A method for evaluating the crispness of fried potato chips

Abstract

The present invention provides a method for evaluating the crispness of fried potato chips, which belongs to the technical field of fried potato chips detection, and includes: comprehensively collecting product feature data through multi-source sensing technology. The method innovatively uses multi-dimensional decomposition of surface images, vibration measurement of micro-accelerometers and infrared imaging technology to extract crispness correlation coefficients from three dimensions: surface morphology, vibration response and thermodynamic characteristics. By constructing a mapping matrix of surface wrinkle degree, porosity, texture characteristics and vibration steady-state, transient components, thermal diffusion and thermal attenuation coefficients, an accurate group of empirical equations for crispness evaluation is established. The innovative method can calculate the comprehensive crispness score and perform grade classification. By correcting with the measurement data of a standard texture analyzer, the self-iterative optimization of the evaluation method is achieved, and a high-precision intelligent evaluation technology solution is provided for the quality control of fried potato chips. The problem that the existing technology is time-consuming and labor-intensive and difficult to meet the needs of rapid evaluation and online monitoring is solved.

Description

Technical Field

[0001] The invention belongs to the technical field of fried potato chips detection, and in particular relates to a method for evaluating the crispness of fried potato chips products. Background Art

[0002] Fried potato chips are a popular snack, and their crispy and delicious taste has always been the focus of consumers. The quality and taste of fried potato chips currently sold on the market vary greatly, which seriously affects consumers' purchasing experience and satisfaction. In order to ensure product quality, manufacturers need to effectively evaluate and control the crispness of products. However, existing evaluation methods generally have limitations and are difficult to fully reflect the actual crispness characteristics of products.

[0003] Traditional evaluation methods mainly rely on manual sensory judgment or standard texture analyzer measurement. This method has the problems of strong subjectivity, poor repeatability and inability to deeply analyze the internal structural characteristics of the product. Manual sensory judgment is easily affected by the operator's experience and emotional state, and it is difficult to quantify the precise brittleness of the product. Although the standard texture analyzer can obtain some objective mechanical indicators, such as maximum destructive force and fracture energy, this single mechanical test cannot fully reflect the overall brittleness characteristics of the product. In addition, these methods require a large number of physical samples for experimental testing, which is time-consuming and labor-intensive, and it is difficult to meet the needs of rapid evaluation and online monitoring.

[0004] Therefore, it is urgent to develop a new method that can quickly evaluate the crispness of fried potato chips products to meet the quality control needs of manufacturers and improve consumers' experience. Summary of the invention

[0005] In view of this, the present invention provides a method for evaluating the crispness of fried potato chips, which can solve the problem that the prior art is time-consuming and labor-intensive and difficult to meet the needs of rapid evaluation and online monitoring.

[0006] The present invention is achieved in that:

[0007] The present invention provides a method for evaluating the crispness of a fried potato chip product, comprising the following steps:

[0008] S10, collecting surface image data, vibration response data, and thermal infrared data of the fried potato chips product as original evaluation data;

[0009] S20, performing multi-dimensional decomposition on the surface image data to extract a surface wrinkle coefficient, a surface porosity coefficient, and a surface texture characteristic coefficient;

[0010] S30, using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under standard wind blowing excitation, and obtaining the vibration steady-state component coefficient and the vibration transient component coefficient through multivariate mixed decomposition calculation;

[0011] S40, using an infrared imager to measure the thermal response data of the fried potato chips product under standard temperature difference conditions to obtain a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix;

[0012] S50, establishing a first mapping matrix between the surface wrinkle coefficient, the surface porosity coefficient, the surface texture characteristic coefficient and the brittleness parameter;

[0013] S60, establishing a second mapping matrix between the vibration steady-state component coefficient, the vibration transient component coefficient and the brittleness parameter;

[0014] S70, establishing a third mapping matrix between the thermal diffusion coefficient matrix, the thermal attenuation coefficient matrix and the brittleness parameter;

[0015] S80, substituting the first mapping matrix, the second mapping matrix, and the third mapping matrix into a group of empirical equations for brittleness assessment to calculate a comprehensive brittleness score;

[0016] S90, classifying the fried potato chip product into crispness grades based on the comprehensive crispness score, and correcting the evaluation result with the standard texture analyzer measurement data, and updating the mapping matrix parameters in the crispness evaluation empirical equation group.

[0017] Furthermore, the calculation steps involved in the present invention are described in detail as follows:

[0018] 1. The multi-dimensional decomposition (S20) of the surface image data is specifically expressed as follows:

[0019]

[0020] Where, T w is the surface wrinkle coefficient; P r is the surface porosity coefficient; F t is the surface texture characteristic coefficient; I(x,y) is the image grayscale value; B(i,j) is the binary image matrix; G x (i,j) and G y (i, j) are the gradient operators in the x-direction and y-direction respectively; n, m are the image sizes;

[0021] 2. The multivariate mixed decomposition (S30) of the vibration response data is specifically expressed as follows:

[0022] V(t)=V s (t)+V t (t)+η(t);

[0023]

[0024] Where V(t) is the total vibration response; V s (t) is the steady-state component; V t (t) is the transient component; η(t) is the noise term; A k ,B l is the amplitude coefficient; f k ,f l is the frequency; φ k ,θ l is the phase angle; α l is the attenuation coefficient; K, L are the number of components;

[0025] 3. The processing of thermal response data (S40) is specifically shown as follows:

[0026]

[0027] Where, T(x, y, t) is the temperature distribution function; D is the thermal diffusion coefficient; Q(x, y, t) is the heat source term; T0 is the initial temperature; β is the thermal attenuation coefficient; J0 is the zero-order Bessel function; λ, γ are the eigenvalues;

[0028] 4. The establishment of the mapping matrix (S50-S70) is specifically expressed as follows:

[0029]

[0030] Where, M1, M2, M3 are the first, second, and third mapping matrices respectively; a ij ,b ij ,c ij is the coefficient to be determined;

[0031] 5. The calculation of the comprehensive brittleness score (S80) is specifically expressed as follows:

[0032]

[0033] Where C is the comprehensive brittleness score; w1, w2, w3 are weight coefficients; ∈ is the error term;

[0034] 6. The principal component analysis of the brittleness characteristic parameters (step 7) is specifically expressed as follows:

[0035]

[0036] |λI-R|=0;

[0037] RC = λC;

[0038] Y = ZC;

[0039] In the formula, Z i is the standardized characteristic parameter; X i is the original characteristic parameter; μ i is the mean; σ i is the standard deviation; R is the correlation coefficient matrix; λ is the eigenvalue; C is the eigenvector; Y is the principal component score;

