Porcine heart blood quality rapid evaluation method based on character, ratio shift value and machine learning

Abstract

The invention relates to a method for detecting pig heart blood quality based on a colorimetric value and an amino acid specific shift value of pig heart blood, which comprises the following steps: acquiring RGB colorimetric values by using digital camera equipment, and quickly detecting the amino acid specific shift value by using thin-layer chromatography. And the problem that conventional mathematical equations cannot be established among detection values such as a chromatic value and an amino acid ratio shift value and the total solid content and the total nitrogen content of the pig heart blood is solved, and a prediction model is successfully established by adopting a machine learning algorithm. The technical problems that the pig heart blood quality cannot be effectively detected due to the fact that the pig heart blood is taken and used at present in the processing process, the storage time is short, the collection and use environment is not suitable for placement of professional detection equipment are solved.

Description

Technical Field

[0001] This patent belongs to the field of quality control of Chinese medicine excipients, and specifically involves a method for determining the total solid content and total nitrogen content in the Chinese medicine excipient pig heart blood through property quantification and multi-component thin layer Rf value combined with machine learning algorithm. Background Art

[0002] Pig heart blood is from Suidae Sus scrofa domestica Brisson The pig heart is obtained by cutting out the heart after slaughtering. It is used fresh as a Chinese medicine auxiliary material and can be stored for a short time. The collection and use environment is mainly limited to slaughterhouses and Chinese medicine processing workshops. According to ancient and modern books, pig heart blood is a traditional Chinese medicine processing auxiliary material and is considered to be a guide for heart-nourishing medicines. It can be used to prepare pig heart blood Danshen, pig heart blood fried with Ziziphus jujuba seeds, etc. Its clinical application can be traced back to the origin of the Menghe School of Medicine, Hua Tuo during the Eastern Han Dynasty and the Three Kingdoms period. His book "Hua Tuo Shen Fang" records: "The heart-nourishing pill for the treatment of heart failure due to shock or restlessness due to excessive thinking is made of cinnabar, realgar, and white aconite, mixed evenly with pig heart blood, which is as big as a sycamore seed, and is coated with cinnabar." It is also often used in compound Zhonghe pills to lead tranquilizers into the brain, which has the effect of calming the mind, soothing the nerves, and replenishing blood. It is used to treat restlessness, palpitations, and phlegm.

[0003] After being collected, pig heart blood has lost its morphological identification characteristics and has serious adulteration. Moreover, traditional properties and physical and chemical identification analysis methods cannot effectively distinguish these blood counterfeits with similar properties but different sources. At present, there is a lack of specific identification and comprehensive quality evaluation indicators for the quality control of auxiliary material pig heart blood. Considering the characteristics of pig heart blood being taken fresh, it is urgent to establish a quick and simple method for identifying the quality of pig heart blood.

[0004] The total solid content of pig heart blood is a relatively intuitive and convenient quality control indicator for liquid auxiliary pig heart blood, and is an important indicator for measuring the nutrient content of animal blood. Protein is an important pharmacologically active ingredient in pig heart blood. After catalytic heating and decomposition, protein undergoes a series of reactions and is finally titrated with a strong acid standard titration solution to convert the nitrogen content. The nitrogen content and protein in common foods can be directly converted according to the conversion coefficient, so the nitrogen content can reflect the protein content of the test sample. Therefore, the total solid content and nitrogen content of pig heart blood can be used as the quality judgment criteria for pig heart blood.

[0005] However, the "Total Solid Determination Method" included in Part IV of the 2020 edition of the "Chinese Pharmacopoeia" requires gradually evaporating the blood to a constant weight in a water bath, which takes a long time and requires repeated weighing with a precision balance, which cannot meet the requirements for real-time quality control of pig heart blood at the processing site. The "nitrogen determination method" such as the Kjeldahl titration method in the prior art requires the use of specific expensive equipment and reagents, such as a digestion furnace, a nitrogen analyzer, etc., and has high requirements for the use environment, requires a dedicated laboratory, and the determination cycle is very long, which also cannot meet the requirements for real-time quality control of pig heart blood at the processing site.

