A fusion spectrum detection method for minced meat gel quality under ultrasonic treatment

Abstract

The present invention discloses a fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment, comprising: preparing a minced meat sample after ultrasonic treatment; determining the gel strength of the minced meat sample; obtaining the spectral information of the minced meat sample, and pre-processing the spectral information to obtain the spectral information after pre-processing; constructing a prediction model for the initial minced meat gel quality, inputting the gel strength of the minced meat sample and the spectral information after pre-processing into the prediction model for the initial minced meat gel quality, and obtaining the prediction model for the minced meat gel quality; inputting the minced meat sample to be predicted into the prediction model for the minced meat gel quality, and obtaining the detection result of the minced meat gel quality. The present invention effectively captures the close relationship between the minced meat gel strength and the near-infrared-Raman fusion spectrum, and finally realizes the non-destructive, rapid and accurate detection of the minced meat gel quality.

Description

Technical Field

[0001] The invention belongs to the technical field of food quality detection, and in particular relates to a fusion spectrum detection method for minced meat gel quality under ultrasonic treatment. Background Art

[0002] In recent years, the poultry industry has shown significant economic growth, which is mainly attributed to the increasing demand for processed meat products. For deeply processed gel-type minced meat products, the gelation quality directly determines the quality and output of the product. Gel strength, as an important indicator for evaluating the gel quality of these products, affects the taste and overall consumer acceptance. However, conventional preparation methods often produce minced meat products of varying quality and low yield. Therefore, improving the preparation process is currently a key breakthrough in improving the quality of minced meat gel. The present invention introduces ultrasonic treatment technology in the preparation process. Ultrasonic technology, as a non-thermal treatment method that can improve food quality through cavitation effect, is continuously invested in food processing today. Ultrasonic technology can improve the gelation ability of protein systems, and change the conformation and intermolecular interaction of proteins through the mechanical effect generated by sound waves during propagation, so as to promote protein cross-linking, aggregation and binding, thereby enhancing the formation of meat protein networks. Although the benefits of ultrasonic treatment are obvious, it is still a major challenge to evaluate the impact of these improvements on the quality of minced meat gel. Traditional detection methods for minced meat gel quality are slow and destructive, and are not suitable for real-time quality control. And as the industry strives to meet growing consumer demand, there is now an urgent need for efficient, accurate and non-destructive quality testing methods. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a fusion spectrum detection method for minced meat gel quality under ultrasonic treatment, adopts near-infrared spectrum and Raman spectrum data fusion strategy, and establishes a long short-term memory neural network model to achieve rapid and non-destructive detection of minced meat gel quality.

[0004] To achieve the above object, the present invention provides a fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment, comprising:

[0005] Preparation of sonicated minced meat samples;

[0006] determining the gel strength of the minced meat sample;

[0007] Acquiring spectral information of the minced meat sample, and preprocessing the spectral information to obtain preprocessed spectral information;

[0008] Constructing a prediction model for the initial minced meat gel quality, inputting the gel strength of the minced meat sample and the pre-treated spectral information into the prediction model for the initial minced meat gel quality, and obtaining the prediction model for the minced meat gel quality;

[0009] The minced meat sample to be predicted is input into the prediction model of the minced meat gel quality to obtain the detection result of the minced meat gel quality.

[0010] According to the fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment provided by the present invention, the method for preparing the minced meat sample after ultrasonic treatment includes:

[0011] Fresh lean meat is taken, ice water, edible salt and sucrose are added to the fresh lean meat, the mixture is placed in a meat grinder to be minced to obtain raw minced meat, the raw minced meat is placed in a centrifuge tube for centrifugation and then subjected to ultrasonic treatment, and the raw minced meat after ultrasonic treatment is heated in a constant temperature water bath to obtain a minced meat sample after ultrasonic treatment.

[0012] According to the fusion spectrum detection method for meat paste gel quality under ultrasonic treatment provided by the present invention, in the process of adding ice water, edible salt and sucrose to the fresh lean meat for mixing, the mass ratio of ice water, edible salt and sucrose ranges from 100:20:2:1 to 100:20:3:1.

[0013] According to the fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment provided by the present invention, the raw minced meat is put into a centrifuge tube and centrifuged, and 35 g to 40 g of the raw minced meat is taken into the centrifuge tube each time, and centrifuged at 3000 r / min to 3600 r / min for 5 min to 10 min.

[0014] According to the fusion spectrum detection method for meat paste gel quality under ultrasonic treatment provided by the present invention, the raw meat paste is placed in a centrifuge tube and then centrifuged and then subjected to ultrasonic treatment. The ultrasonic power range is 220W to 240W, and the ultrasonic time range is 0min to 50min.