[0040] The construction principles and meanings of these equations are as follows:

[0041] 1. The surface feature extraction equation is based on image processing theory, using the Laplace operator to calculate the wrinkle degree, binarization statistics to calculate the porosity, and the Sobel operator to calculate the texture features, which fully characterizes the surface structural characteristics of the product;

[0042] 2. The vibration response decomposition equation is based on Fourier analysis and exponential decay principle, which decomposes the complex vibration signal into steady-state and transient parts, and more accurately reflects the dynamic mechanical properties of the product;

[0043] 3. The thermal response equation is based on the heat conduction equation and Bessel function, taking into account the two-dimensional heat diffusion and attenuation process, and can characterize the internal structural characteristics of the product;

[0044] 4. The mapping matrix adopts the linear transformation form to establish the correlation between characteristic parameters and brittleness, which has good interpretability and scalability;

[0045] 5. The comprehensive scoring equation adopts a weighted summation form, which comprehensively considers multi-dimensional feature information and improves the accuracy and robustness of the evaluation;

[0046] 6. The principal component analysis equation is based on dimensionality reduction theory, which transforms multiple related characteristic parameters into a few independent comprehensive indicators, thus improving the evaluation efficiency.

[0047] The fitting step of the brittleness assessment empirical equation group comprises the following steps:

[0048] Step 1: Construct regression optimization objective function:

[0049]

[0050] In the formula, C i is the measured brittleness score of the i-th sample; To predict the crispness score; M j (k,l) is the element of the jth mapping matrix; w j is the weight coefficient; α, β are regularization coefficients, ranging from [0.01, 0.1]; N is the number of samples;

[0051] Step 2: Design an iterative optimization algorithm:

[0052]

[0053] In the formula, θ t is the parameter vector of the tth iteration, including all parameters to be optimized; η t is the learning rate; η0 is the initial learning rate, which is 0.01; γ is the attenuation coefficient, which is 0.001; L t is the loss function value of the tth iteration;

[0054] Step 3: Calculate parameter gradients:

[0055]

[0056] In the formula, and are the partial derivatives of the predicted value with respect to the mapping matrix and weight respectively;

[0057] Step 4: Introduce momentum term to optimize update strategy:

[0058]

[0059] θ t+1 =θ t -v t+1 ;

[0060] In the formula, v t is the momentum term; μ is the momentum coefficient, which is 0.9;

[0061] Step 5: Design an early stopping strategy:

[0062] ΔL t =|L t -L t-1 |;

[0063]

[0064] In the formula, ΔL t is the change in the loss function; S t is the cumulative change of the most recent K iterations; K is the observation window size, which is 50; when S t <∈, where ∈ is the convergence threshold, which is 10. -6 ;

[0065] Step 6: Cross-validation optimization of hyperparameters:

[0066]

[0067] Where λ is a hyperparameter vector, including α, β, η0, γ, μ, etc.; K is the number of cross-validation folds, which is 5; D k is the k-th fold data set; is the predicted value of sample i by the model trained with the hyperparameter λ;

[0068] The principle and significance of the fitting steps are as follows:

[0069] 1. The optimization objective function adopts a combination of mean square error and regularization term, which not only ensures the fitting accuracy but also avoids the overfitting problem;

[0070] 2. The iterative optimization algorithm uses the stochastic gradient descent method with momentum to improve the optimization efficiency and stability;

[0071] 3. The parameter gradient calculation takes into account the influence of the regularization term, making the optimization process smoother;

[0072] 4. Momentum optimization strategy can help the algorithm escape from the local optimal solution and speed up convergence;

[0073] 5. The early stopping strategy avoids excessive iteration by monitoring the trend of the loss function;

[0074] 6. The cross-validation method is used to select the optimal hyperparameter combination and improve the generalization ability of the model.

[0075] Through the above fitting steps, a group of empirical equations for brittleness assessment with good prediction performance and generalization ability can be obtained.

[0076] Among the data fitted by the brittleness assessment empirical equation group, the steps for obtaining the brittleness score include:

[0077] Step 1, placing the fried potato chips test sample in a constant temperature and humidity environment for pretreatment, so that the fried potato chips test sample is placed in an environment of 25 degrees Celsius and a relative humidity of 60% for at least 4 hours;

[0078] Step 2, using a texture analyzer, set the compression probe diameter to 5 mm, the compression speed to 1 mm per second, the compression deformation to 2 mm, and the trigger force to 5 g;

[0079] Step 3, placing the fried potato chips test product at the center of the texture analyzer test platform, so that the compression probe is perpendicular to the test area of ​​the fried potato chips test product;

[0080] Step 4, starting the texture analyzer to perform a compression test, and recording the stress-strain curve of the fried potato chips during the compression process;

[0081] Step 5, extracting the maximum fracture force, the linear region slope, and the fracture energy value from the stress-strain curve as brittleness characteristic parameters;

[0082] Step 6, establishing the brittleness characteristic parameter database, and performing standardization processing on the maximum breaking force, the linear region slope, and the breaking energy value;

[0083] Step 7: Use the principal component analysis method to reduce the dimension of the standardized brittleness characteristic parameters to obtain a comprehensive brittleness score. The specific formula of step 5 is:

[0084] The calculation formula for extracting characteristic parameters from the stress-strain curve is as follows:

[0085] 1. Maximum breaking force (F max ) calculation:

[0086] F max = max{F(t)|t∈[0,T]},

[0087] Where F(t) is the instantaneous force value during the compression process, and T is the total duration of the compression test.

[0088] 2. Linear region slope (K l ) calculation:

[0089]

[0090] Where [d1, d2] is the linear region range of the stress-strain curve, (F1, d1) and (F2, d2) are the starting and ending coordinates of the linear region.

[0091] 3. Fracture energy value (E b ) calculation:

[0092]

[0093] where d b is the displacement of the fracture point, and F(d) is the functional relationship between the compression force and the displacement.

[0094] The specific formula for step 6 is:

[0095] 1. Calculation formula for standardization:

[0096] Normalized maximum breaking force:

[0097] Normalized linear region slope:

[0098] Normalized fracture energy value:

[0099] in:

[0100] μ F , μ K , μ EF max , K l 、E b The sample mean of ;

[0101] σ F , σ K , σ E F max , K l 、E b The sample standard deviation of .

[0102] The specific formula for step 7 is:

[0103] 1. Build a standardized feature matrix Z:

[0104] Z=[Z F ,Z K ,Z E ].

