Carrot Texture Evaluation Based On Fracture Acoustic Signal

Food texture is the main index of food detection and consumption. Food texture evaluation methods include subjective evaluation and objective evaluation. Due to the subjective factors, sensory evaluation has some defects. So it has been the research hot spots to explore the objective and reliable analysis methods for texture evaluation.The purpose of this paper is to evaluate the carrot texture based on fracture acoustic signal during the process of texture testing. Fracture acoustic signals were analyzed by wavelet decomposition and Hilbert transform theory, acoustic features in time or frequency domain were extracted. Relationship between the acoustic characteristics and the food texture was analyzed. Then texture evaluation models based on carrot acoustic signal were built and verified. Meanwhile, evaluation mechanism of fracture acoustic signals was explored through the changes of main nutrients and structure in carrot. This research can provide theoretical support for the application of acoustic signal technology in food texture evaluation, provide a new method for the detection of food quality. The main research contents in this paper are as follows:(1) Acquisition and analysis method of the carrot fracture acoustic signal were determined. The released acoustic signal was collected by acoustic collection system including computer and acoustic sensor during the compression of texture analyzer,compression and shearing fracture model were used respectively. Wavelet theory was used to eliminate the noise of the fracture acoustic signal. Time domain analysis theory was applied to determine extract method in time domain characteristics mainly including four indicators, waveform index, sound intensity, maximum amplitude and amplitude difference. Hilbert-Huang transform theory was applied to determine extract method in frequency domain characteristics, the characteristic was the frequency range of relative energies.(2) Relationships between the acoustic signal and the crispness and firmness of the carrot under compression fracture mode were studied. The compression probe was TA-4 cylindrical probe, the sample diameter was 10 mm, and the height was 12 mm.The results showed that carrot crispness had significant positive correlation with waveform index(r=0.785, P < 0.01) and sound intensity(r = 0.732, P < 0.01) of acoustic signal characteristics in time domain; had no significant correlation with the frequency characteristics. Carrot firmness had no significant correlation with characteristics in time domain or frequency domain.(3) Relationships between the acoustic signal and the crispness and firmness of the carrot under shearing fracture mode were studied. The probe type was selected TA7 type probe, sample of the cuboid size was 30×20×10 mm(length×width×height). The results showed that carrot crispness showed significant positive correlation with waveform index(r=0.723, P < 0.01) and sound intensity(r=0.725, P < 0.01) of acoustic signal characteristics in time domain; showed significantly negative correlation with relative energy of high frequency(r=-0.676, P < 0.01) in frequency domain values. Carrot firmness showed a highly significant positive correlation with waveform index(r=0.722, P < 0.01) and sound intensity(r=0.806, P < 0.01) of acoustic signal characteristics in time domain; showed significantly negative correlation with relative energy of high frequency(r=-0.726, P < 0.01) in frequency domain. Compared with the compression mode, the relative energy in high frequency,wave form index and sound intensity in time domain all showed significantly correlation with carrot crispness and firmness under shearing fracture model.Therefore the acoustic signal under shearing fracture model can be used to evaluate the carrot texture.(4) Carrot texture evaluation models based on the acoustic signal under shearing fracture mode were constructed and verified. Principal component algorithm was used to reduce dimensions of the acoustic signal characteristics in time domain and frequency domain. Two principal components were obtained, principal component 1was named frequency-domain characteristics, principal component 2 was named time-domain characteristics, and the effective information retention rate of two principal components was up to 75.47%. According to two principal components, the multiple linear regression model for evaluating carrot texture was constructed,average relative error of crispness prediction was 4.11%, the average relative error of firmness prediction was 1.28%. Through the back propagation neural network(BP NN) algorithm, neuron number of hidden layer and the training function parameters were studied. The back propagation neural network model was determined, the topological structure of model was [6-12-2]: the number of input layer neurons is 6,the hidden layer neuron number is 12, the number of output layer neurons is 2. The average relative errors of crispness and firmness were less than 3%. The results showed that the multiple linear regression model can basically evaluate the carrot texture, but the precision is not high, the BP neural network can evaluate the carrot texture slightly better than the multiple linear regression model.(5) Exploration of the sound signal of carrot texture evaluation mechanism was conducted. With the extension of storage time, carrot crispness, firmness, sound signal waveform index and intensity showed downward trend, and the differences in different storage period were obvious(P < 0.05), the total energy of frequency during storage were not significantly different. The three main nutrient contents including water content,reducing sugar and pectin under different storage time were determined by national standard detection methods. The results showed that the water content, pectin showed downward trend with extension of storage time, but the reducing sugar reflected to certain dynamic changes of carrots, increased first and then decreased, the sample after 14 days storage had no significant difference. The ultra-micro structure of carrot under different storage time was observed by scanning electron microscopy. The results showed that fresh carrot cell structure was compact and orderly arrangement,there is no obvious fracture zone, with the extension of storage time, cell wall cellulose polymers became loose, cell structure fracture was appeared after storage 14 days, structure fractures were more serious after storage 28 days.The results defined the evaluation mechanism of carrot texture by acoustic signals:Three main nutrient contents including water content, reducing sugar and pectin declined with the extension of storage time. These changes can influence the carrot cell structure. If the structure changed, the mechanical properties of carrot cells would change too. The changes of mechanical properties may result in decreased of carrot crispness and firmness gradually. Due to these changes, the mechanical characteristics of fracture occurred significantly changes. Once the fracture force declined, the acoustic signal released by fracture energy changed. These results proved that evaluation of carrot texture by acoustic signals is scientific and reliable.In summary, carrot texture evaluation based on the acoustic signal of fracture can provide a reliable theoretical basis for the application of acoustic signal in food quality detection. Characteristics in time or frequency domain extracted by time domain analysis and Hilbert-Huang transform can clearly describe the acoustic signal useful information during fracture. Theses results can provide necessary theoretical basis for the promotion and application of acoustic technology in food quality evaluation and detection.