[0015] According to the fusion spectrum detection method for the gel quality of minced meat under ultrasonic treatment provided by the present invention, the method for determining the gel strength of the minced meat sample includes:

[0016] The gel strength of the minced meat sample was determined by a traditional texture analyzer method. A minced meat sample with a height of 1.5 cm to 2.5 cm was cut. The speed range before the test was 0.5 mm / s to 2 mm / s, the speed range during the test was 1 mm / s to 2 mm / s, the speed range after the test was 1 mm / s to 3 mm / s, and the trigger force range was 4 g to 8 g.

[0017] According to the fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment provided by the present invention, the method for obtaining the spectrum information of the minced meat sample includes:

[0018] The spectral information of the minced meat sample is obtained by a portable near-infrared acquisition spectrometer and a fully automatic confocal micro-Raman spectrometer, wherein the spectral information includes near-infrared spectral information and Raman spectral information.

[0019] According to the fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment provided by the present invention, the preparation work for obtaining the near-infrared spectrum information includes:

[0020] Adjust the integration time range to 800ms to 1000ms, the multiple scan average value range to 3 to 5, and the sliding average width value range to 4 to 6.

[0021] According to the fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment provided by the present invention, the preparation work for obtaining the Raman spectrum information includes:

[0022] The acquisition time ranged from 25 s to 32 s, the spectrum curve smoothing index ranged from 12 to 15, and the maximum number of fitting points ranged from 85 to 95.

[0023] According to the fusion spectrum detection method for minced meat gel quality under ultrasonic treatment provided by the present invention, the prediction model of the initial minced meat gel quality is constructed by adopting a long short-term memory neural network, and the prediction model of the initial minced meat gel quality includes 1 input layer, 2 to 4 LSTM layers, 1 to 3 Relu activation layers, 1 to 3 fully connected layers and 1 output layer, wherein the LSTM layer includes a forget gate, an input gate and an output gate; the long short-term memory neural network adopts the Adam gradient descent algorithm, the maximum number of iterations ranges from 400 to 800, and the initial learning rate ranges from 0.01 to 0.02.

[0024] Technical effect of the invention: The invention discloses a fusion spectrum detection method for meat mince gel quality under ultrasonic treatment. The invention introduces ultrasonic technology into the preparation process, which can significantly improve the gel quality of meat mince compared with the conventional preparation process; a single spectral detection method may not be able to comprehensively evaluate the complex gel strength information of meat mince, and the near-infrared spectrum and Raman spectrum are integrated to perform comprehensive analysis to improve accuracy; the long short-term memory neural network model adopted by the invention is more suitable for processing time series data than other models, and can effectively capture the close connection between gel strength data and near-infrared-Raman fusion spectrum data, thereby realizing non-destructive, rapid and accurate detection of meat mince gel quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0026] Figure 1This is a schematic diagram of a flow chart of a fusion spectrum detection method for meat paste gel quality under ultrasonic treatment according to an embodiment of the present invention;

[0027] Figure 2 This is a schematic diagram of low-level data fusion of near-infrared and Raman spectra of chicken paste according to an embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of the operation process of the long short-term memory neural network model according to an embodiment of the present invention;

[0029] Figure 4 This is a schematic diagram of the prediction results of the long short-term memory neural network model for the gel strength of chicken paste in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0031] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0032] like Figure 1 As shown, this embodiment provides a fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment, comprising the following steps:

[0033] Preparation of sonicated minced meat samples;

[0034] Further, the method for preparing the meat paste sample after ultrasonic treatment includes:

[0035] Take fresh lean meat, add ice water, edible salt and sucrose to the fresh lean meat, mix the mixture, put it in a meat grinder to mince it to obtain raw minced meat, put the raw minced meat into a centrifuge tube, centrifuge it and then perform ultrasonic treatment, and heat the raw minced meat after ultrasonic treatment in a constant temperature water bath to obtain a minced meat sample after ultrasonic treatment.

[0036] Specifically, in the process of adding ice water, edible salt and sucrose to fresh lean meat, the mass ratio of ice water, edible salt and sucrose ranges from 100:20:2:1 to 100:20:3:1. In the process of centrifuging the raw minced meat into a centrifuge tube, 35g to 40g of raw minced meat is taken into the centrifuge tube each time, and centrifuged at 3000r / min to 3600r / min for 5min to 10min. In the process of ultrasonic treatment after the raw minced meat is placed in the centrifuge tube and centrifuged, the ultrasonic power ranges from 220W to 240W, and the ultrasonic time ranges from 0min to 50min.