[0006] In view of the current imperfect quality control standards of pig heart blood and the complicated operation of total solid content and nitrogen content determination, the present invention introduces thin layer chromatography and spectrophotometric colorimetry technology to objectively quantify color values, formulates relevant standards, and screens the trait indicators significantly related to the internal control indicators, establishes a neural network model with good prediction ability, and establishes a model for predicting the total solid content and nitrogen content of pig heart blood by integrating multiple indicators. The method is applied to the determination and quality control of the total solid content and nitrogen content of pig heart blood, the prediction equation is reliable, and the quality of pig heart blood can be stably and quickly predicted. Summary of the invention

[0007] The purpose of the present invention is to provide a method for rapidly detecting the total solid content and nitrogen content of pig heart blood, so as to meet the requirements of fresh use and real-time quality control of pig heart blood, a Chinese medicine auxiliary material.

[0008] In the first aspect of the present invention, the inventor unexpectedly discovered that the chromaticity value of pig heart blood and the amino acid ratio transfer value Rf of pig heart blood have a certain intrinsic correlation with the quality of pig heart blood, that is, the total solid content and nitrogen content. Based on this, the present invention uses the unexpectedly discovered relationship between the chromaticity value, Rf parameter and the intrinsic quality of pig heart blood to establish a method for comprehensive, real-time and accurate pig heart blood quality control, thereby obtaining a pig heart blood quality detection method that is less affected by factors such as equipment, environment, and operators and has strong universality.

[0009] In the second aspect of the present invention, the inventors have discovered through extensive research that it is impossible to establish a conventional mathematical relationship equation between the detection values ​​of pig heart blood, such as the chromaticity value, amino acid ratio shift value Rf, and the total solid amount and total nitrogen amount of pig heart blood, that is, it is impossible to find any suitable mathematical formula that can obtain the correct total solid amount and total nitrogen amount of pig heart blood by inputting parameters such as the chromaticity value, amino acid ratio shift value Rf, etc. In order to solve this technical problem, the inventors creatively introduced a neural network model, and used machine learning to find that a PSO-BiLSTM prediction model can be successfully established between the detection values ​​of pig heart blood, such as the chromaticity value, amino acid ratio shift value Rf, and the total solid amount and total nitrogen amount of pig heart blood, and the correct total solid amount and total nitrogen amount of pig heart blood can be accurately and quickly obtained by inputting parameters such as the chromaticity value, amino acid ratio shift value Rf, thereby completing the present invention.

[0010] Specifically, the quality inspection process of the present invention is to take pictures of pig heart blood and extract and convert RGB color values ​​and Rf values ​​of each component of thin layer chromatography, and bring them into the established neural network model to quickly detect the total solid content and nitrogen content of pig heart blood.

[0011] More specifically, the present invention provides a method for detecting the total solid content and total nitrogen content of pig heart blood, characterized in that it comprises the following steps: 1) obtaining a pig heart blood sample; 2) determining the color value of the pig heart blood; 3) using thin layer chromatography to determine the amino acid ratio shift value Rf of the pig heart blood; 4) using the pig heart blood color value and the amino acid ratio shift value Rf as input parameters, and using a trained neural network model to calculate the total solid content and total nitrogen content of the pig heart blood.

[0012] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the step of obtaining the pig heart blood sample includes: promptly dissecting the pig heart after slaughter, collecting its fresh blood after pouring out and sieving, adding an equal volume of rice wine, and refrigerating and transporting and storing it at -20°C.

[0013] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the step of determining the color value of pig heart blood includes: taking a pig heart blood sample, placing it in a clean evaporating dish, taking a photo, extracting and converting the photo into RGB values, and recording data.

