,a new feeding mode combining dynamic group feeding of sows and mixed feeding of double feeding lines was proposed.(3)A composite expert system based on CLIPS and ELM was developed by using the superiority of CLIPS development environment and the better classification performance of ELM model.Intelligent decisions are made on the most suitable feeding amount and the most suitable feed grade for sows.And based on the internet of things system,the feeding decision conclusions can be executed in real time.Meanwhile,the network structure of Rete pattern matching algorithm in CLIPS inference machine was optimized to effectively improve the inference efficiency.According to the simulation experiment results of ELM model,the average prediction accuracy reached 92.7% when the number of neurons in the hidden layer was 78.(4)In order to verify whether the intelligent feeding system has practical application effects,a field test was conducted in a pig farm in Yixing,Jiangsu.The results showed that the feeding accuracy and stability of feeding equipment met the actual use requirements.The average daily feed intake of gestating sows reared in this system was more in line with the feeding rules during gestation than that of sows reared manually in restricted corrals.When sows were transferred to this system,their reproductive performance was better than that of the restricted corrals period,with an increase of 4.55% and 2.78% in the number of healthy piglets and litter weight of piglets,respectively.The double feeding line model for group sows has obvious advantages in feed utilization,saving feed expenses and labor costs while meeting the nutritional needs and welfare of sows.In summary,combining the new feeding mode of sows with internet of things and expert system,the system is dedicated to building an intelligent feeding system for sows with comprehensive perception as the premise,real-time transmission as the guarantee,expert decision-making as the core and automatic control as the goal.The reliability and practicability of the system has been verified.It is beneficial to achieve the fundamental purpose of improving the reproductive performance of sows,reducing operating costs of pig farms,and expanding breeding benefits.