A novel microwave Doppler method for mass estimation of machine-picked seed cotton

Obtaining information on cotton yields and creating a spatial map of cotton production is the starting point for implementing precision agriculture. Microwave technology has been used for yield estimation in China for crops such as corn and soybeans and has shown strong applicability. However, research on cotton yield estimation using microwave technology is weak. Therefore, this paper proposes a method for seed cotton mass estimation based on the microwave Doppler method and designs bench tests to prove the feasibility of the method. First, a seed cotton speed estimation model was established based on the microwave Doppler effect and fast Fourier transform. On this basis, a seed cotton velocity estimation experiment was designed to study the effects of sensor installation angle and position, pipe collision friction, air resistance, and the interaction force between the seed cotton on the seed cotton movement velocity. The power spectral density (PSD) of the acquired microwave data is then estimated using the modified periodogram method. The least squares method was used to establish a linear regression model between seed cotton mass and the corresponding PSD, and finally the free-fall mass experiment and pneumatic conveying mass experiment yielded that the estimation errors of seed cotton mass were 7.30% and 6.17%, respectively. The experimental results show that the estimation error is small when using microwave technology for seed cotton mass estimation, which meets the requirements of actual operation.