[0014] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the step of determining the color value of pig heart blood includes: placing an evaporating dish on a blue background plate at room temperature, using a digital camera to capture a high-definition image, and using a chromaticity value extraction code based on a pillow library to extract image color information and convert it into RGB values.

[0015] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the step of using thin layer chromatography to determine the amino acid ratio transfer value Rf of pig heart blood includes: selecting leucine, valine, alanine, and glycine as indicators for thin layer chromatography detection, using silica gel G thin layer plate, and using n-butanol-glacial acetic acid-water 3:1:1 as the developing agent.

[0016] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the step of using thin layer chromatography to determine the amino acid ratio transfer value Rf of pig heart blood includes: accurately aspirating 200 μL of pig heart blood sample, adding 400 μL of protein precipitant acetonitrile, vortex mixing and standing for 10 minutes, centrifuging at 12000 rpm / min for 10 minutes, taking the supernatant, taking appropriate amounts of leucine, valine, alanine, glycine, and lysine reference substances, adding 0.1 mol / L hydrochloric acid to prepare mixed solutions with concentrations of 0.2 mg / mL, respectively, as reference substance solutions; aspirating 10 μL each, spotting them on the same silica gel G thin layer plate, using n-butanol-glacial acetic acid-water 3:1:1 as the developing agent, developing, taking out, drying, spraying with ninhydrin test solution, heating at 105°C until the spots are clearly colored, and calculating the Rf values ​​of the five amino acid components.

[0017] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that the neural network model is a PSO-BiLSTM model, and the data training set used in its training method is 53% of the original data divided by the algorithm, the test set is 20% of the data divided by the algorithm, and the verification set is 27% of the data divided by the algorithm.

[0018] The method for detecting the total solid content and total nitrogen content of pig heart blood is characterized in that: the PSO-BiLSTM model uses four indicators to evaluate the performance of the model: mean absolute error (MAE), mean error (ME), root mean square error (RMSE), and correlation coefficient R2.

[0019] The pig heart blood computer vision extraction conversion color value and thin layer chromatography TLC technology of the present application can quickly and accurately obtain the total solid content and total nitrogen content of pig heart blood, which is suitable for the quality control of the processing auxiliary material pig heart blood. It overcomes the defects of the prior art that the nitrogen content in pig heart blood is determined by the Kjeldahl titration method and other content determination methods, which are complicated to operate, high cost, high requirements for the detection environment, and unsuitable for large-scale pig heart blood quality inspection.

[0020] At the same time, the present application can comprehensively reflect the total solid content and nitrogen content of pig heart blood through the color value of pig heart blood and the results of thin layer chromatography (TLC) inspection. First, it overcomes the technical defect of only being able to indicate the quality inspection of pig heart blood used as an auxiliary material for medicinal preparation through colorimetric values. Second, by adding the amino acid Rf value of thin layer chromatography (TLC) as one of the indicators for predicting the quality of pig heart blood, the precision and accuracy of the calculation results are significantly improved. Third, it overcomes the defect of the prior art that there is no rapid determination method for the nitrogen content and total solid content of pig heart blood, and finds a method for rapid detection of quality indicators in pig heart blood, thereby enabling real-time control of the quality of the auxiliary material pig heart blood for preparation.

[0021] In addition, since the color of pig heart blood is the most intuitive factor in judging the quality of pig heart blood, and the naked eye's judgment of color is subjective, the inventor of the present application also reduces the quality judgment deviation caused by personal subjective influence by objectively quantifying the color index. In the present invention, the colorimetry / colorimetric value strategy is a method of using digital values ​​as characteristics to reflect the surface color of Chinese medicinal materials by extracting color analysis with the help of a high-definition digital camera. Among them, R, G, and B represent the three primary colors of red, green, and blue, respectively. The average chromaticity value is extracted by homogenizing the picture. Since the quality control indicators and regulations for pig heart blood are not obvious in existing research, the inventor of the present application further predicts the total solid content and nitrogen content of pig heart blood by measuring the Rf values ​​of each component of thin layer chromatography amino acids as parameters. The chromaticity value is combined with the above-mentioned amino acid Rf value as input data to import the established PSO-BiLSTM prediction model, which can quickly predict the total solid content and nitrogen content that reflect the quality of pig heart blood.

