Optimization of safflower seed extraction process based on ant colony optimization and backpropagation neural network

Abstract

To provide a method for optimizing a safflower seed extraction process based on an ant colony optimization method and back propagation neural network.SOLUTION: A method for optimizing a safflower seed extraction process includes: extracting safflower seeds by ultrasonic waves, and conducting a single-factor experiment in which ultrasonic output, ethanol concentration, solid-liquid ratio, and extracted temperature are selected as affectors in the single-factor experiment; calculating the extraction ratio of CS and FS by using a HPLC method, and determining the optimal value of the affectors on the basis of the extraction ratio; and calculating the total evaluation value by using an entropy weighting method, and obtaining the optimal process parameter by using an ant colony optimization method and back propagation neural network model. Specific steps of the ant colony optimization method and back propagation neural network include a step of constructing a back propagation neural network, and a step of constructing the optimal neural network model of the ant colony optimization method.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to the technical field of traditional Chinese medicine extraction, specifically to the optimization of safflower seed extraction process based on ant colony optimization and backpropagation neural network. [Background technology]

[0002] Safflower (Carthamus tinctorius L.), a member of the Asteraceae family and also known as Suetsumuhana, is native to India and distributed throughout Central and Southwest Asia and the Mediterranean region. Safflower is a medicinal and edible crop, and its corollas, seeds, stems, and leaves are rich in various active substances. Safflower contains nutrients such as fat, protein, various vitamins, and trace elements, which lower blood lipids, inhibit bacteria, prevent cardiovascular disease, and enhance cellular metabolism. Safflower seeds, which are white, inverted ovate achenes, have the same effects as the flowers and are used both as food and as medicine. Safflower seed extract contains various components, the most common of which are N-(p-coumaroyl)-serotonin (CS) and N-feruloyl-serotonin (FS). CS has been shown to have therapeutic effects against cardiovascular disease, glioma, and cholesterol. FS can treat atherosclerosis by reducing associated inflammation.

[0003] Ultrasonic extraction primarily uses vibration to increase contact between the solvent and medicinal materials, increasing the penetration of the solvent into the cells and improving extraction efficiency. Ultrasonic extraction does not require heating equipment and has a protective function for medicinal substances that are unstable and easily decomposed by heat. Compared to traditional methods, ultrasonic extraction time is also significantly reduced, and generally, an ultrasonic extraction time of 24 to 40 minutes can achieve an ideal extraction rate. The general advantage of ultrasonic extraction is that it can extract most types of herbal medicines and a variety of ingredients. Ultrasonic-assisted extraction has the advantages of short extraction time, high extraction efficiency, high extraction yield, high product purity, and is environmentally friendly and safe.

[0004] Artificial neural networks (ANNs) are models that use machines to train and predict data, and can handle a variety of complex or non-complex problems. Ant colony optimization (ACO) is a multivariate clustering algorithm that obtains optimal solutions for data through positive and negative feedback. ACO-BPNN is a new model that combines ant colony optimization and backpropagation neural networks. Artificial neural networks are gradually being applied in fields such as plant science, medicine, and information transmission.

[0005] Safflower seed oil has been widely studied in the fields of food, cosmetics, and livestock. However, safflower seed extract has received little research. Currently, there is a lack of research into obtaining the optimal extraction process for safflower seed through model optimization. Summary of the Invention [Problem to be solved by the invention]

[0006] The objective of this invention is to provide a method for optimizing the safflower seed extraction process based on ant colony optimization and backpropagation neural network. The safflower seed extract obtained under the optimized conditions has significant antioxidant effects, providing theoretical support for large-scale safflower seed extraction. [Means for solving the problem]

[0007] In order to achieve the above object of the invention, the technical solutions of the present invention are as follows:

[0008] In one aspect, the present invention provides a method for optimizing the extraction process of safflower seeds based on ant colony optimization and backpropagation neural network, Step (1) of ultrasonically extracting safflower seeds and conducting a single-factor experiment in which ultrasonic power, ethanol concentration, solid-liquid ratio, and extraction temperature are selected as influencing factors of the single-factor experiment; (2) calculating the extraction rates of N-(p-coumaroyl)-serotonin (CS) and N-feruloylserotonin (FS) using an HPLC method, and determining the optimal values ​​of the influencing factors based on the extraction rates; (3) calculating a comprehensive evaluation value by the entropy weighting method and obtaining optimal process parameters by using an ant colony optimization method and a backpropagation neural network model; The specific steps of the Ant Colony Optimization and Backpropagation Neural Network are as follows: The method includes: a step (I) of constructing a backpropagation neural network model; and a step (II) of constructing an optimization neural network model of an ant colony optimization method; The step (I) of constructing the backpropagation neural network model includes: Step 1) construct a neural network model with n single-layer hidden layer neurons, where n = 2, 3, 4, 5, 6; The hidden layer activation function is a sigmoid function a Step 2) where the output layer activation function is a linear function; The backpropagation optimizer settings are: Step 3) where the optimizer is Adam and the learning step size Lr is 0.01; The loss function is MSEloss b Step 4) where Step 5) using a backpropagation optimizer to optimize the neural network model based on the initial weights for 400 cycles, which is the number of epochs, to search for the minimum value of the loss function; The step (II) of constructing an optimized neural network model of the ant colony optimization method includes: Step 1) randomly generates multiple ants distributed at different spatial locations and decodes the spatial locations, i.e., initializes the weights; The constructed backpropagation neural network model optimizes the spatial positions of different ants, searches for the optimal positions of each of the multiple ants, and records the pheromone F. c Step 2) and Encode the weights of each of the ants and record the optimal position among them. d Step 3) and Normalizing the pheromone F of multiple ants e Step 4) and The magnitude relationship between normalized pheromone F and transition probability p0 f Based on this, a global update is made to the spatial position of the ants. g or local update h Step 5) Update pheromones of multiple ants based on volatility rate lf i Step 6) and and step 7) of repeating steps 1) to 6) above until a predetermined number of cycles of the ant colony optimization method is reached, and outputting an optimal model result.

