An Acoustic System For The Detection Of Sugar Content And The Hollow In Watermelon

The sugar content and hollow heart of watermelons was detected with the acoustic system, which was set up with the acoustic non-destructive technology in home and abroad. The main contents and achievements of this research were as follows:Firstly, the former acoustic detection system was improved in this research, including a plastic striking ball, six accelerate sensors, a set of signal processing circuit, an electric amplifier (YE5853B), a computer with a data acquisition board PCL-1800 (Advantech Co, Ltd.) and a set of software including the data acquisition window and the signals analysis procedure specially designed for the system. The accelerate sensors can get better acoustic signals than that of microphone without being disturbed by the noise in the atmosphere. The data acquisition software are brief and easy to handle. Sampling frequency was 10KHz, MATLAB, TQ were used for the data analysis. The system was proved all right for watermelon detection.Secondly, a new acoustic parameter, the acoustic transmissivity (8) at frequencies were used in the research. The number of the parameters from 1024δand the different tapping-receiving groups were disused with the help of MATLAB software. Then,20 transmissivities were used for testing models, and the equator part can had better results.Thirdly, with the help of the SMLR function in TQ software, the relationship between the number of parameters used in the model and the R, RMSEP, RMSEC were discussed respectively. Different tapping-receiving points were compared with each other, and 6 characteristic frequencies were chosen for setting models of watermelon sugar content:y=8.03+0.39δ1-1.18δ2+0.82δ3-0.09δ4-0.17δ5+0.0986, by tapping the equator part and receiving the signals from the opposite side with watermelon being put in its growing position, the R value was 0.80753, RMSEC was 0.646, RMSEP was 0.655.Finally, the stepwise multiple linear regression (SMLR) and the (discriminant analysis) DA statistic method in the software TQ were used for detection of hollow heart in watermelons. By comparing the prediction rates of different tapping-receiving groups, the accuracy of SMLR was higher than that of DA, of 0.8.