[0022] Beneficial Effects The invention detects the color value of pig heart blood and the Rf value of amino acid components by thin layer chromatography (TLC) technology, and the range setting is reasonable. The neural network model equation for detecting the total nitrogen content and the total solid content of pig heart blood by the property index is established, and the prediction ability is good after experimental verification. The invention is applied to the quality of pig heart blood auxiliary materials, and the calculation ability is reliable, and the quality of pig heart blood can be stably and quickly determined. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 : Characteristics of 15 batches of pig heart blood samples. From left to right and then from top to bottom in order of batch numbers.

[0025] Figure 2 : Amino acid TLC identification results of 15 batches of pig heart blood samples. From left to right in the order of batch numbers, the 8th column from the left is: amino acid reference (from top to bottom, leucine, valine, alanine, glycine, lysine).

[0026] Figure 3 : The prediction process of the neural network model.

[0027] Figure 4 : Training progress of total solid content of pig heart blood. The upper figure is the RMSE graph, where the light-colored continuous curve with dots is: training, the continuous curve without dots is: training (smoothed), and the intermittent curve with dots is: validation. The lower figure is the loss graph, where the light-colored continuous curve with dots is: training, the continuous curve without dots is: training (smoothed), and the intermittent curve with dots is: validation.

[0028] Figure 5 : Training progress of pig heart blood nitrogen content. The upper figure is the RMSE graph, where the light-colored continuous curve with dots is: training, the continuous curve without dots is: training (smoothed), and the intermittent curve with dots is: validation. The lower figure is the loss graph, where the light-colored continuous curve with dots is: training, the continuous curve without dots is: training (smoothed), and the intermittent curve with dots is: validation.

[0029] Figure 6 : Prediction training set of pig heart blood nitrogen content. The dots are the true values, and the squares are the predicted values.

[0030] Figure 7 : Training set for predicting total solid content of pig heart blood. The dots are true values, and the squares are predicted values.

[0031] Figure 8 : Prediction test set of pig heart blood nitrogen content. The dots are the true values, and the squares are the predicted values.

[0032] Fig. 9 : The prediction test set of total solid content of pig heart blood. The dots are the true values, and the squares are the predicted values.

[0033] Fig.10 : Prediction and validation set of pig heart blood nitrogen content. The dots are the true values, and the squares are the predicted values.

[0034] Fig.11 : Prediction and validation set of pig heart blood nitrogen content. The dots are the true values, and the squares are the predicted values. DETAILED DESCRIPTION

[0035] The technical solution of the present invention is further explained below in conjunction with specific embodiments, which is not intended to limit the technical solution.

[0036] The sources of biological materials in the present invention are as follows: Pig heart blood was obtained from the designated slaughterhouse of Nanjing Dachang Xinle Meat Food Center, and rice wine (14% vol) was obtained from Shaoxing Jinyu Wine Co., Ltd.

[0037] Example 1 Obtaining pig heart blood Take 30g of fresh pig heart blood and 30g of rice wine and mix them well until they are completely absorbed. The source of pig heart blood is shown in Table 1.

[0038] Table 1 Source information of pig heart blood samples for processing auxiliary materials

[0039] Example 2 Conversion of 15 batches of pig heart blood properties and chromaticity values Take the pig heart blood of Example 1, place it in a clean evaporating dish, use a professional camera (Canon EOS5D) of the Resource Survey Office of Nanjing University of Chinese Medicine to take a high-definition image at room temperature, place it on a blue background board to take a photo, transfer the photo into Python software, and use the chromaticity value extraction code based on the pillow library to extract the color information and convert it into RGB value. This product is reddish brown, has good fluidity, no impurities and clots, has a slightly fishy smell, and has an alcoholic taste. It should not be dark black or have a corrupt smell. Figure 1 , RGB values ​​of 15 batches of pig heart blood are shown in Table 2.