[0009] Specifically, the sigmoid function at a is JPEG0007811045000001.jpg2976, where x is the weight, The loss function MSEloss at b is JPEG0007811045000002.jpg25113, JPEG0007811045000003.jpg1212 is the true value, JPEG0007811045000004.jpg1312 is the predicted value and i represents different samples.

[0010] The function in c is JPEG0007811045000005.jpg1559, where k represents different ants, The function in d is JPEG0007811045000006.jpg1455, The function in e is JPEG0007811045000007.jpg2388, The function in f is JPEG0007811045000008.jpg3295, The function in g is: Global updates are JPEG0007811045000009.jpg10126, In the above equation, lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function in h is: Local Update: JPEG0007811045000010.jpg10160, where arctan is the arctangent function, lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function in i is JPEG0007811045000011.jpg1486, In the formula, lf is the volatility rate of the Ant Colony Optimization method, and t represents the t-th Ant Colony Optimization cycle.

[0011] Specifically, the solid-liquid ratio described in step (1) is the ratio of the mass of safflower seed powder to the volume of ethanol.

[0012] Preferably, the chromatography column in the HPLC method in step (2) is an Agilent Eclipse XDB-C 18 (4.6×250mm, 5μm).

[0013] Preferably, the mobile phase of the HPLC method in step (2) is a 0.4% formic acid solution in water (A) and a methanol solution (B).

[0014] Preferably, the flow rate of the HPLC method in step (2) is 1 mL / min.

[0015] Preferably, the gradient elution method of the HPLC method in step (2) is 90% A for 0 min; 90% A for 3 min; 50% A for 20 min; and 10% A for 30 min.

[0016] Preferably, the injection volume of the HPLC method in step (2) is 10 μL, the column temperature is 35° C., and the detection wavelength is 310 nm.

[0017] Preferably, the extraction rate (%) described in step (2) = compound concentration in the extract (mg / mL) / [herb weight (mg) / herb extraction solvent volume (mL)] x 100%.

[0018] Preferably, the optimal values ​​of the influencing factors determined in step (2) are an ultrasonic power of 160-320 W, an ethanol concentration of 70-90%, a solid-liquid ratio of 15-25 mL / g, and an extraction temperature of 60-80°C.

[0019] Preferably, the overall evaluation value (Y) in step (3) = (extraction rate CS )×0.5036+(extraction rate FS )×0.4964.

[0020] Preferably, the optimal process parameters in step (3) are: ultrasonic power 320W, ethanol concentration 80%, solid-liquid ratio 15mL / g, and extraction temperature 80°C.

[0021] In yet another aspect, the present invention provides a safflower seed extract prepared by the above method.

[0022] In yet another aspect, the present invention provides use of a safflower seed extract prepared by the above method or the above safflower seed extract in the preparation of an antioxidant.

[0023] Specifically, the antioxidant can reduce the content of MDA and increase the content of SOD.

[0024] Specifically, the antioxidants can increase HO-1 mRNA levels and decrease NOX2 mRNA levels. [Effects of the Invention]

[0025] The beneficial effects of the present invention are as follows: The present invention provides a method for optimizing the extraction process of safflower seeds based on ant colony optimization and backpropagation neural network. The safflower seed extract under optimized conditions has significant antioxidative stress effects, providing a theoretical basis for the clinical development of safflower seeds. The present invention further expands the application field of the ACO-BPNN model and provides reference for large-scale extraction of safflower seeds. [Brief explanation of the drawings]