[0105] 2. Calculate the correlation coefficient matrix R:

[0106]

[0107] Where n is the number of samples.

[0108] 3. Solve the characteristic equation:

[0109] |λI-R|=0,

[0110] Where λ is the eigenvalue and I is the identity matrix.

[0111] 4. Solve the eigenvector equation:

[0112] RC=λC,

[0113] Where C is the eigenvector.

[0114] 5. Calculate the principal component score Y:

[0115] Y=ZC.

[0116] 6. Calculate the comprehensive brittleness score S:

[0117]

[0118] Where p is the number of principal components selected, λ i is the i-th eigenvalue, Y i Score the i-th principal component.

[0119] Variable Description:

[0120] F(t): instantaneous force value during compression, in Newton (N);

[0121] T: total duration of compression test, in seconds (s);

[0122] F max : Maximum breaking force, in Newton (N);

[0123] K l : Linear region slope, in Newton / millimeter (N / mm);

[0124] E b : Fracture energy value, in millijoules (mJ);

[0125] d: compression displacement, in millimeters (mm);

[0126] d b : Displacement of the breaking point, in millimeters (mm);

[0127] Z F , Z K , Z E : standardized dimensionless characteristic parameter;

[0128] μ: sample mean, the same unit as the original data;

[0129] σ: sample standard deviation, in the same unit as the original data;

[0130] R: correlation coefficient matrix, dimensionless;

[0131] λ: eigenvalue, dimensionless;

[0132] C: eigenvector, dimensionless;

[0133] Y: principal component score, dimensionless;

[0134] S: comprehensive brittleness score, the calculation result is normalized to the interval [0,100]

[0135] Note:

[0136] 1. The linear region is determined by the point-by-point slope change rate method. When the slope change rate of two adjacent points is less than a threshold (usually 1%), it is considered to belong to the linear region;

[0137] 2. The breaking point is determined by the first obvious sudden drop in compression force, which usually exceeds 20% of the current force value.

[0138] 3. During principal component analysis, the principal components with cumulative contribution rates of eigenvalues ​​greater than 85% are selected for comprehensive score calculation;

[0139] 4. The comprehensive crispness score needs to be normalized so that the score falls within the range of 0-100 to facilitate subsequent grading.

[0140] The crispness evaluation method of fried potato chips proposed in the present invention systematically extracts multi-dimensional features of the product, including surface wrinkle degree, surface porosity, surface texture features, as well as vibration steady-state component, vibration transient component, thermal diffusion coefficient and thermal attenuation coefficient, etc., by collecting and analyzing surface image data, vibration response data and thermal infrared data of the product. These characteristic parameters comprehensively reflect the structural characteristics, dynamic response characteristics and thermal conductivity characteristics of the product, and provide a rich source of information for crispness evaluation. For the above multi-dimensional characteristic parameters, the present invention establishes a series of mapping matrices to accurately correspond these characteristic parameters with the crispness parameters of the product, and realizes the quantitative conversion from multi-source information to comprehensive crispness evaluation. This method based on mapping matrix not only overcomes the limitations of strong subjectivity and singleness of traditional methods, but also establishes the mathematical relationship between characteristics and crispness in the form of matrix, and has good interpretability and scalability. In addition, the present invention also designs a crispness characteristic parameter optimization method based on principal component analysis, which converts multiple related parameters into a few independent comprehensive indicators, greatly improving the efficiency of evaluation. At the same time, this method also integrates the measurement data of standard texture analyzer to correct and optimize the brittleness assessment results, ensuring the accuracy and reliability of the assessment.

[0141] Compared with the prior art, the beneficial effects of the crispness evaluation method for fried potato chips provided by the present invention are: 1) it can deeply explore the internal structure and performance characteristics of the product and comprehensively characterize the crispness state of the product; 2) a quantitative mapping relationship between characteristic parameters and crispness is established, which has good interpretability; 3) the introduction of principal component analysis and standard test data correction greatly improves the accuracy and robustness of the evaluation; 4) the evaluation can be completed quickly to meet the online monitoring needs of manufacturers. BRIEF DESCRIPTION OF THE DRAWINGS

[0142] Figure 1 A flow chart of the method provided by the present invention;

[0143] Figure 2 Schematic diagram of the equipment composition for measuring vibration response data. DETAILED DESCRIPTION

[0144] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0145] like Figure 1 FIG. 1 is a flow chart of a method for evaluating the crispness of fried potato chips provided by the present invention. The method comprises the following steps:

[0146] S10, collecting surface image data, vibration response data, and thermal infrared data of the fried potato chips product as original evaluation data;

[0147] S20, performing multi-dimensional decomposition on the surface image data to extract the surface wrinkle coefficient, surface porosity coefficient, and surface texture characteristic coefficient;

[0148] S30, using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under standard wind blowing excitation, and obtaining the vibration steady-state component coefficient and the vibration transient component coefficient through multivariate mixed decomposition calculation;

[0149] S40, using an infrared imager to measure thermal response data of a fried potato chip product under standard temperature difference conditions to obtain a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix;

[0150] S50, establishing a first mapping matrix between the surface wrinkle coefficient, the surface porosity coefficient, the surface texture characteristic coefficient and the brittleness parameter;

[0151] S60, establishing a second mapping matrix between the vibration steady-state component coefficient, the vibration transient component coefficient and the brittleness parameter;

[0152] S70, establishing a third mapping matrix among the thermal diffusion coefficient matrix, the thermal attenuation coefficient matrix and the brittleness parameter;

[0153] S80, substituting the first mapping matrix, the second mapping matrix, and the third mapping matrix into the brittleness evaluation empirical equation group to calculate the comprehensive brittleness score;

[0154] S90. Classify the fried potato chip products into crispness grades based on the comprehensive crispness scores, calibrate the evaluation results with the standard texture analyzer measurement data, and update the mapping matrix parameters in the crispness evaluation empirical equation group.

[0155] The specific implementation methods of the above steps are described in detail below:

[0156] The specific implementation method of step S10 is: collecting the surface image data, vibration response data, and thermal infrared data of the fried potato chips product as the original evaluation data. First, an industrial camera is used to obtain the image data of the surface of the fried potato chips product under standard light source conditions, and the resolution of the industrial camera is not less than 12 million pixels to ensure the acquisition quality of image details. Secondly, a micro acceleration sensor is used to measure the vibration response data of the fried potato chips product under the standard wind blowing excitation, and the wind speed of the standard wind blowing excitation is 2 meters per second to simulate the dynamic mechanical characteristics in the real use scenario. Finally, an infrared imager is used to measure the thermal response data of the fried potato chips product at an ambient temperature of 25 degrees Celsius, and the temperature resolution of the infrared imager is not less than 0.1 degrees Celsius to accurately capture the heat conduction process inside the product. These rich original data lay the foundation for the subsequent multi-dimensional feature extraction.

