Registering feeding pigs in group-housing

A High Frequency Radio Frequency Identification (HF RFID) system was designed for registering feeding pigs. The system consists of transponders, antenna(s) and reader(s). The RFID antennas were installed on commercially available round feeders for growing-finishing pigs. Eight of these feeders were placed in four pens with partially slatted floors. In each pen 59 growing-finishing pigs were housed with an RFID transponder in their ear. In one pen, RFID ear tag transponders were inserted in both ears of the pigs. When a pig’s transponder comes close to the RFID antenna and thus to the feeder, the transponder’s unique identification code is registered. The RFID reader ensured data communication between the RFID system and the PC and decoding of the transponder messages. It was determined whether the system was able to discriminate between pigs that were feeding and pigs that were not feeding. This was done both online (using feeding pigs) and offline (using stationary transponders). For the online validation 20 pigs (from the pen with 2 transponders per pig) were marked and observed during 11.5 hours. The percentage of RFID registrations within observed feeding visits was 77.11 % and 92.23 % of the RFID registrations occurred within 10 s from a feeding visit. The percentage of feeding visits containing RFID registrations was 89.17 %. To investigate the possible causes for the missed feeding visits an offline validation was performed where stationary transponders were placed in several orientations and on several heights under the antenna. The range of the RFID system was measured. The range of the system covered the feed trough well. However, the orientation of the transponders relative to the antenna had an influence on the registrations and for some orientations, the registration range was small. From this observation, it was inferred that the movement of the pigs’ ears during feeding can bring the transponders out of the registration range, thus creating irregular gaps between registrations. However, this problem can be solved by proper data-analysis where the RFID registrations are clustered into feeding visits and meals to determine the feeding patterns of the pigs. By clustering the registrations using a time window of 9 s, the sensitivity and specificity of the system were 88.58 % and 98.34 %, respectively. Changes in feeding patterns can then be used to signal problems with the pigs’ health, welfare and productivity in an online way.