Development Of Impact-Based Grain Yield Monitor Sensor And Its Experiment Stand

Acquisition of yield information is an important part of the precision agriculture. And, the grain flow sensor is a core unit of the yield monitor system. However, the impact-based grain flow sensor could be enormously affected by the vibration noise. So, a yield monitor experimental platform was constructed based on JOHN DEERE-3080combine. Then, some studies such as scheme selection of the stand design, structure selection of the flow sensor and methods of data processing were conducted to reduce the vibration influence. The main research results were as follows:(1) The experimental stand was constructed by transforming the JOHN DEERE-3080combine. Then, the working vibration of combines in the field could be well simulated. The actual grain flow rate, moreover, could be obtained for yield monitor calibration by measuring the weight of grain tank of experimental stand in real time. And, two kinds of flow sensors with dual parallel beams and single parallel beam were designed.(2) A data acquisition system for the experimental stand was designed using flow sensor, weighing sensor, rotate speed sensor, acceleration transducer, data acquisition card and so on. The software of the data acquisition system was programmed using LABVIEW to synchronously acquire and memory multipath signals.(3) Vibration noise elimination for the flow sensor with dual parallel beams was studied. The frequency component of the vibration noise non-overlapping the harmonic frequency was eliminated by the harmonic extraction method, while the noise overlapping the harmonic frequency was eliminated by the adaptive interference cancellation method. The experimental results showed the relative error was less than1.6%. And, the effectiveness of the vibration noise cancellation method was verified.(4) Vibration noise elimination for the flow sensor with single parallel beam was studied. The BP neural network was used to identify the vibration system of the parallel beam. Then, the vibration noise in the output of the parallel beam could be obtained by the acceleration signal. A three layered BP neural network with the100-201-1structure was selected using the trial method. The experimental results showed the relative error was less than3.2%. Good measurement accuracy was obtained on the condition that the structure of flow sensor was well simplified.(5) The influence of the installing angles to the impact-based flow sensor was studied. Three different angles were selected for experiment to study the influence on the flow sensors with dual parallel beams and single parallel beam. The experimental results showed installing angle had a significant impact on the impact-base flow sensor.