A kind of synthesis process of apigenin

Abstract

The present invention relates to the technical field of pharmaceutical chemical synthesis, and specifically to a synthesis process of apigenin. The process comprises: a step of synthesizing apigenin raw materials: after mixing and stirring naringenin and pyridine completely, iodine is added, the temperature of the solution in the reaction container is increased, and the solution is refluxed; during the reflux process, a hyperspectral image is obtained; a material uniformity index is obtained based on the similarity between different pixels in the hyperspectral image and the characteristics of the pixels themselves; the temperature is adjusted according to the material uniformity index to obtain a reaction solution; a step of preparing crude apigenin: the reaction solution is evaporated to obtain a solid, hydrochloric acid and water are added to the solid until it is neutral, and crude apigenin is obtained; a step of refining apigenin: for crude apigenin, ethanol is added, and after heating and refluxing, suction filtration is performed to obtain a filtrate, the filtrate is concentrated and allowed to stand, suction filtration is performed after standing, and fine apigenin is obtained after drying, and the synthesis of apigenin is completed. The present invention improves the purity of the synthesized apigenin.

Description

Technical Field

[0001] The invention relates to the technical field of pharmaceutical chemical synthesis, and in particular to a synthesis process of apigenin. Background Art

[0002] Apigenin is a naturally occurring flavonoid compound with biological activities such as anti-tumor, cardiovascular and cerebrovascular protection, and antibacterial properties. However, the content of apigenin in natural products is relatively low. Therefore, producing apigenin through synthetic methods can reduce costs and increase output.

[0003] During the reflux process of the apigenin synthesis process, the heating temperature of the reaction solution will affect the purity of apigenin. By measuring the purity of apigenin in solution samples at different sampling times, the temperature of the reaction solution can be adaptively adjusted according to the change of apigenin purity over time. However, the uneven distribution of substances in the reaction solution may cause the measured value of the purity of apigenin in the solution sample to fail to accurately represent the overall purity of apigenin in the reaction solution. The use of hyperspectral images can evaluate the uniformity of substance distribution in the reaction solution. However, bubbles may appear in the reaction solution due to the evaporation of the organic solvent pyridine and local boiling of the solution. The appearance of bubbles may cause erroneous substance distribution characteristics in the collected hyperspectral images, thereby affecting the evaluation of the uniformity of substances in the entire reaction solution. Summary of the invention

[0004] In order to solve the technical problem of inaccurate temperature regulation caused by bubbles, the present invention provides a synthesis process of apigenin, and the technical scheme adopted is as follows:

[0005] The present invention provides a synthesis process of apigenin, which comprises the following steps:

[0006] The synthesis steps of apigenin raw materials are as follows: after mixing naringenin and pyridine, iodine is added, the temperature of the solution in the reaction container is raised to 116°C, and the solution is refluxed; when no naringenin is detected by thin layer detection, the reflux is stopped to obtain the reaction solution; the reflux process includes the following steps:

[0007] Collecting hyperspectral images of the reaction liquid and the heating temperature;

[0008] Construct a spectral curve based on the reflectance and wavelength of the hyperspectral image band to obtain the flatness of the spectral curve of the pixel;

[0009] Obtain the pixel spectral similarity between pixels according to the spectral curves and spectral curve flatness of two pixels of the hyperspectral image;

[0010] Clustering is performed through pixel spectral similarity to obtain the bubble pixel suspicion of each type of pixel;

[0011] The material uniformity index of the hyperspectral image is obtained according to the bubble pixel suspicion;

[0012] Obtain the temperature adjustment value according to the material uniformity index and adjust the heating temperature adaptively;

[0013] Preparation step of crude apigenin: evaporating the reaction liquid obtained by stopping reflux in the apigenin raw material synthesis step to obtain a solid, adding hydrochloric acid and water of preset mass to the solid until it is neutral, and obtaining crude apigenin;

[0014] Apigenin refining steps: For crude apigenin, add a preset mass of ethanol, heat and reflux, and then filter to obtain a filtrate. After concentrating the filtrate, let it stand for a preset time, filter the filtrate after standing, and obtain fine apigenin after drying, thereby completing the synthesis of apigenin.

[0015] Preferably, during the synthesis, the mass ratio of the raw materials naringenin, pyridine and iodine is 1:6:0.8.

[0016] Preferably, the thin layer conditions for detecting the absence of naringenin are toluene:ethyl acetate:formic acid=5:4:1.