[0026] [Figure 1a-1b] 1A and 1B are HPLC chromatograms, where FIG. 1A is the HPLC chromatogram of the sample and FIG. 1B is the HPLC chromatogram of the mixed reference. [Figure 2] The response values ​​of CS and FS, where (a) is the ultrasonic frequency response value of CS and FS, (b) is the ethanol concentration response value of CS and FS, (c) is the solid-liquid ratio response value of CS and FS, and (d) is the extraction temperature response value of CS and FS. [Figures 3a-3f] 3a is a three-dimensional response surface diagram of ultrasonic frequency and ethanol concentration, FIG. 3b is a three-dimensional response surface diagram of ultrasonic frequency and solid-liquid ratio, FIG. 3c is a three-dimensional response surface diagram of ultrasonic frequency and solid-liquid ratio, FIG. 3d is a three-dimensional response surface diagram of ethanol concentration and solid-liquid ratio, FIG. 3e is a three-dimensional response surface diagram of ethanol concentration and extraction temperature, and FIG. 3f is a three-dimensional response surface diagram of solid-liquid ratio and extraction temperature. [Figure 4] The overall evaluation value in the ACO-BPNN model, where (a) is the relationship between the predicted value and the actual value, and (b) is the residual diagram. [Figure 5]Figure 5 shows the results of in vivo antioxidant activity, where (a) is serum SOD activity, (b) is serum MDA activity, (c) is HO-1 mRNA level in brain tissue, and (d) is NOX2 mRNA level in brain tissue. n = 6 per group. In Figure 5, statistically, identical letters (e.g., a and a, b and b, c and c, d and d) indicate no significant difference (no significance), adjacent letters (e.g., a and b, b and c, c and d) indicate significance at P < 0.05, and spaced letters (e.g., a and c, a and d, b and d) indicate significance at P < 0.01. DETAILED DESCRIPTION OF THE INVENTION

[0027] In order to facilitate understanding of the technical means, creative features, objectives and effects achieved by the present invention, the present invention will be further described below with reference to specific examples. However, the following examples are only preferred examples of the present invention and are not all-inclusive. Based on the examples in the embodiments, other examples obtained by those skilled in the art without requiring creative labor fall within the scope of protection of the present invention. In the following examples, unless otherwise specified, all operating methods used are conventional operating methods, all devices used are conventional devices, and all materials of the devices used in each example are the same.

[0028] Example 1 1. Materials and Methods 1.1 Medicines, reagents and equipment Dried safflower seeds (lot number: 20221010) were purchased from Jimxar County, Changji Hui Autonomous Prefecture, Xinjiang, China. Standards N-(p-coumaroyl)-serotonin, N-feruloylserotonin, and 2,2-diphenyl-1-picrylhydrazyl (DPPH) were purchased from Shanghai Yuanye Biological Technology Co., Ltd., China. The purity of the standards was ≥98%. The FRAP kit was purchased from Shanghai Biyuntian Biological Technology Co., Ltd., China. The solvents used in the mobile phase were HPLC grade. TRIzol® Invitrogen was purchased from Thermo Fisher Scientific, China. 5x All-In-One RT Master Mix was purchased from Applied Biological Materials Inc., a Canadian company. 2x Color SYBR Green qPCR Master Mix was purchased from Shanghai Taitan Scientific Co., Ltd., China.

[0029] The Ezra multi-function grinder was purchased from Ezela Electric Co., Ltd. The Agilent 1200 high-performance liquid chromatograph was purchased from Agilent Technologies, Inc. The xP105DR analytical balance was purchased from Mettler Toledo Instruments Co., Ltd. The xM-400ULF liquid crystal low-frequency ultrasonic cleaner was purchased from Xiao Mei Ultrasonic Instruments Co., Ltd.

[0030] 1.2 Preparation of sample and standard solutions Dried safflower seeds were crushed in a grinder. The crushed safflower seed powder was mixed with different volumes of solvent and extracted using different ultrasonic frequencies. After extraction, the sample was allowed to stand and cool to room temperature. The supernatant was collected by centrifugation and passed through a 0.45 μm microfiltration membrane to obtain the sample solution.

[0031] N-(p-coumaroyl)-serotonin (CS) and N-feruloyl-serotonin (FS) were precisely weighed into 1.5 mL centrifuge tubes and dissolved in methanol to obtain standard solutions. The CS and FS standard solutions were accurately weighed and diluted with methanol to obtain five different concentrations of standard solutions.

[0032] 1.3 High-performance liquid chromatography analysis The chromatography column was an Agilent Eclipse XDB-C. 18 The column diameter was 4.6 × 250 mm, 5 μm. The mobile phase consisted of 0.4% formic acid (A) and methanol (B) at a flow rate of 1 mL / min. The gradient elution was 90% A at 0 min; 90% A at 3 min; 50% A at 20 min; and 10% A at 30 min. The injection volume was 10 μL, the column temperature was 35 °C, and the detection wavelength was 310 nm.

[0033] 1.4 Methodological research 1.4.1 Specificity The test substance solution and the binary mixed standard solution were injected into a liquid chromatograph for analysis. The specificity of the method was confirmed according to the above chromatographic conditions.