[0040] Table 2 RGB values ​​of 15 batches of pig heart blood

[0041] Example 3 Thin layer chromatography TLC study of 15 batches of pig heart blood Use acetonitrile as the protein precipitant, accurately draw 200 μL of the pig heart blood sample obtained in Example 1, add 400 μL of acetonitrile, vortex mix and let stand for 10 minutes, centrifuge at 12000 rpm / min for 10 minutes, take the supernatant, take appropriate amounts of leucine, valine, alanine, glycine, and lysine reference substances, add 0.1 mol / L hydrochloric acid to make a mixed solution with a concentration of 0.2 mg / mL, as the reference substance solution. Draw 10 μL each and spot on the same silica gel G thin layer plate, use n-butanol-glacial acetic acid-water (3:1:1) as the developing agent, develop, take out, dry, spray with ninhydrin test solution, heat at 105°C until the spots are clearly colored, and calculate the Rf values ​​of the five amino acid components. Figure 2 The distance between each spot and the front and the baseline was measured and the Rf value was calculated. The results are shown in Table 3.

[0042] Table 3 Rf values ​​of 15 batches of pig heart blood

[0043] Example 4 Determination of the total solid content of 15 batches of pig heart blood The total solids of the auxiliary material pig heart blood were determined according to the "Total Solid Determination Method" (General Rule 3101) included in Part IV of the 2020 edition of the "Chinese Pharmacopoeia". Take the pig heart blood test sample of Example 1 and mix it evenly, accurately measure 5 mL, place it in an evaporating dish dried to constant weight, evaporate it in a water bath, and dry it under reduced pressure at 75-85°C to constant weight to obtain the total solids. The results are shown in Table 4.

[0044] Table 4 Total amount of solids from 15 batches of pig heart blood

[0045] Example 5 Determination of nitrogen content in 15 batches of pig heart blood Add water to dissolve and prepare the following reagents in appropriate amounts: boric acid solution (20 g / L), sodium hydroxide solution (400 g / L); add 95% ethanol to dissolve and prepare the following reagents in appropriate amounts: methyl red ethanol solution (1 g / L), bromocresol green ethanol solution (1 g / L), 1 part of methyl red ethanol solution and 5 parts of bromocresol green ethanol solution are mixed to prepare a mixed indicator before use. Accurately add 1 mL of the pig heart blood sample of Example 1 to the digestive tube, and then add 0.40 g of copper sulfate, 6.0 g of potassium sulfate, and 10 mL of sulfuric acid for digestion. When the temperature of the digestion furnace reaches 240°C, keep it for 1 hour, continue to heat it to 440°C and digest for 1.5 hours. At this time, the liquid in the digestive tube is green, clear and transparent, take it out and cool it, and prepare a reagent blank in the same way. The nitrogen analyzer automatically adds water and automatically distills for about 15 minutes. While distilling, add 10 mL of boric acid solution and 2 drops of mixed indicator to the conical flask, and the lower end of the condenser should be below the liquid surface. Titrate with 0.1000 mol / L hydrochloric acid standard solution until the end point is light gray-red. The nitrogen content was calculated and the results are shown in Table 5.

