Abstract: In order to achieve the optimization of the feeding amount of group sows and the remote monitoring of feeding condition, so as to control the body condition of the capable sows, improve the reproductive capacity, and alleviate the problem of the sharp decline in the number of live pigs in the past year, an intelligent feeding Internet of things system for group sows based on Internet of things and sow feeding expert system was designed. The terminal device obtained the ear tag number and the temperature and humidity of the pig house in real time. The data packets of each child node were collected at the aggregation node through the LoRa star network, and then uploaded to the cloud server by the GPRS module. The monitoring program of the upper computer deployed in the cloud can view the identity information of the sow, feeding information and environmental temperature and humidity, and can query relevant historical data through the database. The expert system was developed based on the Doorls rule engine and inferred the optimal daily feed amount for the sow by automatic matching rules of Rete algorithm. The upper computer feeds back the decision results to the control module, and the feeder feeds accurately. The average relative error of feed volume is 1.10% through the inspection, which meets the demand of high-precision feeding. Compared with the sows in the limited rail and in small pigpen, the average feed intake of pregnant sows fed by this system was closer to the reasonable value, the proportion of sows whose weight and backfat thickness controlled within the reasonable range was significantly higher, and the average number of healthy piglets per birth is increased by about 0.3. The results indicated that the system had high transmission accuracy and good real-time performance, which met the practical application requirements of large-scale pig farms.