[0157] like Figure 2 As shown, the equipment for measuring vibration response data includes:

[0158] Micro accelerometer: A high-precision sensor using MEMS technology that can capture tiny vibrations;

[0159] Wind speed generating device: equipped with precise wind speed control system to ensure stable airflow of 2m / s±0.1m / s;

[0160] Data acquisition system: high-speed sampling, recording detailed information of vibration response;

[0161] Sample fixing bracket: ensure the stable position of the sample and reduce installation errors;

[0162] Environmental control box: controls temperature and humidity to eliminate the influence of external environmental factors.

[0163] The specific implementation method of step S20 is: multi-dimensional decomposition of the surface image data is performed to extract the surface wrinkle coefficient, surface porosity coefficient, and surface texture characteristic coefficient. First, the surface image data is gradient calculated by quadratic partial derivative operation, and the second-order partial derivatives of each pixel in the horizontal and vertical directions of the surface image data are summed to obtain the surface wrinkle coefficient representing the degree of surface wrinkles. Secondly, the surface image data is binarized, and the ratio of the number of black pixels to the total number of pixels in the surface image data is calculated to obtain the surface porosity coefficient representing the surface pore distribution. Finally, the surface image data is edge detected by using a gradient operator, and the sum of the absolute values ​​of the gradient values ​​in the horizontal and vertical directions is calculated to obtain the surface texture characteristic coefficient representing the surface texture characteristics. These three coefficients comprehensively characterize the structural characteristics of the surface of the fried potato chip product, laying a foundation for subsequent mapping analysis.

[0164] The specific implementation method of step S30 is: using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under standard wind blowing excitation, and obtaining the vibration steady-state component coefficient and the vibration transient component coefficient through multivariate mixed decomposition calculation. First, the vibration response data is decomposed into a steady-state vibration component, a transient vibration component and a noise component. Secondly, the steady-state vibration component is decomposed by Fourier series expansion to obtain the amplitude coefficient, frequency and phase angle parameters of multiple sinusoidal components. Then, the transient vibration component is fitted by an exponential decay function to obtain the attenuation coefficient, amplitude coefficient and phase angle parameter. Finally, the amplitude coefficient, the frequency, the phase angle parameter and the attenuation coefficient are combined to form the vibration steady-state component coefficient and the vibration transient component coefficient. These dynamic characteristic parameters provide key information for subsequent crispness evaluation.

[0165] The specific implementation method of step S40 is: using an infrared imager to measure the thermal response data of the fried potato chips product under standard temperature difference conditions to obtain a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix. First, a two-dimensional heat conduction partial differential equation model is established, and the thermal response data is substituted into the two-dimensional heat conduction partial differential equation model. Secondly, the thermal response data is decomposed using a zero-order Bessel function to obtain a thermal diffusion coefficient and a thermal attenuation coefficient that characterize the internal structure characteristics. Finally, the thermal diffusion coefficient and the thermal attenuation coefficient are matrixed on a two-dimensional plane to form a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix. These thermal response characteristic parameters provide necessary input for subsequent multi-source information fusion.

[0166] The specific implementation of step S50 is: establish a first mapping matrix between the surface wrinkle coefficient, the surface porosity coefficient, the surface texture characteristic coefficient and the brittleness parameter. First, establish a first mapping matrix with 3 rows and 3 columns, and the row vectors of the first mapping matrix correspond to the surface wrinkle coefficient, the surface porosity coefficient, and the surface texture characteristic coefficient, respectively. Secondly, fit the coefficient parameters in the first mapping matrix by the least squares method. Finally, map the surface wrinkle coefficient, the surface porosity coefficient, and the surface texture characteristic coefficient to the brittleness parameter space through the first mapping matrix. The mapping matrix establishes a quantitative relationship between surface characteristics and brittleness, providing an important basis for comprehensive evaluation.

[0167] The specific implementation of step S60 is: establish a second mapping matrix between the vibration steady-state component coefficient, the vibration transient component coefficient and the brittleness parameter. First, establish a second mapping matrix with 2 rows and 2 columns, and the row vectors of the second mapping matrix correspond to the vibration steady-state component coefficient and the vibration transient component coefficient respectively. Secondly, fit the coefficient parameters in the second mapping matrix by the least squares method. Finally, map the vibration steady-state component coefficient and the vibration transient component coefficient to the brittleness parameter space through the second mapping matrix. The mapping matrix establishes a quantitative relationship between dynamic response characteristics and brittleness, providing an important basis for comprehensive evaluation.

[0168] The specific implementation method of step S70 is: establish a third mapping matrix between the thermal diffusion coefficient matrix, the thermal attenuation coefficient matrix and the brittleness parameter. First, establish a third mapping matrix with 2 rows and 2 columns, and the row vectors of the third mapping matrix correspond to the thermal diffusion coefficient matrix and the thermal attenuation coefficient matrix respectively. Secondly, fit the coefficient parameters in the third mapping matrix by the least squares method. Finally, map the thermal diffusion coefficient matrix and the thermal attenuation coefficient matrix to the brittleness parameter space through the third mapping matrix. The mapping matrix establishes a quantitative relationship between thermal response characteristics and brittleness, providing an important basis for comprehensive evaluation.

[0169] The specific implementation method of step S80 is: substitute the first mapping matrix, the second mapping matrix, and the third mapping matrix into the brittleness evaluation empirical equation group to calculate the comprehensive brittleness score. First, assign weight coefficients to the first mapping matrix, the second mapping matrix, and the third mapping matrix respectively. Secondly, multiply the surface feature coefficient, the vibration component coefficient, and the thermal response coefficient with the corresponding mapping matrix respectively. Finally, multiply the product result with the corresponding weight coefficient and sum them to obtain the comprehensive brittleness score. The comprehensive scoring equation fully integrates multi-dimensional feature information and improves the accuracy and robustness of the evaluation.