[0034] 1.4.2 Linear Relationships The concentrations of CS were 0.024, 0.034, 0.048, 0.068, and 0.08 mg / mL, and the concentrations of FS were 0.068, 0.09, 0.12, 0.15, and 0.2 mg / mL. The linear regression equations of CS and FS were established with concentration as the abscissa and the peak area of ​​the chromatographic peak as the ordinate.

[0035] 1.4.3 Accuracy The mixed standard solution was taken and injected six times in replicates according to the above chromatographic conditions. The chromatographic peak areas of each component were recorded and the relative standard deviation (RSD) values ​​were calculated.

[0036] Preparation of mixed standard solution: An appropriate amount of 1 mg / mL stock solution of CS and FS standards was aspirated and diluted with methanol until the concentrations of CS and FS became 0.048 mg / mL and 0.15 mg / mL, respectively.

[0037] 1.4.4 Stability Freshly prepared sample solutions were taken and tested at 0, 2, 4, 6, 8, 12, and 24 hours, respectively. The chromatographic peak areas of each component were recorded, and the relative standard deviation (RSD) values ​​were calculated.

[0038] 1.4.5 Reproducibility Six samples of the same batch of medicinal materials were prepared in parallel under the same conditions and detected. The chromatographic peak areas of each component were recorded and the relative standard deviation (RSD) values ​​were calculated.

[0039] 1.4.6 Sample collection Safflower seed samples of known concentrations were re-extracted using the above extraction conditions and spiked with a reference solution. Six parallel runs were performed according to the above chromatographic conditions. The relative standard deviation (RSD) of each component was calculated.

[0040] 1.5 Single-factor experimental design Safflower seeds were weighed and analyzed using ultrasonic frequency (80-400W), ethanol concentration (60-100%), solid-liquid ratio (5-25mL / g), and extraction temperature (40-80°C). Univariate analysis was used for single-factor studies. When one factor was investigated, other factors included ultrasonic frequency 240W, ethanol concentration 100%, solid-liquid ratio 15mL / g, extraction temperature 60°C, and extraction time 60min. The extracted supernatant was analyzed by HPLC.

[0041] The concentration of each component was calculated based on the calibration curve. The yield was calculated using the following formula: Extraction rate (%) = compound concentration in the extract (mg / mL) / [herb weight (mg) / herb extraction solvent volume (mL)] × 100% (Equation 1).

[0042] 1.6 Optimization of safflower seed extraction production process 1.6.1 Calculation of comprehensive evaluation value by entropy weighting method To evaluate the overall extraction rate, the entropy weighting method was used to assign coefficients to the two components. The overall evaluation score was calculated as follows: Overall evaluation value (Y) = (extraction rate CS )×0.5036+(extraction rateFS )×0.4964 (Equation 2).

[0043] Entropy weighting method: Normalization was performed on the two compounds, and the weight of each compound was calculated using the entropy weighting method. Finally, the overall evaluation result of the extract was calculated based on the weight.

[0044] In the above formula 2, 0.5036 was the weighting coefficient for CS and 0.4964 was the weighting coefficient for FS.

[0045] 1.6.2 RSM Model Based on the single-factor experiment, the overall evaluation value of safflower seeds was selected as the dependent variable. The response surface experiment was designed using Design Expert 13 software according to the Box-Benhnken central composite experimental design (BBD) principle. There were a total of six center points for the 30 sets of experiments. The conditions and levels are listed in Table 1, and the specific design of the 30 sets of experiments is shown in Table 5.

[0046] Table 1 Experimental conditions and levels JPEG0007811045000012.jpg45163

[0047] 1.6.3 ACO-BPNN model The 30 sets of data results designed by BBD were used as the input and output layers for training prediction. Related settings and methods: A neural network model with 4 hidden layer neurons was constructed, and the mean squared error (MSE) was used as the loss function. The model was optimized using gradient descent (optimizer: Adam, learning rate lr: 0.01, number of cycles: 300). Since the initial weights of the neural network have a significant impact on the results, the initial weights were optimized using ant colony optimization.

[0048] 1.6.3.1 Building a Backpropagation Neural Network Model (1) Construct a neural network model with a single hidden layer of neurons, n = 2, 3, 4, 5, 6. (2) The hidden layer activation function is a sigmoid functiona and the output layer activation function is a linear function. (3) Backpropagation optimizer settings: Optimizer: Adam, learning step size Lr: 0.01. (4) Loss function: MSEloss b . (5) Using the backpropagation optimizer, we optimize the neural network model based on the initial weights for 400 cycles, which is the number of epochs, to find the minimum value of the loss function.

[0049] 1.6.3.2 Construction of an optimized neural network model using ant colony optimization (1) Randomly generate multiple ants distributed at different spatial positions, and decode the spatial positions, i.e., initialize the weights. (2) Using the constructed backpropagation neural network model, we optimize the spatial positions of different ants, search for the optimal positions of each of the multiple ants, and record the pheromone F. c . (3) Encode the weights of each ant and record the optimal position among the ants. d . (4) Normalize the pheromone F of multiple ants e . (5) The magnitude relationship between the normalized pheromone F and the transition probability p0 f Based on this, a global update is made to the spatial position of the ants. g or local update h Do the following. (6) Update the pheromones of multiple ants based on the volatility rate lf. i . (7) Repeat the above steps (1) to (6) until the specified number of cycles of the Ant Colony Optimization method is reached, and output the optimal model result.