[0046] Table 5 Results of pig heart blood nitrogen content

[0047] Example 6 Model training and verification The present invention uses computer vision and ratio shift value differences to quantify the properties of pig heart blood and thin layer chromatography ratio shift value Rf data, and converts the original one-dimensional data into multidimensional data with 8 features for long-term and short-term memory neural network learning. Previous studies have found that traditional regression methods cannot effectively predict the quality of pig heart blood through feature data. By introducing a deep learning algorithm with strong learning ability to predict the correlation of data, after model construction and adaptive optimization of parameters, it is verified that the obtained results are good. Specific implementation method is as follows Figure 3:First, divide the results obtained from the above experiment into a test set, a training set, and a validation set for the next step. Input the training set data into the original PSO-BiLSTM algorithm for model training, and use the quality parameters of the pig heart blood samples in the test set to evaluate the model performance. If the model performance is poor after evaluation by the four indicators of R2, MAE, ME, and RMSE, return to the model training part for retraining. If the model evaluation is good, save the model parameters, and import the quality parameters of the pig heart blood samples in the validation set into the model for calculation. The calculation result is the actual prediction effect. In order to make the validation set and the test set have similar distributions and ensure the parallelism and integrity of the data, the data set is randomly divided into a training set, a test set, and a validation set. The PSO-BiLSTM model combines forward LSTM and backward LSTM, and can obtain the relationship between data points in both directions. It has unique advantages in solving nonlinear problems. At the same time, in order to further improve the prediction performance of the model, the PSO algorithm is introduced to optimize the hyperparameters of the model to improve the accuracy and precision of pig heart blood quality prediction. The specific calculation steps are as follows:

[0048] (1) Use BiLSTM to obtain the network weight matrix and apply it to determine the initial position of the initialized particles.

[0049] (2) The fitness value of the PSO algorithm is defined as the loss function of the BiLSTM based on the validation set. The fitness value is inversely proportional to the loss function value, that is, the smaller the loss function value, the stronger the adaptability of the particle.

[0050] (3) According to the PSO algorithm optimization process, the position and velocity of each particle are iteratively updated until the end condition is reached (such as reaching the maximum number of iterations). Finally, the position matrix of the global optimal solution is applied to BiLSTM to determine the weight matrix of the optimal solution. In this way, the PSO algorithm can find the optimal solution under the constraint of the maximum number of iterations.

[0051] (4) Train the BiLSTM network and iterate and update it repeatedly until the specified end condition is reached (in this article, the end condition is reaching the maximum number of iterations). Through continuous iterative updates, the BiLSTM network gradually optimizes its weight matrix to minimize the training error, thereby improving the accuracy and generalization ability of the model.

[0052] (5) After the training and optimization in the above steps, the network weight matrix needs to be saved for subsequent testing. In this way, we can evaluate the performance of the model on the test set and obtain indicators such as the accuracy and generalization ability of the model. At the same time, saving the model can also facilitate us to adjust and optimize the model.

[0053] The present invention uses four indicators to evaluate the model: mean absolute error (MAE), mean error (ME), root mean square error (RMSE), and correlation coefficient R2. MAE is the average value of the absolute error between the predicted value and the true value. The calculation formula is: MAE = Σ(|predicted value-true value|) / number of samples. The smaller the value of MAE, the smaller the deviation between the predicted value and the true value of the model. ME is the average value of the error between the predicted value and the true value. The calculation formula is: ME = Σ(predicted value-true value) / number of samples. The value of ME can be positive or negative, indicating the direction and magnitude of the overall error. RMSE is the average value of the square of the error between the predicted value and the true value, and then the square root is taken. The calculation formula is: RMSE = √(Σ(predicted value-true value)^2 / number of samples). The smaller the value of RMSE, the smaller the volatility of the model prediction value. R 2 It is an indicator used to measure the degree of model fit, with a value range of 0 to 1. 2 The closer it is to 1, the better the model fits the observed data. 2 When R is equal to 1, it means that the model fits the data perfectly; 2 When it is 0, it means that the model cannot explain the variability of the data. The present invention uses the mean of the four indicators of the training set, test set, and validation set to represent the results.