[0170] The specific implementation of step S90 is: based on the comprehensive crispness score, the crispness grade of the fried potato chips product is classified, and the evaluation result is corrected with the standard texture analyzer measurement data, and the mapping matrix parameters in the crispness evaluation empirical equation group are updated. First, a crispness grade classification standard is established based on the comprehensive crispness score, and the crispness grade is divided into five levels. Secondly, the crispness grade classification standard is corrected using the standard texture analyzer measurement data. Finally, the coefficient parameters and the weight coefficients in the first mapping matrix, the second mapping matrix, and the third mapping matrix are updated according to the correction results. This step realizes the continuous optimization and improvement of the method, so that it can better adapt to different product characteristics and application scenarios.

[0171] Specifically, the principle of the present invention is: by collecting and analyzing the surface image data, vibration response data and thermal infrared data of the product, the system extracts multi-dimensional characteristic parameters that can fully reflect the crispness characteristics of the product, and establishes a quantitative mapping relationship between these characteristic parameters and crispness, and finally realizes accurate evaluation of the crispness of the product. The technical principles of this method are mainly reflected in the following aspects:

[0172] First, the structural characteristics of the product surface are one of the key factors affecting brittleness. The present invention uses image processing technology to extract three key indicators: surface wrinkle degree, surface porosity, and surface texture characteristics. Among them, surface wrinkle degree reflects the degree of undulation on the product surface and is the main location where the product is damaged under the action of external force; surface porosity reflects the density of the internal structure of the product. The more pores, the easier it is for the product to break; surface texture characteristics reflect the fineness of the product surface. Rough surfaces are more likely to be damaged. These three characteristic parameters constitute important dimensions for characterizing the surface structure of the product.

[0173] Secondly, the dynamic response characteristics of the product are also a key factor affecting its brittleness. The present invention extracts two parameters, the steady-state vibration component and the transient vibration component, through vibration response analysis. The steady-state vibration component reflects the inherent vibration characteristics of the product under the action of external force, while the transient vibration component reflects the transient response characteristics of the product when the external force changes. These two characteristic parameters together constitute a comprehensive characterization of the dynamic mechanical characteristics of the product.

[0174] Thirdly, the thermal conductivity of the product is also an important factor affecting brittleness. The present invention extracts two parameters, thermal diffusion coefficient and thermal attenuation coefficient, through thermal infrared analysis. The thermal diffusion coefficient reflects the ability of internal heat conduction of the product, and the thermal attenuation coefficient reflects the rate of heat attenuation on the surface of the product. These two characteristic parameters can deeply reflect the internal structure and thermal conduction process of the product, providing an important supplement for brittleness assessment.

[0175] Finally, the present invention establishes a series of mapping matrices between surface characteristics, dynamic response characteristics, thermal conductivity characteristics and product crispness. These mapping relationships in the form of matrices can not only realize the quantitative conversion of characteristic parameters to crispness parameters, but also have good interpretability and scalability. At the same time, the present invention also introduces the principal component analysis method to optimize the combination of characteristic parameters, further improving the efficiency of the evaluation. In addition, the accuracy and reliability of the evaluation results are ensured by calibration with the standard texture analyzer measurement data.

[0176] In summary, the crispness evaluation method for fried potato chips proposed in the present invention makes full use of the technical means of multi-source information fusion and mathematical modeling, realizes a comprehensive analysis of the internal structure, dynamic response and thermal conductivity characteristics of the product, establishes a quantitative mapping relationship between these characteristic parameters and crispness, and finally realizes fast, accurate and reliable crispness evaluation.

[0177] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows: The specific implementation of step S10: For the image data collected by the industrial camera, the resolution requirement can be expressed as R c =N x ×N y ≥1.2×10 7 Pixels, where R c Indicates the camera resolution, N x and N y Respectively represent the number of pixels in the horizontal and vertical directions of the image; the illumination range of the standard light source is E l ∈[800,1200] lux, color temperature T l =5500 Kelvin, color rendering index R a ≥90, illumination uniformity U l =E min / Emax ≥0.85, where E min and E max are the minimum and maximum illumination values ​​of the measured area respectively; the sensitivity S of the micro acceleration sensor a ≥100 mV / gravity acceleration, frequency response range f a ∈[0,1000] Hz, sampling rate f s ≥2000 Hz; Standard wind speed v for wind excitation w =2.0±0.1 m / s, duration t w = 10 seconds, angle θ with the sample surface w =45±5 degrees; the temperature resolution of the infrared imager is ΔT≤0.1 degrees Celsius, and the temperature measurement range is T r ∈[-20,100] degrees Celsius, spatial resolution IFOV ≤ 1.0 millirads, image frequency f i ≥30 Hz, ambient temperature controlled at T e =25±0.5 degrees Celsius.

[0178] Specific implementation of step S20: First, define the gray value function I(x,y) of the image, where x∈[1,N x ],y∈[1,N y ] is the pixel coordinate, the gray value range is i∈[0,255]; the surface wrinkle coefficient is calculated using the Laplace operator After discretization, it can be expressed as The image binarization uses the OTSU algorithm and sets the threshold t b , so that the between-class variance The maximum, where w1, w2 is the ratio of the two types of pixels, μ1, μ2 is the average gray value of the two types of pixels, the white pixel value in the binary image matrix B (i, j) is 1, and the black pixel value is 0; the surface porosity coefficient calculation formula is: The surface texture feature uses the Sobel operator and the horizontal operator Vertical operator Texture feature coefficient Where * indicates the convolution operation.

[0179] Specific implementation of step S30: The original vibration signal measured by the acceleration sensor is recorded as a(t), t∈[0,T], where T is the sampling time; the vibration displacement signal V(t)=∫∫a(t)dt is obtained by integration operation 2 ; Use fast Fourier transform (FFT) to analyze the displacement signal and get the spectrum function Determine the main frequency component f according to the power spectral density distribution k,k=1,2,…,K, where K is the number of main frequencies selected; perform least squares fitting on each frequency component to obtain the amplitude coefficient A k and phase angle So we get the steady-state component The transient component is fitted using an exponential decay model. Where L is the number of transient components, B l ,α l ,θ l are the amplitude, attenuation coefficient and phase angle respectively; the noise component η(t) is suppressed by bandpass filtering, and the cutoff frequency range of the filter is [f low ,f high ] = [10,500] Hz.

[0180] Specific implementation of step S40: The temperature field data collected by the infrared imager is expressed as T(x, y, t), the spatial resolution is Δx = Δy = 0.5 mm, and the time resolution is Δt = 1 / f i Seconds; Establish two-dimensional heat conduction partial differential equation Where D is the thermal diffusion coefficient and Q(x,y,t) is the heat source term. The equation is discretized using the finite difference method to obtain The superscript n represents the time layer, and the subscripts i and j represent the spatial grid points; the zero-order Bessel function expansion Where T0 is the initial temperature, β is the thermal attenuation coefficient, λ,γ are the eigenvalues; the thermal diffusion coefficient matrix D is obtained by least squares fitting. i,j and the thermal attenuation coefficient matrix β i,j .