[0050] The functions of a, b, c, d, e, f, g, h, and i above are as follows: The sigmoid function at a is JPEG0007811045000013.jpg2976, where x is the weight, The loss function MSEloss at b is JPEG0007811045000014.jpg29113, JPEG0007811045000015.jpg1212 is the true value, JPEG0007811045000016.jpg1211 is the predicted value and i represents different samples.

[0051] The function in c is JPEG0007811045000017.jpg1559, where k represents different ants, The function in d is JPEG0007811045000018.jpg1455, The function in e is JPEG0007811045000019.jpg2689, The function in f is JPEG0007811045000020.jpg2793, The function in g is: Global updates are JPEG0007811045000021.jpg9126, In the above equation, lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function in h is: Local updates are JPEG0007811045000022.jpg10131, where arctan is the arctangent function, lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function in i is JPEG0007811045000023.jpg1386, In the formula, lf is the volatility rate of the Ant Colony Optimization method, and t represents the t-th Ant Colony Optimization cycle.

[0052] 1.7 Model Validation Verification experiments were conducted on the safflower seed extraction conditions obtained by optimizing the two models, and the concentrations of the two components were calculated using the peak areas of the chromatograms. Each set of results was repeated six times. The two models were evaluated by their relative error, which was calculated using the following formula: Relative error (%) = (predicted value - experimental value) / predicted value × 100% (Equation 3).

[0053] 1.8 In vitro antioxidant experiment of safflower seed extract The antioxidant capacity of safflower seed extract under two extraction conditions was evaluated using the DPPH and FRAP methods. The DPPH method involves precisely weighing an appropriate amount of DPPH powder and adding absolute ethanol to prepare a 0.3 mM / L DPPH working solution. 180 μL of DPPH working solution and 20 μL of sample solution, Trolox solution, or solvent were placed in a 96-well plate and thoroughly mixed. After 30 minutes in the dark at room temperature, the absorbance was measured at 517 nm. The calculation formula is as follows: JPEG0007811045000024.jpg19120

[0054] Abs DPPH is the absorbance of the mixture of DPPH and absolute ethanol. Abss and Absc are the absorbances of the sample solutions containing DPPH or absolute ethanol, respectively. For the FRAP method, refer to the kit instructions. Trolox was used as a positive reference.

[0055] 1.9 In vivo antioxidant experiment of safflower seed extract 1.9.1 Animal experiments and design Eighteen adult male Sprague-Dawley rats (200±20g) were provided by the Experimental Animal Center of Zhejiang University of Traditional Chinese Medicine (ethics code: IACUC-20220516-15). After one week of dietary adaptation, the rats were divided into a sham-operated group, a model group, and a safflower seed extract-treated group, with six rats in each group.

[0056] A modified Longa beam embolization method was used to create an MCAO model, and the modeling method is as follows. After weighing the rats, atropine (0.04 mg / kg) was injected intramuscularly, followed 5 min later by an intraperitoneal injection of Zoletil 50 (40 mg / kg). After successful anesthesia, the rats were fixed in a supine position and the neck hair was shaved. After disinfection with alcohol, the subcutaneous tissue and muscle were separated along the center of the neck to expose the blood vessels on the right side of the rat's neck (depending on the rat's left or right side). The right common carotid artery (CCA), internal carotid artery (ICA), and external carotid artery (ECA) were then isolated. The ECA and CCA were ligated, and the ICA was closed with an artery clamp. A small incision was then made in the common carotid artery, and a fine silicone-coated surgical nylon monofilament was inserted into the internal carotid artery. After 60 min of ischemia, it was removed and the wound was sutured. After the rats woke up, a Longa score of 1 to 3 was used as the criterion for successful modeling. Three days after administration, the animals were sacrificed and tissue samples were collected. The rats in the treatment group were intragastrically administered with safflower seeds (1.3g / kg, the dosage was calculated based on the amount of herb, i.e., the weight of safflower seeds), and the rats in the remaining groups were intragastrically administered with the corresponding saline solution.

[0057] The specific preparation method for the safflower seed extract of the treatment group was as follows: Extraction was performed according to an optimized method: ultrasonic frequency 320W, ethanol volume concentration 80%, solid-liquid ratio 15mL / g, extraction temperature 80℃, extraction time 60min. After extraction was completed, the supernatant was filtered and rotary evaporated until the alcohol odor disappeared, and water was added to reach the desired concentration.

