Research On Rapid Measurement Method And Device Of Apple Hardness Based On Acoustic Characteristics

Apple's edible quality and economic value are closely related to apple's internal quality,which is the main index of apple quality testing and an important reference basis for consumers to choose apple.Hardness has an important impact on the post-harvest quality testing and grading of apple.Therefore,it is of great significance to study the hardness prediction method of apple for its production,harvest,processing and other related processes.The hardness evaluation methods of apple include sensory evaluation and objective evaluation.As sensory evaluation is easily affected by subjective factors,it has some defects.The existing objective evaluation methods are mainly destructive test,which is harmful to apple and inefficient.Therefore,it is of great significance to seek a rapid nondestructive testing method for apple hardness.In this paper,red Fuji apple is selected as the research object to predict the hardness of apple based on acoustic characteristics.In the research process,the sound signal is preprocessed,fast Fourier transform and power spectrum analysis,etc.to extract the corresponding time domain and frequency domain features,and analyze the correlation between the sound signal features and apple hardness.An apple hardness prediction model based on acoustic characteristics was established.This research can provide support for the application of sound signal technology in quality evaluation of agricultural products and provide a new method for nondestructive testing of agricultural products.The main research contents and conclusions of this paper are as follows:(1)This paper firstly analyzes the prediction principle of apple hardness based on acoustic characteristics.The acoustic characteristics of agricultural products change with the internal organization of agricultural products,and the acoustic characteristics of the same kind of agricultural products with different internal quality are often different,according to the acoustic characteristics of agricultural products can be judged how its quality.Therefore,acoustic technology can be applied to the research of nondestructive testing and grading of hardness of agricultural products.(2)According to the acoustic testing principle and the physical characteristics of the apple itself,an apple hardness testing device is designed and manufactured.The device can collect the sound signal of apple hardness prediction,and adjust the relative position of hitting plate and hitting shaft to adapt to the hardness testing of apples of different sizes.The optimal position of the acoustic signal sensor and the detection of impact force are determined.(3)The acquisition and analysis method of apple hardness prediction sound signal is determined.Using the test device to hit the apple,the sound signal is collected by the sound signal sensor,the sound signal is preprocessed,and the fast Fourier transform is used to transform the time frequency of the signal.To extract the sound signal characteristics,including: the maximum amplitude,amplitude and waveform symmetry,resonance frequency,power spectrum peak,mean,standard deviation of the time domain,time domain signal energy,RMS of the mean frequency domain,frequency domain and frequency domain variance,standard deviation of frequency domain,frequency domain signal energy,center frequency,frequency variance maximum length,total power,power spectrum of signal energy and time domain RMS,etc.The correlation between the characteristic value and hardness was analyzed.(4)The prediction model of apple hardness based on acoustic characteristics was built and verified.Stepwise regression method was used to establish a multiple linear regression apple hardness prediction model for the extracted characteristic values,and 10 apple samples with different hardness were randomly selected for prediction.The average accuracy of the verification model was 78.14%.Neural network apple hardness prediction model based on principal component analysis,neural network apple hardness prediction model based on stepwise regression and neural network hardness prediction model based on all characteristics are established respectively.10 apple samples were selected to verify the models,and the results showed that the average errors of the three models were 75.14%,92.06 percent and 73.96% respectively.By comparing the four prediction models,the conclusion is drawn: the neural network prediction model of apple hardness based on stepwise regression has the highest average accuracy,and the prediction trend of hardness is closest to the actual change trend of apple hardness,which is the best model.Design is suitable for many occasions,many varieties of apple based on the acoustic properties of signal collecting device and hardness prediction,The relationship between acoustic characteristics and apple hardness is studied and the mathematical model of apple hardness prediction based on acoustic characteristics is established.The fast and nondestructive testing of apple hardness is realized.The application of acoustic nondestructive testing technology in agricultural products is realized.It provides technical support for better testing of apple quality and accurate and rapid classification after apple picking.