Signal processing method of impact-based grain flow sensor for predicted yield

Abstract: Background vibration noise of combine harvester has a significant influence on measuring accuracy of impact-based grain flow sensor. An indoor calibration test bench of impact-based grain flow sensor was developed to study extraction and rejection methods of the background vibration noise from output signal of impact-based grain flow sensor. The test bench was mainly composed of a grain supplying bin, a grain conveying auger, a drag-flight elevator and a grain receiving bin. The auger and the elevator were driven by a three-phase AC asynchronous motor. Grain flow rate was regulated by a spile plate which was mounted at the bottom of the grain supplying bin. Actual grain flow rate was measured with three weight sensors which were mounted between the grain supplying bin and the frame. Revolution speed of the elevator was measured with a revolution speed transducer. Different field working conditions of combine harvester were simulated by adjusting power frequency of the driving motor with a variable frequency generator. A dual-plate differential impact-based grain flow sensor was used in the study which consisted of a measuring plate, a reference plate and 2 strain bridges. The measuring plate accepted impact of the grain flow, and the reference plate sensed background vibration of combine harvester. The measuring plate and the reference plate had a same structure and were parallel mounted to make them have an approximately same mode of vibration. An industrial control system was used to control the variable frequency generator and to acquire output signals of the measuring plate, the reference plate, the weight sensors and the revolution speed transducer. Signal processing was also performed on the industrial control system. Output signals of the measuring plate and the reference plate were acquired synchronously with 3 kHz sampling frequency, and mean filtering was performed to the signals respectively to attenuate random noise. DFT (Discrete Fourier Transformation) was executed to the filtered signals respectively, and the transformed signal of the reference plate was subtracted from the transformed signal of the measuring plate, and IDFT (Inverse Discrete Fourier Transformation) was executed to the differential signal afterwards, thereby frequency-domain differential was fulfilled and yield signal was obtained. Compared with the time-domain differential results the yield signal obtained through the frequency-domain differential method had a less standard deviation. Experiment under different power frequencies of the driving motor showed frequency-domain differential method had a steady effective attenuation effect on background vibration noise rejection for output signal of the dual-plate differential impact-based grain flow sensor. Zero calibration and scale conversion calibration were carried out on the indoor calibration test bench, and yield test experiment under different power frequencies of the driving motor and different actual grain flow rate were performed on the indoor calibration test bench. Yield test experiment results showed the dual-plate differential impact-based grain flow sensor with frequency-domain differential signal processing method had a maximum measuring error of 3.1% with the actual grain flow rate in the range of 0.9 to 2.3 kg/s.