[0058] The Longa score criteria are as follows: 0 points: normal, no neurological impairment. Score 1: The contralateral forelimb could not be fully extended, and mild neuropathy occurred. 2 points: When walking, the rat turned around to the opposite (paralyzed) side, and moderate neurological impairment occurred. 3 points: When walking, the rat's body leaned to the opposite side (paralyzed side) and fell, resulting in severe neurological damage. 4 points: Unable to walk independently and loss of consciousness.

[0059] 1.9.2 Antioxidant Levels Peripheral blood was collected from the abdominal aorta of rats. After 30 minutes at room temperature, it was centrifuged at 4500 rpm for 15 minutes to obtain the supernatant. The corresponding SOD and MDA assays were then performed according to the kit instructions (the SOD kit was purchased from Nanjing Jiancheng Institute of Biological Engineering, product number A001-3, and the MDA kit was purchased from Solarbio Biotechnology Co., Ltd., product number BC0025).

[0060] 1.9.3 HO-1 mRNA levels and NOX2 mRNA levels Total RNA was extracted from the brain tissue of each group of rats using TRIzol® Reagent. Reverse transcription was performed according to the manufacturer's instructions. Real-time qPCR amplification reactions were performed using Universal SYBR Green qPCR Master Mix. HO-1 and NOX2 mRNA levels were detected using real-time quantitative PCR. GAPDH was used to quantify HO-1 and NOX2 mRNA levels. The primer sequences used for RT-qPCR analysis are listed in Table 2.

[0061] Table 2 JPEG0007811045000025.jpg52155

[0062] The reaction system for real-time quantitative PCR contained the following components: final reaction system (10 μL), 5 μL of 2x SYBR QPCR MIX, 1 μL of cDNA, 0.2 μL of forward primer (10 μM), 0.2 μL of reverse primer (10 μM), and sterile water added up to 10 μL.

[0063] The PCR reaction program was 95°C for 5 min; 95°C for 10 s; 60°C for 30 s; 95°C for 10 s; 60°C for 1 min; 95°C for 30 s, for a total of 40 cycles.

[0064] 1.10 Statistical analysis The software used was GraphPad Prism 9 software, Design Expert 13 software, and Visual Studio Code to run Python. Single-factor analysis of variance (ANOVA) was used to evaluate the effect of each group. P<0.05 was considered significant. Significance was expressed as *P<0.05, **P<0.01.

[0065] 2.Results 2.1. Methodological study 2.1.1 Specificity The chromatographic peaks of the test substance solution and the binary mixed standard solution are shown in Figures 1a and 1b. The chromatographic peak resolution of the two quantitative components at 310 nm was greater than 1.5, demonstrating ideal chromatographic separation. No interfering peaks were observed in the extraction solvent, demonstrating the good specificity of the method.

[0066] 2.1.2 Linear Relationships The linear regression equations for the two components are shown in Table 3. The regression equation for each component is R 2 >0.999, showing a positive linear relationship, and can be used to measure the content of each component.

[0067] Table 3 Linear relationships JPEG0007811045000026.jpg23153

[0068] 2.1.3 Precision, stability, reproducibility, and sample recovery The method validation results are shown in Table 4. The RSD for precision and repeatability was less than 2%, and the RSD for stability was less than 1%. The results showed that the instrument was stable, the method was reliable, and the samples were stable within 24 hours.

[0069] Table 4. Methodological findings JPEG0007811045000027.jpg301592.2 Single factor experiment 2.2.1 Ultrasonic frequency The yields of the two components at different ultrasonic frequencies are shown in Figure 2(a). The yield of CS remained stable from ultrasonic frequencies of 80 to 160 W, increased from 160 to 240 W, and remained stable from 240 to 400 W. The yield of FS initially increased in the 80 to 240 W range and then slowly decreased. This is because, when the ultrasonic frequency is within a certain range, ultrasound destroys the plant cell wall, and as the frequency increases, the cells and the solvent tend to complete the material exchange. Therefore, for subsequent studies, we selected the higher yield of 160 to 320 W.

[0070] 2.2.2 Ethanol concentration The yield results for the two components at different ethanol concentrations are shown in Figure 2(b). The CS yield gradually increased in the ethanol concentration range of 60-80% and gradually decreased in the 80-100% range. The FS yield increased in the ethanol concentration range of 60-70% and decreased in the 70-100% range. When the concentration was 90-100%, the yields of CS and FS decreased. Based on the principle that like dissolves like, CS and FS showed better yields and higher solubility at ethanol concentrations of 70-90%, so this invention selected this range for subsequent studies.

[0071] 2.2.3 Solid-liquid ratio The yield results for the two components at different solid-liquid ratios are shown in Figure 2(c). The CS yield increased in the range of 5 to 20 mL / g and decreased in the range of 20 to 25 mL / g. The FS yield increased in the range of 5 to 10 mL / g, decreased in the range of 10 to 15 mL / g, increased in the range of 15 to 20 mL / g, and decreased in the range of 20 to 25 mL / g. The yield was high in the range of 15 to 25 mL / g. Therefore, the present invention selects the range of 15 to 25 mL / g for subsequent studies.