[0037] Preferably, the mass ratio of ice water, edible salt and sucrose is 100:20:2:1 to 100:20:3:1. This mass ratio range can not only maintain the moistness of the meat, improve the moisture content and flavor of the minced meat, but also enhance the solubility and emulsification ability of the protein, making it easier for the minced meat to form a stable gel structure during subsequent processing.

[0038] Preferably, the centrifugal rate is 3000 r / min to 3600 r / min and the centrifugal time is 5 min to 10 min. Within the range of the centrifugal rate and the centrifugal time, the air bubbles in the minced meat can be more effectively removed, thereby improving the density and uniformity of the minced meat.

[0039] Preferably, the ultrasonic power is 220W to 240W and the ultrasonic time is 0min to 50min. Within this ultrasonic power and ultrasonic time range, the minced meat can be effectively micro-transformed by ultrasonic high-frequency vibration, thereby promoting protein dissociation and reorganization in the minced meat and enhancing the gelling ability.

[0040] Determination of gel strength of minced meat samples;

[0041] Further, the method for determining the gel strength of the minced meat sample comprises:

[0042] The gel strength of the minced meat samples was determined by the traditional texture analyzer method. Minced meat samples with a height of 1.5 cm to 2.5 cm were cut. The speed range before the test was 0.5 mm / s to 2 mm / s, the speed range during the test was 1 mm / s to 2 mm / s, the speed range after the test was 1 mm / s to 3 mm / s, and the trigger force range was 4 g to 8 g.

[0043] Preferably, the minced meat sample is 1.5 cm to 2.5 cm high, and this height range can prevent measurement errors caused by the minced meat sample being too thin or too thick.

[0044] Preferably, the rate before the test is 0.5 mm / s to 2 mm / s, the rate during the test is 1 mm / s to 2 mm / s, and the rate after the test is 1 mm / s to 3 mm / s. Within these rate ranges, accurate gel strength can be obtained during the measurement process. On the one hand, the rupture of the minced meat sample due to too fast a rate and the excessive deformation of the minced meat sample due to too slow a rate can be avoided. On the other hand, the release process of the minced meat sample after the pressure is applied is ensured to be smooth, thereby reducing the sample rebound or rebound effect caused by the sudden release force.

[0045] Preferably, the trigger force is 4 g to 8 g. This trigger force range can effectively eliminate the influence of tiny surface irregularities on the test results and ensure that the starting conditions for gel strength measurement are consistent.

[0046] Acquiring spectral information of the minced meat sample, and preprocessing the spectral information to obtain preprocessed spectral information;

[0047] Furthermore, the method for obtaining spectral information of the minced meat sample includes:

[0048] The spectral information of the minced meat sample is obtained by a portable near-infrared acquisition spectrometer and a fully automatic confocal micro-Raman spectrometer, wherein the spectral information includes near-infrared spectral information and Raman spectral information.

[0049] Specifically, the preparation work for obtaining near-infrared spectral information includes:

[0050] Adjust the integration time range to 800ms to 1000ms, the multiple scan average value range to 3 to 5, and the sliding average width value range to 4 to 6.

[0051] Preferably, the integration time is 800 ms to 1000 ms, within which the photon signal can be ensured to be fully accumulated, so that the acquired spectral data is clearer and more accurate.

[0052] Preferably, the average value of multiple scans is 3 to 5. This average value range can help improve the reproducibility and signal quality of the spectrum, making the subsequent quantitative analysis more reliable and accurate.

[0053] Preferably, the sliding average width is 4 to 6, and within this sliding average width range, the requirements of noise suppression and signal retention can be balanced, making the spectral signal clearer and more identifiable.

[0054] The preparation work for obtaining Raman spectroscopy information includes:

[0055] The acquisition time ranged from 25 s to 32 s, the spectrum curve smoothing index ranged from 12 to 15, and the maximum number of fitting points ranged from 85 to 95.

[0056] Preferably, the acquisition time is 25 s to 32 s, which allows a sufficiently long exposure time to enhance the intensity of the spectral signal while avoiding drift or other time-dependent errors caused by too long a time, so that the final spectral data is stable and accurate.

[0057] Preferably, the spectrum curve smoothing index is 12 to 15. Within the range of the spectrum curve smoothing index, the interference of high-frequency noise on the spectrum data can be effectively reduced, and the main characteristic peaks of the spectrum can be clearly displayed.

[0058] Preferably, the maximum number of fitting points is 85 to 95, within which the fitting accuracy of the spectral data can be optimized, ensuring that the detailed features of the spectral curve are fully preserved, and improving the reliability and accuracy of the analysis results.