[0181] Specific implementation of step S50: establish a first mapping matrix M1 between surface features and brittleness parameters, with a dimension of 3×3; use the least squares method to perform parameter fitting, and the objective function is Among them C k is the crispness parameter vector of the kth sample, X k =[T w ,P r ,F t ] T is the corresponding surface feature vector, N is the number of samples, α is the regularization coefficient, ∥·∥2 represents the 2-norm, ∥·∥ Y represents the Frobenius norm; solving the optimization problem to obtain the element a of the mapping matrix M1 ij ,i,j=1,2,3.

[0182] Specific implementation of step S60: establish a second mapping matrix M2 between vibration characteristics and brittleness parameters, with a dimension of 2×2; use the least squares method to perform parameter fitting, and the objective function is where Y k =[V s ,V t ] T is the vibration eigenvector; according to the vibration steady-state component coefficient and transient component coefficient Construct the eigenvector; solve the optimization problem to obtain the element b of the mapping matrix M2 ij ,i,j=1,2.

[0183] Specific implementation of step S70: establish a third mapping matrix M3 between thermal response characteristics and brittleness parameters, with a dimension of 2×2; use the least squares method to perform parameter fitting, and the objective function is Where Z k =[D,β] T is the thermal response feature vector; the feature vector is constructed according to the statistical characteristics of the thermal diffusion coefficient matrix and the thermal attenuation coefficient matrix, including the mean and Standard Deviation and Solve the optimization problem to obtain the element c of the mapping matrix M3 ij ,i,j=1,2.

[0184] Specific implementation of step S80: The comprehensive brittleness score is calculated using a weighted summation form C = w1(M1X) + w2(M2Y) + w3(M3Z) + ε, where w1, w2, w3 are weight coefficients that satisfy The weight coefficients are determined by cross-validation, and the data set is randomly divided into a training set and a validation set with a ratio of 7:3. For each set of weight coefficients, the mean square error on the validation set is calculated. Where N v is the number of samples in the validation set; the weight combination that minimizes the MSE is found through the grid search method; the error term ε has a mean of 0 and a standard deviation of σ ε The normal distribution of .

[0185] Specific implementation of step S90: Establish a five-level crispness classification standard based on the comprehensive crispness score C, define the level boundary values ​​[b0, b1, b2, b3, b4, b5], where b0 = 0, b5 = 100, and the intermediate boundary values ​​are determined by the K-means clustering algorithm; use the reference crispness value C obtained by standard texture analyzer measurement ref Perform correction and establish the correction function C corr =f(C)=k1C 3 +k2C2 +k3C+k4, where k1, k2, k3, k4 are polynomial coefficients; update the mapping matrix parameters according to the correction results, and the new mapping matrix is ​​M i ′=(1-η)M i +ηΔM i ,i=1,2,3, where η is the learning rate, ΔM i is the parameter update amount; the weight coefficient is updated using the gradient descent method. in To correct the error.

[0186] In order to better understand and implement the present invention, Example 2 of a specific application scenario of the present invention is provided below: A food company A introduced the crispness evaluation method of fried potato chips proposed in the present invention; first, the company's R&D team sorted out and analyzed the existing quality inspection process, and found that it currently mainly relies on two methods: manual sensory evaluation and standard texture analyzer testing, which have problems such as strong subjectivity and low efficiency, and it is difficult to meet the needs of fast and accurate evaluation.

[0187] To this end, the team decided to fully implement the fried potato chips product crispness assessment method proposed in the present invention. The specific implementation steps are as follows:

[0188] Step S10: collecting surface image data, vibration response data, and thermal infrared data of the fried potato chips product as original evaluation data.

[0189] To ensure the accuracy and repeatability of data collection, the team purchased the following instruments and equipment:

[0190] Industrial camera: Basler acA2500-14gm, with a resolution of 2592×1944 pixels (about 5 million pixels), GigE interface, and an imaging frame rate of up to 14 frames per second. The camera is equipped with a 16mm fixed-focus lens with an F1.4 specification, which can clearly capture the detailed texture of the product surface.

[0191] Miniature accelerometer: Model: PCB 356A01, range: ±500g, sensitivity: 10mV / g, frequency response range: 0.5-10,000Hz. This sensor is installed at the center of the product surface and can accurately measure the vibration response of the product under standard wind excitation.

[0192] Infrared imager: Model is FLIR SC7700, thermal sensitivity <0.025℃, spatial resolution <0.025mm, frame rate 7.5Hz. The instrument can capture the thermal response changes of the product surface with high resolution at an ambient temperature of 25℃.

[0193] During the acquisition process, the product was first placed under standard lighting conditions and its surface image was captured with an industrial camera; then, under standard 2m / s wind speed conditions, the vibration response of the product was measured with a micro accelerometer; finally, under an ambient temperature of 25°C, the thermal response changes of the product were recorded with an infrared imager. These rich raw data laid the foundation for subsequent feature extraction and brittleness assessment.

[0194] Step S20: performing multi-dimensional decomposition on the surface image data to extract the surface wrinkle coefficient, surface porosity coefficient, and surface texture feature coefficient.

[0195] After careful analysis, the team decided to use the following method to extract surface feature parameters:

[0196] Surface wrinkle coefficient T w :

[0197] First, the second-order gradients in the horizontal and vertical directions of each pixel in the image are calculated using the quadratic partial derivative operation, and then the absolute values ​​of the two gradient values ​​are summed to obtain the surface wrinkle coefficient:

[0198]

[0199] Among them, n and m are the number of rows and columns of the image respectively, and I(x, y) is the grayscale value of the image.

[0200] Surface porosity coefficient P r :

[0201] First, the image is binarized, and the pixels with brightness below a certain threshold are marked as black, and the rest are marked as white. Then the proportion of black pixels to the total number of pixels is counted to obtain the surface porosity coefficient:

[0202]

[0203] Among them, B(i,j) is the binary image matrix. When the pixel at (i,j) is black, B(i,j)=1, otherwise B(i,j)=0.

[0204] Surface texture characteristic coefficient F t :

[0205] The Sobel operator is used to detect the edge of the image and the sum of the absolute values ​​of the gradient values ​​in the horizontal and vertical directions is calculated to obtain the surface texture feature coefficient:

[0206]

[0207] Among them, G x (i,j) and G y(i, j) represents the horizontal and vertical gradient values ​​at (i, j) respectively.