[0072] 2.2.4 Extraction temperature The results of the yields of the two components at different extraction temperatures are shown in Figure 2(d). The CS yield increased in the temperature range of 40 to 70 °C and tended to stabilize in the 70 to 80 °C range. The FS yield also showed an increasing trend in the 40 to 80 °C range. As the temperature increased, the movement of solvent molecules became faster, promoting the dissolution of the components. When the temperature increased to a certain value, the heat disturbed the stability of the components, and the extraction rate decreased. Therefore, the present invention selects the range of 60 to 80 °C for subsequent studies.

[0073] 3.3 RSM model results The range of factors was screened based on the results of each single factor. The conditions and results of the RSM experiment are shown in Table 5. A multiple linear regression analysis was performed on 30 sets of data, and the relationship between the overall evaluation and each factor was as follows: Y 総合評価値 =0.185956+0.00365422A+0.00288556B-0.00324564C+0.000973157D-0.00154132AB-0 .00391808AC-0.0012893AD+0.00995856BC+0.00123389BD+0.00071382CD-0.00988253A 2 -0.00611679B 2 -0.00162696C 2 -0.00686053D 2 .

[0074] Table 6 shows the analysis of variance for the overall evaluation score using the RSM model. The results indicated good model fit (P<0.01). The significance of the P and F values ​​indicated that the influence of the four factors on the safflower seed extraction rate was in the order A > C > B > D, with the ultrasonic frequency having the greatest effect. The interactions among the four factors are shown in Figures 3a–3f. The steepness and contour plots of the three-dimensional response surface can be visualized to reflect the influence of the interaction between each factor on the overall evaluation score of the safflower seeds. The steeper the arch of the response surface, the stronger the interaction between the two factors. The optimal extraction results and conditions predicted by the RSM method are shown in Table 7.

[0075] Table 5. 30 sets of RSM experimental conditions and extraction results JPEG0007811045000028.jpg227153

[0076] Table 6. Results of the overall evaluation value of the analysis of variance for the RSM design JPEG0007811045000029.jpg155153

[0077] Table 7. Prediction conditions and predicted values ​​for RSM and ACO-BPNN JPEG0007811045000030.jpg31160

[0078] 2.4 Results of the ACO-BPNN model ACO-BPNN optimization was performed on 30 sets of data from the RSM experiment. The optimization results of the ACO-BPNN model are shown in Table 7. From the simulation process in Figure 4, the training values ​​and predicted values ​​are uniformly distributed, which indicates that the ACO-BPNN model has a good fit.

[0079] 2.5 Verification of extraction rate under optimization conditions of RSM and ACO-BPNN Using two different modeling calculation methods, two sets of different optimal extraction conditions were obtained and verified (the verification conditions are shown in Table 8). 2 The correlation coefficients were 0.7915 for RSM and 0.9154 for ACO-BPNN, respectively. This indicates the good applicability and high predictive value of the ACO-BPNN model. Considering the actual situation, all factors were approximate integers when conducting the validation experiments. The actual conditions for the RSM validation experiment were an ultrasonic frequency of 280 W, an ethanol concentration of 74%, a solid-liquid ratio of 15 mL / g, and an extraction temperature of 70°C. The actual conditions for the ACO-BPNN validation experiment were an ultrasonic frequency of 320 W, an ethanol concentration of 80%, a solid-liquid ratio of 15 mL / g, and an extraction temperature of 80°C. The predicted and actual results of the two models are shown in Table 8. Furthermore, the actual extraction results showed that the overall evaluation score of ACO-BPNN was higher than that of RSM, and the relative error of ACO-BPNN was lower than that of RSM, indicating that ACO-BPNN has a stronger predictive effect.

[0080] The reason why the actual values ​​are lower than the predicted values ​​may be due to limitations of the equipment and conditions, and the actual extraction conditions do not match the predicted conditions, resulting in errors.

[0081] Table 8 Actual and predicted values ​​under the prediction conditions of RSM and ACO-BPNN JPEG0007811045000031.jpg254157

[0082] 2.6. In vitro antioxidant activity The extraction conditions were determined by comparing the actual values ​​from the two models, and the extracts were subjected to total antioxidant index measurements (composed of DPPH and FRAP). The procedure was performed according to the kit instructions (DPPH was purchased from Shanghai Yuanyeh Biotechnology Co., Ltd., product number S30629, and the FRAP kit was purchased from Shanghai Biyuntian Biotechnology Co., Ltd., product number S0116), with triepoxide used as a positive control. The DPPH and FRAP results for the RSM model were 1.9889 ± 0.2418 and 1.4643 ± 0.0440 mmol Trolox / g, respectively. The DPPH and FRAP results for the ACO-BPNN model were 2.2819 ± 0.4199 and 1.6367 ± 0.1173 mmol Trolox / g, respectively. All calculations were based on the dry weight of safflower seeds. According to the antioxidant results of the two models, the antioxidant activity of safflower seed extract in the ACO-BPNN model was higher than that in the RSM model.