[0059] constructing a prediction model for the initial minced meat gel quality, inputting the gel strength of the minced meat sample and the spectral information after pretreatment into the prediction model for the initial minced meat gel quality, and obtaining the prediction model for the minced meat gel quality;

[0060] Furthermore, the prediction model of initial minced meat gel quality is constructed using a long short-term memory neural network, which includes 1 input layer, 2 to 4 LSTM layers, 1 to 3 Relu activation layers, 1 to 3 fully connected layers and 1 output layer, wherein the LSTM layer contains a forget gate, an input gate and an output gate; the long short-term memory neural network adopts the Adam gradient descent algorithm, with a maximum number of iterations ranging from 400 to 800 and an initial learning rate ranging from 0.01 to 0.02.

[0061] Preferably, there are 2 to 4 LSTM layers, 1 to 3 Relu activation layers, and 1 to 3 fully connected layers. The LSTM layer range can deeply explore the time series relationship of the data, especially the long-term and short-term dependencies, thereby improving the model's predictive ability for the quality of minced meat gel; the Relu activation layer range can more efficiently capture the complex nonlinear relationships in the data and enhance the expressive power of the model; the fully connected layer range can appropriately increase the model complexity, further optimize the accuracy and stability of the prediction, and make it better fit the data, thereby improving the prediction performance.

[0062] Preferably, the maximum number of iterations is 400 to 800. This maximum number of iterations can ensure that the generalization ability and stability of the prediction model are improved while saving model training time.

[0063] Preferably, the initial learning rate is 0.01 to 0.02. This initial learning rate range can ensure that the model can quickly find a better direction in the early stage of training, improve the overall training efficiency, and avoid the problem of unstable convergence caused by too high a learning rate.

[0064] The minced meat sample to be predicted is input into the prediction model of the minced meat gel quality to obtain the detection result of the minced meat gel quality.

[0065] A specific application example of the present invention is implemented in the following steps:

[0066] Take the pure lean meat part of fresh chicken breast, mince the lean meat with a meat grinder, and then mix the lean meat: ice water: edible salt: sucrose in a mass ratio of 100:20:2.4:1, and mince it again with a meat grinder to obtain raw minced meat. Take 40g of raw minced meat and put it into a centrifuge tube, centrifuge it at 3600r / min and 4℃ for 10min to remove bubbles. Then put it into an ultrasonic cleaner and ultrasonicate it at 240W for 0, 10, 20, 30, 40, and 50min respectively. Finally, heat it in a constant temperature water bath at 78℃ for 40min to obtain a chicken minced meat sample.

[0067] After the preparation of chicken mince, the gel strength of chicken mince was measured using a TA.XT PlusC texture analyzer. The specific steps are as follows: cut a chicken mince sample with a height of 12 mm, select a P / 0.5 probe, and set the parameters as follows: 2 mm / s before the test, 1.5 mm / s during the test, 2 mm / s after the test, 5 g trigger force, 10 mm penetration distance, and the product of the breaking distance and the breaking force is the gel strength.

[0068] The spectral information of the chicken mince sample was collected by a portable near-infrared acquisition spectrometer and a fully automatic confocal micro-Raman spectrometer, and the spectra were processed separately. Finally, low-level data fusion was performed. Figure 2 The specific steps are as follows: When collecting near-infrared spectral information, set the wavelength range to 950-1700cm in the Overview software. -1 , integration time 850ms, multiple scan average value 3, sliding average width value 5; when collecting Raman spectrum information, set the Raman shift range 400-3200cm in the labspec6 software -1 , excitation wavelength 532nm, acquisition time 31s, preprocessing method: spectral curve smoothing index 15, maximum number of points for fitting 8. Finally, the near-infrared spectra (128 groups) and Raman spectra (796 groups) were fused into near-infrared-Raman spectra (924 groups) through a low-level fusion strategy.

[0069] The measured chicken paste gel strength value was combined with the corresponding near-infrared-Raman fusion spectrum data, with 80 training sets and 40 prediction sets. The long short-term memory neural network was constructed using the software MATLAB 2023B. The long short-term memory neural network can efficiently mine the close connection between the gel strength data and the near-infrared-Raman fusion spectrum data. The training set and the prediction set were imported into the long short-term memory neural network. Among them, the long short-term memory neural network contains 1 input layer (input size is 924), 1 LSTM layer (the number of hidden units is 4), 1 Relu activation layer, 1 fully connected layer and 1 regression layer. Figure 3As shown. Adopt the Adam gradient descent algorithm, set the maximum number of iterations to 400, and the initial learning rate to 0.01. Finally, run the prediction results as shown Figure 4 As shown, the training correlation coefficient of chicken paste gel strength is Rc=0.9855, RMSEC=44.2715 (g×mm), the prediction correlation coefficient is Rp=0.9694, RMSEP=63.3595 (g×mm), and RPD=4.1255, indicating that the model has strong prediction accuracy.