[0208] Through the above methods, the team successfully extracted three key parameters representing the surface characteristics of the product, laying the foundation for subsequent brittleness assessment.

[0209] Step S30: using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under standard wind blowing excitation, and obtaining the vibration steady-state component coefficient and the vibration transient component coefficient through multivariate mixed decomposition calculation.

[0210] In order to accurately extract the steady-state and transient characteristics of the vibration response data, the team adopted the following methods:

[0211] First, the original vibration response data V(t) is decomposed into steady-state vibration components V s (t), transient vibration component V t (t) and the noise component η(t):

[0212] V(t)=V s (t)+V t (t)+η(t);

[0213] Among them, V s (t) can be expressed as the superposition of multiple sinusoidal components:

[0214]

[0215] A k 、f k and φ k represent the amplitude, frequency and phase angle of the kth sinusoidal component respectively.

[0216] V t (t) can be expressed as an exponential decay function:

[0217]

[0218] B l 、f l ,θ l and α l They represent the amplitude, frequency, phase angle and attenuation coefficient of the lth exponential decay component respectively.

[0219] Finally, the extracted amplitude coefficient A k and B l , frequency f k and f l , phase angle φ k and θ l , and the attenuation coefficient α l Combined into the vibration steady-state component coefficient Vs and the vibration transient component coefficient V t , providing key input for subsequent brittleness assessment.

[0220] Step S40: using an infrared imager to measure the thermal response data of the fried potato chips product under standard temperature difference conditions to obtain a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix.

[0221] In order to extract the key parameters reflecting the heat conduction characteristics, the team adopted the following method:

[0222] First, a two-dimensional heat conduction partial differential equation model is established to describe the spatiotemporal distribution of thermal response data T(x, y, t):

[0223]

[0224] Where D is the thermal diffusion coefficient and Q(x,y,t) is the heat source term.

[0225] Then, the thermal response data is decomposed using the zero-order Bessel function J0 to obtain the thermal diffusion coefficient D and the thermal attenuation coefficient β:

[0226]

[0227] λ and γ are eigenvalues.

[0228] Finally, the thermal diffusion coefficient D and thermal attenuation coefficient β are matrixed on a two-dimensional plane to form a thermal diffusion coefficient matrix D and a thermal attenuation coefficient matrix β, which provide necessary input for subsequent brittleness evaluation.

[0229] Steps S50-S70: Establishing a mapping relationship between surface features, vibration features, thermal response features and brittleness parameters.

[0230] Based on the various characteristic parameters extracted in the previous steps, the team used the least squares method to fit three mapping matrices, which correspond to the relationship between surface characteristics, vibration characteristics, thermal response characteristics and brittleness parameters:

[0231] The first mapping matrix M1:

[0232]

[0233] Among them, a ij are unknown coefficients, corresponding to the surface wrinkle coefficients T w , surface porosity coefficient P r and surface texture feature coefficient F t The mapping relationship between the brittleness parameter C.

[0234] The second mapping matrix M2:

[0235]

[0236] Among them, b ij are unknown coefficients, corresponding to the steady-state component coefficients V s and the vibration transient component coefficient V t The mapping relationship between the brittleness parameter C.

[0237] The third mapping matrix M3:

[0238]

[0239] Among them, c ij are unknown coefficients, corresponding to the mapping relationship between the thermal diffusion coefficient matrix D and the thermal attenuation coefficient matrix β and the brittleness parameter C respectively.

[0240] By establishing the above three mapping matrices, the team achieved the quantitative conversion from multi-dimensional characteristic parameters to brittleness assessment, laying the foundation for subsequent comprehensive evaluation.

[0241] Step S80: Substitute the first mapping matrix, the second mapping matrix, and the third mapping matrix into the brittleness evaluation empirical equation group to calculate the comprehensive brittleness score.

[0242] Based on the three mapping matrices established in the previous steps, the team constructed the following empirical equation for brittleness assessment:

[0243]

[0244] Among them, w1, w2 and w3 are the weight coefficients of the three characteristic parameters respectively, and ∈ is the error term.

[0245] Through repeated experiments, the team finally determined the specific parameter values ​​of the brittleness assessment empirical equation:

[0246] w1=0.4,w2=0.3,w3=0.3;

[0247] This indicates that surface characteristics have the greatest impact on brittleness in this evaluation, with dynamic response characteristics and thermal conductivity also playing an important role.

[0248] With the above empirical equation for brittleness assessment, the team can quickly calculate the comprehensive brittleness score C of the product based on the various characteristic parameters extracted in the previous steps. This score can be directly used to reflect the overall brittleness of the product.

[0249] Step S90: classifying the fried potato chip product into crispness grades based on the comprehensive crispness score, and correcting the evaluation result with the standard texture analyzer measurement data, and updating the mapping matrix parameters in the crispness evaluation empirical equation group.

[0250] In order to better interpret and apply the evaluation results, the team divided the brittleness score C into the following five levels:

[0251] Table 1 Brittleness grade classification standards

[0252] grade Crispness score range excellent C≥90 good 80≤C<90 generally 70≤C<80 Poor 60≤C<70 Very bad C<60

[0253] Through this grading, production personnel can intuitively understand the brittleness of the product and take targeted improvement measures.

[0254] At the same time, the team also organized personnel to use a standard texture analyzer to conduct destructive tests on a batch of products, and obtained three brittleness characteristic parameters: maximum breaking force, linear region slope, and breaking energy. These test data were compared and corrected with the above brittleness assessment results, and it was found that there was a certain error between the two. Therefore, the team adjusted the three mapping matrix parameters a in the brittleness assessment empirical equation. ij 、b ij and c ij It has been optimized and updated to better reflect the actual brittleness characteristics.

[0255] Through the above steps, Company A successfully established a crispness evaluation system for fried potato chips based on multi-dimensional feature fusion. Compared with traditional methods, this evaluation method has the following advantages:

[0256] 1. Comprehensiveness: By collecting and analyzing surface images, vibration responses and thermal infrared data, it is possible to deeply explore the internal structural characteristics, dynamic mechanical properties and thermal conductivity characteristics of the product, providing multi-dimensional information support for brittleness assessment.

[0257] 2. Accuracy: A quantitative mapping relationship between characteristic parameters and brittleness is established, which overcomes the limitations of traditional subjective judgment and can more accurately quantify the actual brittleness of the product.

