Development and denoising test of grain combine with remote yield monitoring system

Abstract: Since grain yield in farmland has spatial variability, and the size of production can reflect the growth and management situation of grain, it is necessary to obtain accurate information on spatial distribution of production for implementing precision agriculture. However, it is still lacking of yield monitoring systems that are suitable for grain combine harvester and field conditions in China. The current developed systems in China mostly cannot reduce the vibration from the harvester, and tend to produce a large error in dynamic measurement of production. Therefore, in this study, a new type of intelligent grain yield monitoring system was developed in order to minimize the influence of the field vibration on accuracy of grain yield monitoring system and improve its practicality. The system included a remote monitoring subsystem based on computer networking technology and a vehicle-mounted subsystem based on controller area network (CAN) bus technology. The remote monitoring subsystem could realize on-site yield measurement, yield mapping, remote monitoring and harvest management. The vehicle-mounted subsystem consisted of industrial computer, CAN bus module, general packet radio service (GPRS), GPS receiver module and a variety of signal sensors. It could detect grain yield, generate yield map and remote wireless communication. Meanwhile, it collected impulse sensor data, elevator shaft speed, grain moisture, harvester travel speed and cutting width to establish mathematical model and measured the grain yield accurately. In addition, it also could get information on geographical location from GPS receiver to draw grain yield distribution map. Moreover, through the GPRS network, it sent the data to a remote personal computer (PC) for processing and displaying. The vehicle-mounted subsystem here adopted mechanical denoising method and double plates differential method to reduce the influence of harvester vibration on measurement accuracy, but the minor differences in output signals between pre-plate and rear-plate of the impact sensor could be observed, which might be caused by difference in installation location of the two plate bracket in fixed end distance and the different force on the sensitive beam resulted from mechanical vibration of the combine. For this reason, a regression difference method was proposed, by which the vibration signal of rear plate approximated the vibration signal of first board before difference processing. In the subsystem, digital threshold filtering was used to improve the estimating accuracy of grain yield, and the filtered data was used for fitting mathematical models of total yield and yield of per unit area. Field test results showed that by regression difference method, the average error of the yield estimate was 3.27% and the maximum error was 8.03%, which was reduced by 7.12% compared with the direct difference method. It suggested that the regression difference method was superior to the direct difference method in eliminating vibration interference. The remote subsystem developed a friendly interface, which realized the remote monitoring and managing grain harvest. The system had a good performance to meet the needs of yield measurement in China.