Wheat Hardness Of Determination Method Of Acoustic Ptimization Research

The merits of wheat quality directly affect national food quality, so the research ondetection methods of wheat quality has important theoretical significance and practical value.Wheat hardness plays an important role in wheat processing quality. One of methods fordetecting wheat hardness is the acoustic method, which has the characteristics ofnondestructive and convenient. This paper is applied to optimization study on wheat hardnessacoustic measurement method with drop height, sampling frequency and sampling device cost,aiming at analyzing the differences among wheat acoustic signals with different samplingfrequencies and different drop heights, and acquiring the relationship between the wheatacoustic characteristics and its hardness index. Based on this, we can further optimize theexisting acoustic detection method.During the detection signal acquisition process, different sampling frequencies anddifferent drop heights of the acoustic detection signal are obtained separately, and then we usedifferent signal processing algorithms to analyze the characteristics of the acoustic detectionsignal in the time domain and frequency domain. Next the signal features are extracted tomake a correlate analysis with wheat hardness index. Namely, in the time domain, wheatacoustic signal characteristic parameters with different sampling frequencies and differentdropping heights are extracted, such as the average absolute amplitude TF3, energy TF4andpulse factor TF6; In the frequency domain, we extract the characteristic parameters of FFT1and FFT2based on fast Fourier transform, DCT1and DCT2based on discrete cosine transform,WT1and WT2based on the wavelet transform, DF1and DF2based on wavelet transform anddiscrete cosine transform, DT1and DT2based on wavelet transform and fast Fourier transformin different sampling frequency and drop height wheat acoustic signal respectively. Studieshave shown that, whether the time domain or the frequency domain, when the samplingfrequency is200kHz and drop height is40cm, the signal characteristics have better correlationwith wheat hardness index, and the correlation coefficient is0.95. Finally, the wheat hardnessacoustic detection model is built through linear regression analysis and neural network technology respectively to analyze the wheat hardness index test with different samplingfrequencies and different drop height. Experiments show that the prediction model based onDT2and neural network has better performance as wheat hardness acoustic measurementmodel, where its maximum absolute error is0.21and the average relative error is0.0051%.Through these studies, this paper can not only improve the precision of acoustic detectionmethod and reduce the cost of measuring instruments, but also provide technical support forthe application of wheat hardness acoustic method in the future.