[0258] 3. Efficiency: The use of optimization methods such as principal component analysis has greatly improved the computational efficiency of the evaluation, enabling fast, online brittleness monitoring.

[0259] 4. Reliability: It integrates standard test data and continuously corrects and updates the evaluation results to ensure the accuracy and adaptability of the evaluation.

[0260] It should be noted that the explanations of the variables involved in the present invention are shown in Table 2 below:

[0261] Table 2 Variable explanation table

[0262]

[0263]

[0264] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for evaluating the crispness of fried potato chips, characterized in that: The following steps are involved: S10, collecting surface image data, vibration response data, and thermal infrared data of the fried potato chips product as original evaluation data; S20, performing multi-dimensional decomposition on the surface image data to extract a surface wrinkle coefficient, a surface porosity coefficient, and a surface texture characteristic coefficient; S30, using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under standard wind blowing excitation, and obtaining the vibration steady-state component coefficient and the vibration transient component coefficient through multivariate mixed decomposition calculation; S40, using an infrared imager to measure the thermal response data of the fried potato chips product under standard temperature difference conditions to obtain a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix; S50, establishing a first mapping matrix between the surface wrinkle coefficient, the surface porosity coefficient, the surface texture characteristic coefficient and the brittleness parameter; S60, establishing a second mapping matrix between the vibration steady-state component coefficient, the vibration transient component coefficient and the brittleness parameter; S70, establishing a third mapping matrix between the thermal diffusion coefficient matrix, the thermal attenuation coefficient matrix and the brittleness parameter; S80, substituting the first mapping matrix, the second mapping matrix, and the third mapping matrix into a group of empirical equations for brittleness assessment to calculate a comprehensive brittleness score; S90, classifying the fried potato chip product into crispness grades based on the comprehensive crispness score, and correcting the evaluation result with the standard texture analyzer measurement data, and updating the mapping matrix parameters in the crispness evaluation empirical equation group.

2. The method for evaluating the crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S10 includes: Step 101: using an industrial camera to obtain image data of the surface of the fried potato chips product under standard light source conditions, wherein the resolution of the industrial camera is not less than 12 million pixels; Step 102: using a micro acceleration sensor to measure the vibration response data of the fried potato chips product under the action of standard wind blowing excitation, wherein the wind speed of the standard wind blowing excitation is 2 meters per second; Step 103: Use an infrared imager to measure the thermal response data of the fried potato chips product at an ambient temperature of 25 degrees Celsius, wherein the temperature resolution of the infrared imager is not less than 0.1 degrees Celsius.

3. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S20 includes: Step 201: perform gradient calculation on the surface image data using a quadratic partial derivative operation, sum the second-order partial derivatives in the horizontal and vertical directions of each pixel in the surface image data, and obtain a surface wrinkle coefficient; Step 202: binarize the surface image data, calculate the ratio of the number of black pixels to the total number of pixels in the surface image data, and obtain a surface porosity coefficient; Step 203: Use a gradient operator to perform edge detection on the surface image data, calculate the sum of the absolute values ​​of the gradient values ​​in the horizontal direction and the vertical direction, and obtain the surface texture feature coefficient.

4. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S30 includes: Step 301, decomposing the vibration response data into a steady-state vibration component, a transient vibration component and a noise component; Step 302: Decompose the steady-state vibration component by Fourier series expansion to obtain amplitude coefficients, frequencies and phase angle parameters of multiple sinusoidal components; Step 303: fitting the transient vibration component using an exponential decay function to obtain an attenuation coefficient, an amplitude coefficient and a phase angle parameter; Step 304: Combine the amplitude coefficient, the frequency, the phase angle parameter and the attenuation coefficient to form a vibration steady-state component coefficient and a vibration transient component coefficient.

5. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S40 includes: Step 401, establishing a two-dimensional heat conduction partial differential equation model, and substituting the thermal response data into the two-dimensional heat conduction partial differential equation model; Step 402: Decompose the thermal response data using a zero-order Bessel function to obtain a thermal diffusion coefficient and a thermal attenuation coefficient; Step 403: Matrix the thermal diffusion coefficient and the thermal attenuation coefficient on a two-dimensional plane to form a thermal diffusion coefficient matrix and a thermal attenuation coefficient matrix.

6. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S50 includes: Step 501, establishing a first mapping matrix with 3 rows and 3 columns, wherein the row vectors of the first mapping matrix correspond to the surface wrinkle coefficient, the surface porosity coefficient, and the surface texture characteristic coefficient respectively; Step 502: Fit the coefficient parameters in the first mapping matrix by the least square method; Step 503: Map the surface wrinkle coefficient, the surface porosity coefficient, and the surface texture feature coefficient to the brittleness parameter space through the first mapping matrix.

7. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S60 includes: Step 601, establishing a second mapping matrix with 2 rows and 2 columns, wherein the row vectors of the second mapping matrix correspond to the coefficients of the steady-state component of vibration and the coefficients of the transient component of vibration respectively; Step 602: Fit the coefficient parameters in the second mapping matrix by least square method; Step 603: Map the vibration steady-state component coefficient and the vibration transient component coefficient to the brittleness parameter space through the second mapping matrix.

8. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S70 includes: Step 701, establishing a third mapping matrix with 2 rows and 2 columns, wherein the row vectors of the third mapping matrix correspond to the thermal diffusion coefficient matrix and the thermal attenuation coefficient matrix respectively; Step 702: Fit the coefficient parameters in the third mapping matrix by the least square method; Step 703: Map the thermal diffusion coefficient matrix and the thermal attenuation coefficient matrix to the brittleness parameter space through the third mapping matrix.

9. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S80 includes: Step 801: assigning weight coefficients to the first mapping matrix, the second mapping matrix, and the third mapping matrix respectively; Step 802, multiplying the surface feature coefficient, the vibration component coefficient, and the thermal response coefficient with corresponding mapping matrices respectively; Step 803: multiply the product result by the corresponding weight coefficient and then sum them to obtain a comprehensive brittleness score.

10. The method for evaluating crispness of fried potato chips according to claim 1, characterized in that: The specific implementation of step S90 includes: Step 901: establishing a brittleness grade classification standard based on the comprehensive brittleness score, and classifying the brittleness grade into five grades; Step 902: using the standard texture analyzer measurement data to calibrate the crispness grade classification standard; Step 903: Update the coefficient parameters and the weight coefficients in the first mapping matrix, the second mapping matrix, and the third mapping matrix according to the correction result.

Images