[0083] 2.7 Antioxidant activity in MCAO model rats Three days after cerebral ischemia-reperfusion, the serum SOD and MDA levels in each group were measured as shown in Figure 5(a)-(b). This reflects the strength of the rats' ability to protect against oxidative stress after cerebral ischemia-reperfusion. Compared with the sham-operated group, serum SOD levels in the model group were significantly decreased (P<0.01), and MDA levels were significantly increased (P<0.01). Compared with the model group, serum SOD levels in the treatment group were significantly increased (P<0.01), and MDA levels were significantly decreased (P<0.05).

[0084] The mRNA expression levels of HO-1 and NOX2 in the brain tissue of each group are shown in Figure 5(c)-(d). Compared with the sham-operated group, the mRNA expression levels of HO-1 and NOX2 in the model group were significantly elevated (P<0.01). The HO-1 level in the treatment group was significantly higher than that in the model group (P<0.01), and the NOX2 level was significantly lower than that in the model group (P<0.05).

[0085] The above description is only a preferred embodiment of the present invention, and does not limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the extraction process of safflower seeds based on ant colony optimization and backpropagation neural network, comprising: Step (1) of ultrasonically extracting safflower seeds and conducting a single factor experiment in which ultrasonic power, ethanol concentration, solid-liquid ratio and extraction temperature are selected as influencing factors of the single factor experiment; (2) calculating the extraction rates of N-(p-coumaroyl)-serotonin and N-feruloyl serotonin using HPLC method, and determining the optimal values ​​of the influencing factors based on the extraction rates; (3) calculating a comprehensive evaluation value by the entropy weighting method and obtaining optimal process parameters by using an ant colony optimization method and a backpropagation neural network model; The specific steps of the ant colony optimization backpropagation neural network include: step (I) of constructing a backpropagation neural network model; and step (II) of constructing an optimized neural network model of the ant colony optimization method; The step (I) of constructing the backpropagation neural network model includes: Step 1) construct a neural network model with n single-layer hidden layer neurons, where n = 2, 3, 4, 5, 6; The hidden layer activation function is a sigmoid function ａ Step 2) where the output layer activation function is a linear function; The backpropagation optimizer settings are: Step 3) where the optimizer is Adam and the learning step size Lr is 0.01; The loss function is MSEloss ｂ Step 4) where 5) using a backpropagation optimizer to optimize the neural network model based on the initial weights for 400 cycles, which is the number of epochs, to search for a minimum of the loss function; The step (II) of constructing an optimized neural network model of the ant colony optimization method includes: Step 1) randomly generate multiple ants distributed at different spatial locations and decode the spatial locations, i.e., initialize the weights; The constructed backpropagation neural network model optimizes the spatial positions of different ants, searches for the optimal positions of each of the multiple ants, and records the pheromone F. ｃ Step 2) and Encode the weights of each of the ants and record the optimal position among them. ｄ Step 3) and Normalizing the pheromone F of multiple ants ｅ Step 4) and The magnitude relationship between the normalized pheromone F and the transition probability p0 ｆ Based on this, a global update is made to the spatial position of the ants. ｇ or local update ｈ Step 5) of performing Update the pheromones of multiple ants based on the volatility rate lf ｉ Step 6) and and step 7) of repeating steps 1) to 6) above until a specified number of cycles of the Ant Colony Optimization method is reached, and outputting an optimal model result.

2. The sigmoid function at a is and where x is the weight, The loss function MSEloss in b is and is the true value, 2. The method of claim 1, wherein i is a predicted value and i represents a different sample.

3. The function in c is and where k represents a different ant, The function in d is and The function in e is and The function in f is and The function in g is: Global updates are and where lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function in h is: Local updates are and where arctan is the arctangent function, lr is the learning rate of the ant colony optimization method, and R(i,j) is a random matrix between 0 and 1. The function at i is and 2. The method of claim 1, wherein lf is the volatility rate of the Ant Colony Optimization method, and t represents the t-th Ant Colony Optimization cycle.

4. The method according to claim 1, characterized in that the extraction rate (%) described in step (2) = compound concentration in the extract (mg / mL) / [herb weight (mg) / herb extraction solvent volume (mL)] x 100%.

5. 2. The method of claim 1, wherein the optimal values ​​of the influencing factors determined in step (2) are: ultrasonic power of 160-320 W, ethanol concentration of 70-90%, solid-liquid ratio of 15-25 mL / g, and extraction temperature of 60-80°C.

6. Comprehensive evaluation value (Y) in step (3) = (extraction rate ＣＳ )×0.5036+(extraction rate ＦＳ ) × 0.4964.

7. 2. The method of claim 1, wherein the optimal process parameters in step (3) are: ultrasonic power 320 W, ethanol concentration 80%, solid-liquid ratio 15 mL / g, and extraction temperature 80°C.

Images