[0070] The invention discloses a fusion spectrum detection method for meat mince gel quality under ultrasonic treatment. The invention introduces ultrasonic technology into the preparation process, and can significantly improve the meat mince gel quality compared with the conventional preparation process; a single spectrum detection method may not be able to comprehensively evaluate the complex gel strength information of meat mince, and near-infrared spectrum and Raman spectrum are fused to perform comprehensive analysis to improve accuracy; the long short-term memory neural network model adopted by the invention is more suitable for processing time series data than other models, and can effectively capture the close connection between gel strength data and near-infrared-Raman fusion spectrum data, so as to realize non-destructive, rapid and accurate detection of meat mince gel quality.

[0071] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment, characterized in that: include: Preparation of sonicated minced meat samples; The method for preparing the meat paste sample after ultrasonic treatment includes: Take fresh lean meat, add ice water, edible salt and sucrose to the fresh lean meat, mix the mixture, put it in a meat grinder and grind it to obtain raw minced meat, put the raw minced meat into a centrifuge tube, centrifuge it and then perform ultrasonic treatment, and heat the raw minced meat after ultrasonic treatment in a constant temperature water bath to obtain a minced meat sample after ultrasonic treatment; determining the gel strength of the minced meat sample; Acquiring spectral information of the minced meat sample, and preprocessing the spectral information to obtain preprocessed spectral information; The method for obtaining the spectral information of the minced meat sample comprises: Acquiring spectral information of the minced meat sample by a portable near-infrared acquisition spectrometer and a fully automatic confocal micro-Raman spectrometer, wherein the spectral information includes near-infrared spectral information and Raman spectral information; Constructing a prediction model for the initial minced meat gel quality, inputting the gel strength of the minced meat sample and the pre-treated spectral information into the prediction model for the initial minced meat gel quality, and obtaining the prediction model for the minced meat gel quality; The prediction model of the initial minced meat gel quality is constructed by using a long short-term memory neural network, and the prediction model of the initial minced meat gel quality includes 1 input layer, 2 to 4 LSTM layers, 1 to 3 Relu activation layers, 1 to 3 fully connected layers and 1 output layer, wherein the LSTM layer includes a forget gate, an input gate and an output gate; the long short-term memory neural network adopts an Adam gradient descent algorithm, the maximum number of iterations ranges from 400 to 800, and the initial learning rate ranges from 0.01 to 0.02; The minced meat sample to be predicted is input into the prediction model of the minced meat gel quality to obtain the detection result of the minced meat gel quality.

2. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: In the process of adding ice water, edible salt and sucrose to the fresh lean meat and mixing them, the mass ratio of ice water, edible salt and sucrose ranges from 100:20:2:1 to 100:20:3:

1.

3. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: During the centrifugation of the raw minced meat into a centrifuge tube, 35 g to 40 g of the raw minced meat is taken out each time and put into the centrifuge tube, and centrifuged at 3000 r / min to 3600 r / min for 5 min to 10 min.

4. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: The raw minced meat is placed in a centrifuge tube and then centrifuged and then subjected to ultrasonic treatment. The ultrasonic power ranges from 220 W to 240 W, and the ultrasonic time ranges from 0 min to 50 min.

5. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: The method for determining the gel strength of the minced meat sample comprises: The gel strength of the minced meat sample was determined by a traditional texture analyzer method. A minced meat sample with a height of 1.5 cm to 2.5 cm was cut. The speed range before the test was 0.5 mm / s to 2 mm / s, the speed range during the test was 1 mm / s to 2 mm / s, the speed range after the test was 1 mm / s to 3 mm / s, and the trigger force range was 4 g to 8 g.

6. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: The preparation work for obtaining the near infrared spectrum information includes: Adjust the integration time range to 800ms to 1000ms, the multiple scan average value range to 3 to 5, and the sliding average width value range to 4 to 6.

7. The fusion spectrum detection method for the quality of minced meat gel under ultrasonic treatment according to claim 1, characterized in that: The preparation work for obtaining the Raman spectrum information includes: The acquisition time ranged from 25 s to 32 s, the spectrum curve smoothing index ranged from 12 to 15, and the maximum number of fitting points ranged from 85 to 95.

Images