[0054] The present invention uses Matlab to design and implement the PSO-BiLSTM prediction model. The specific experimental results are shown in the attached figure and the results are shown in Table 6. The results in Table 6 show that the model described in the present invention has a good degree of fitting, and the model has good robustness and generalization ability. Figure 4 ,5 It can be seen that in the process of iterative learning of the model, the training accuracy of solid content and nitrogen content tends to be stable, and the loss of model accuracy is basically zero. The training set of the model ( Figure 6 ,7) is the main data set for training the model. The input features in the training set are learned with the corresponding target values. The model can find the relationship between the features and the target so that it can make predictions when faced with new data. The test set of the model ( Figure 8 ,9) Used to adjust and optimize the model's hyperparameters and select the best parameter configuration to improve the model's generalization ability. During the model adjustment process, the test set can help detect whether the model is overfitting. Validation set ( Fig.10 ,11) is used to finally evaluate the performance of the model. After the model is trained and tested, the validation set is used to test the predictive ability and robustness of the model. The validation set is data that the model has not seen before, so it can more realistically reflect the performance of the model in actual applications.

[0055] Table 6 Regression prediction results

Claims

1. A method for detecting the total solid content and total nitrogen content of pig heart blood, characterized in that: The method comprises the following steps: 1) obtaining a pig heart blood sample; 2) determining the color value of the pig heart blood; 3) determining the amino acid ratio shift value Rf of the pig heart blood by thin layer chromatography; 4) using the pig heart blood color value and the amino acid ratio shift value Rf as input parameters, and using a trained neural network model to calculate the total solid content and total nitrogen content of the pig heart blood.

2. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 1, characterized in that: The steps of obtaining the pig heart blood sample include: promptly removing the pig heart after slaughtering, collecting the fresh blood through sieving after pouring out, adding an equal volume of rice wine, and refrigerating and transporting and storing at -20°C.

3. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 1, characterized in that: The step of determining the color value of pig heart blood comprises: taking a pig heart blood sample, placing it in a clean evaporating dish and taking a photo, extracting and converting the photo into RGB value, and recording data.

4. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 3, characterized in that: The step of determining the color value of pig heart blood comprises: placing an evaporating dish on a blue background plate at room temperature, using a digital camera to capture a high-definition image, and using a chromaticity value extraction code based on a pillow library to extract image color information and convert it into an RGB value.

5. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 1, characterized in that: The step of using thin layer chromatography to determine the amino acid ratio transfer value Rf of pig heart blood includes: selecting leucine, valine, alanine, and glycine as indicators for thin layer chromatography detection, using silica gel G thin layer plate, and using n-butanol-glacial acetic acid-water 3:1:1 as the developing agent.

6. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 5, characterized in that: The steps of using thin layer chromatography to determine the amino acid ratio transfer value Rf of pig heart blood include: accurately aspirating 200 μL of pig heart blood sample, adding 400 μL of protein precipitant acetonitrile, vortexing and mixing, standing for 10 minutes, centrifuging at 12000 rpm / min for 10 minutes, taking the supernatant, taking appropriate amounts of leucine, valine, alanine, glycine, and lysine reference substances, adding 0.1 mol / L hydrochloric acid to prepare mixed solutions with concentrations of 0.2 mg / mL, respectively, as reference substance solutions; aspirating 10 μL of each, spotting on the same silica gel G thin layer plate, using n-butanol-glacial acetic acid-water 3:1:1 as a developing agent, developing, taking out, drying, spraying with ninhydrin test solution, heating at 105°C until the spots are clearly colored, and calculating the Rf values ​​of the five amino acid components.

7. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 1, characterized in that: The neural network model is a PSO-BiLSTM model, and its training method uses a data training set of 53% of the original data divided by the algorithm, a test set of 20% of the data divided by the algorithm, and a validation set of 27% of the data divided by the algorithm.

8. The method for detecting the total solid content and total nitrogen content of pig heart blood according to claim 7, characterized in that : The PSO-BiLSTM model uses four indicators to evaluate the performance of the model: mean absolute error MAE, mean error ME, root mean square error RMSE, and correlation coefficient R2.

Images