Study On Dough Fermentation Monitoring And Quality Assessment Of Chinese Steamed Bread By Intelligent Sensor Technology

Chinese steamed bread(CSB),as a typical and traditional Chinese staple food,has a long history of consumption.However,its processing methods are extremely traditional,and process control and quality assessment also rely on manual methods.Therefore,it is great practical significance to explore the intelligent methods for its process monitoring and quality assessment.Dough fermentation is the crucial link in the processing of CSB.This study intends to systematically analyze the dynamic characteristics of dough during the fermentation process of CSB.And select the characteristic indicators that can accurately reflect the fermentation state.Then establish the intelligent monitoring method of dough fermentation state combining with the corresponding sensor technology.When performing artificial sensory assessment on the quality of CSB,the interaction of various sensory information will affect the objectivity and accuracy of the results.In addition,the cognitive psychology method is used in this study,to investigate the interaction between sensory information.And design a sensory assessment model of CSB to improve the objectivity and accuracy of the results.Eventually establish the intelligent assessment method for CSB sensory quality using multi-sensor technology based on the designed assessment model.The main contents of this research are as follows:(1)Research on the dynamic characteristics of dough during the fermentation process of CSBScreening out suitable indicators that can accurately reflect the dough fermentation state is the crucial step for the establishment of intelligent monitoring method for dough fermentation.This study conducted a systematic study on the dynamic changes of the moisture distribution,pasting property of starch,gluten structure,free amino acids,internal structure of dough,gas production characteristics and other indicators during the dough fermentation process.The results showed that there are 8 characteristic peaks on the Fourier infrared spectrum of the gluten,they changed significantly during the dough fermentation(p<0.05).And the time nodes of significant changes were similar to those of different fermentation stages.In the insufficiently fermented and moderately fermented stages of dough,the gas production increased significantly with the fermentation time(p<0.05);when the dough was over-fermented,the dough began to leak,and the leakage showed a significant increase trend(p<0.05).Therefore,the structural changes of the gluten and the gas production characteristics were the suitable indicators could reflect the characteristics of different dough fermentation stages.Based on those two indicators,the appropriate sensor technology can be selected to study the intelligent monitoring method of dough fermentation.(2)Research on the monitoring method of dough fermentation based on electronic noseBased on the measurement results of the volatile components of the fermented dough with Gas chromatography-ion mobility spectroscopy(GC-IMS)technology,an intelligent monitoring method of dough fermentation was established using electronic nose technology.Principal component analysis of the electronic nose information of the dough at different fermentation times shows that the cumulative contribution rate of the first two principal components can reach 99.84 %.According to the weight coefficient in the two principal components and the dispersion of the response value of each sensor,the sensor array that can reflect the characteristic changes of the dough fermentation process was screened out.The analysis showed that the optimal sensor response value was extremely significantly correlated with the gas production in the initial and moderate stages of dough fermentation,while the sensor response value was extremely significantly negatively correlated with the air leakage in the over-fermentation stage(p<0.01).The results of the K-nearest neighbor(KNN)and Support Vector Machine(SVM)recognition models constructed based on the electronic nose technology for the dough fermentation state show that the recognition accuracy rates of the training set and test set of the SVM model are respectively 97 %,92 %,better than the KNN model.(3)Research on the monitoring method of dough fermentation based on near-infrared spectroscopy technologyIn this study,a handheld near-infrared spectrometer was used to collect the spectrum information of the dough during the fermentation process,and the SG-convolution smoothing(frame size was 15)was used to reduce the noise of the spectrum.The full spectrum was divided into 10 intervals by the synergy interval partial least squares method(Si-PLS)and 205 spectral variables in 4 intervals were selected.The Competitive Adaptive Reweighted Sampling(CARS)method was further used to optimize the variables,and finally 13 effective variables were selected.The comparison of the results of the KNN and SVM models of dough fermentation state monitoring constructed with different variable sets showed that the CARS-Si-PLS joint algorithm can effectively remove redundant variables and improve the accuracy of the model.The results of the model built with the variables selected by the joint algorithm showed that the recognition accuracy of the training set and the test set of the SVM model are 94 % and 92 %,respectively,which is better than the KNN model.(4)Research on the interaction of multi-sensory information in the assessment of CSB quality based on cognitive psychologyBased on the Stroop effect of cognitive psychology,three assessment models of CSB quality under different Stroop paradigms were designed to explore the interaction between various sensory information.The results showed that the color of CSB had the greatest Stroop effect on the morphology(internal structure and surface structure)score,and has little effect on other indicators.The morphology had a slight influence on other sensory indicators of CSB.When the whiteness of the raw material was less than 75,the color of the CSB was about 8 points,which is the watershed for the formation of the Stroop effect.The color higher or lower than 8 points can respectively promote or interfere with other indicators.The amount of Stroop effect was not only affected by the difference in color,but also due to the obvious difference in the quality of CSB(p<0.05).Comprehensive analysis showed that when assessing the sensory quality of CSB,there are multi-dimensional and differentiated interactions between various sensory indicators,and there is no obvious rule to follow.Therefore,a sensory assessment model of CSB that can reduce sensory information interaction should be designed to ensure the accuracy and objectivity of the assessment results,so as to improve the generalization ability when building an intelligent assessment model of CSB sensory quality based on sensor technology.(5)Research on the intelligent assessment method of CSB sensory quality based on sensor technologyIn this study,the sensory quality score of CSB in the non-interactive assessment mode was used as the output.First,computer vision,texture analyzer and electronic nose were used to establish intelligent assessment methods for the visual,texture and odor quality of CSB,respectively.The shape image of CSB was analyzed and processed under different color components,and characteristic variables such as color,defect,texture,etc.were extracted,and an intelligent assessment models of the surface color,surface structure and internal structure of CSB were established.A texture analyzer was used to collect texture characteristic variables,and intelligent assessment models for the elasticity,toughness and viscosity of CSB were established.An intelligent assessment model of the odor quality of CSB was also established based on the electronic nose technology.The results showed that there is a high correlation between the response information of each sensor technology and the corresponding artificial sensory assessment scores.The results also showed that,except for the viscosity index,the performance of the Support vector regression(SVR)prediction model of other indexes was better than that of the partial least squares(PLS)model.This study further used the feature fusion technology of multi-source information to construct an intelligent assessment model of CSB comprehensive sensory quality based on the combination of multi-sensor technology.The results showed that the Rc and Rp of the CSB comprehensive sensory assessment of SVR model were 0.9808 and 0.9573,respectively,which had better generalization ability than the individual sensory assessment models.This study analyzed the changing laws of a series of indicators during the dough fermentation,and provided a theoretical basis for using sensor technology to establish an intelligent monitoring method of dough fermentation.This study also investigated the interaction effect of multi-sensory information from the perspective of cognitive psychology,and opened up a new idea for improving the generalization ability of the intelligent assessment model of sensory quality.Based on the above theoretical research foundation,intelligent models for dough fermentation monitoring and CSB quality assessment were established.At the same time,this research could meet the urgent needs of the current industrial development and provide a research foundation for the development of new intelligent equipment for traditional food.The implementation of the research results will not only help improve the production efficiency of enterprises and enhance the intelligent level of traditional pasta production,but also help to carry forward our country's traditional food culture.