Abstract:An adequate and consistent depth positioning of seeds is vital for uniform crop germination to achieve the optimum yield of agricultural crops. However, the downforce variations from the row units will affect the stability of sowing depth because of the irregular and inconsistent soil resistance of the seedbed. Therefore, controlling the seeding downforce to compensate for changes in soil resistance can improve seeding quality. At present, most of the downforce control methods are driven by hydraulic pressure, which requires a high level for the tractor hydraulic system. In addition, previous studies have found that the existing downforce detection methods have problems of low sensitivity and lack of fast and precise control model, which can not achieve real-time accurate downforce control. To solve the problems, a new downforce control method based on the air-spring pressure and the four-link angle was proposed, and a corresponding pneumatic downforce control system was designed. The system consisted of pneumatic driving device, tilt sensor for profiling mechanism, pressure sensor for air-spring, downforce sensor for gauge-wheel, date acquisition and control module, and an upper computer. The pneumatic driving device, which mainly included air-spring, electric-gas proportional valve, air pump, gas tank and oil separation filter, was used to provide the necessary downforce on the profiling mechanism to ensure the optimum and consistency of sowing depth. The downforce sensor and tilt sensor were applied to generate downforce and the four-link angle signals in real time. After first-order low-pass filtering and model calculation by the upper computer, these actual downforce was displayed on the interface programmed by LabVIEW and the control signals were sent to the electrical-gas proportional valve through the date acquisition and control module based on RS485 communication. A modeling experiment was conducted to establish the relationship between the sensor values and the actual downforce under different air-spring pressures and four-link angles. Regression analysis showed that the model fitted the best, being 0.9743 in adjusted determination coefficient (R2Adj) and 49.41N in root mean square error (RMSE). The verification test showed that the predicted root mean square error (PRMSE) was 39.51N, which showed that the model had better control accuracy for downforce. Further, an air-spring response test and a field test were carried out respectively to test system control performance. The results showed that the air-spring inflation step response average overshoot was 3.83%, the average steady state error was 0.0052MPa, and the average adjustment time was 0.42s when the pressure was set in the range of 0.1~0.6MPa. The field tests indicated that the system had stable and reliable control performance for sowing depth in the speed range of 6~10km/h. Within the industry standard error range of 10mm, the qualified rate of sowing depth of the system was not less than 98.91%. Especially when the speed of the planter was over 10km/h, the standard deviation (SD) of sowing depth was 3.46mm and the coefficient of variation (CV) was 6.97%, which was significantly better than the passive downforce control system with the SD of 6.70mm and the CV of 13.07% respectively.