[0017] Preferably, in the step of preparing crude apigenin, the mass of the added hydrochloric acid is the mass of the raw material naringenin, and when hydrochloric acid and water are added at the same time, the mass ratio of the two is 1:1.

[0018] Preferably, in the apigenin refining step, the mass of ethanol added to the crude apigenin is 50 times that of the crude apigenin.

[0019] Preferably, in the apigenin refining step, a filtrate and insoluble matter are obtained after heating and reflux, the insoluble matter is added again with the same mass of ethanol and heated and refluxed again, and the two filtrates are combined to obtain a concentrated filtrate.

[0020] Preferably, the method of constructing a spectral curve according to the reflectivity and wavelength of the hyperspectral image band and obtaining the flatness of the spectral curve of the pixel is:

[0021] The reflectance of each band of the pixel in the hyperspectral image in the visible light and near-infrared bands is set as the ordinate, and the minimum wavelength of the band is set as the abscissa, and the spectral curve is obtained by fitting;

[0022] Obtain the peak points and trough points of the spectrum curve, as well as the peak height and peak width of the peaks where the peak points and trough points are located;

[0023] The expression of spectral curve flatness is:

[0024]

[0025] Where n arepresents the sum of the number of peaks and troughs in the spectral curve corresponding to the a-th pixel; h ′ a It represents the mean of the peak heights of all the peaks and valleys of the spectrum curve corresponding to the ath pixel, h ′ a ′ represents the standard deviation of the peak height values ​​of all peaks and troughs of the spectral curve corresponding to the ath pixel; w a ′ represents the mean of the peak widths of all peaks and valleys of the spectral curve corresponding to the ath pixel, w a ″ Represents the standard deviation of the peak width values ​​of all peaks and valleys of the spectral curve corresponding to the ath pixel, G a represents the flatness of the spectral curve of the a-th pixel; β represents the parameter adjustment factor.

[0026] Preferably, the method for obtaining the pixel spectral similarity between pixels according to the spectral curves of two pixels of the hyperspectral image and the flatness of the spectral curves is:

[0027] The spectral angle value between two pixels is obtained by using the spectral angle mapping algorithm;

[0028] The expression of pixel spectral similarity is:

[0029]

[0030] In the formula, G a Indicates the flatness of the spectral curve of the a-th pixel, G a′ Indicates the first ′ The flatness of the spectral curve of each pixel, θ a,a′ Indicates the a-th pixel and the a-th ′ The spectral angle value between pixels, β represents the parameter adjustment factor, S a,a′ Indicates the a-th pixel and the a-th ′ The pixel spectral similarity between pixels.

[0031] Preferably, the method of clustering the pixels by the spectral similarity of the pixels to obtain the bubble pixel suspicion of each type of pixel is:

[0032] The inverse of the pixel spectral similarity between pixels in the hyperspectral image is used as the metric distance of the clustering algorithm to obtain several pixel sets;

[0033] Obtain a two-dimensional image of the hyperspectral image. For each pixel set, mark it as 1 in the two-dimensional image, and mark the remaining pixels as 0. After connected domain analysis, obtain the clustering area of ​​each pixel set; calculate the circularity of each clustering area;

[0034] Calculate the coefficient of variation of the flatness of the spectral curves of all pixels in each clustering area; let the product of the inverse of the coefficient of variation of the clustering area and the circularity of the clustering area be the characteristic value of the bubble pixel in the clustering area;

[0035] The coefficient of variation of the bubble pixel characteristic values ​​of all clustered areas in each pixel set is calculated, and the inverse of the coefficient of variation of the pixel set is recorded as the bubble pixel suspicion degree of the pixel set.

[0036] Preferably, the method for obtaining the material uniformity index of the hyperspectral image according to the bubble pixel suspicion is:

[0037] A preset number of hyperspectral images without bubbles and hyperspectral images containing bubbles are obtained as standard spectral images, the bubble pixel suspicion of all standard spectral images is calculated, the bubble pixel suspicion of all standard spectral images is taken as input, and the bubble suspicion threshold is obtained using the maximum inter-class variance;

[0038] The pixel set whose bubble pixel suspicion degree in the hyperspectral image is greater than the bubble suspicion degree threshold is recorded as the bubble pixel set;

[0039] The hyperspectral image after removing the bubble pixel set from the hyperspectral image is recorded as a de-bubble spectrum image;

[0040] The pixel set in the de-bubble spectrum image is converted into a two-dimensional image according to the number of pixels, and the value of each pixel in the image is the label of its corresponding pixel set, so as to obtain a de-bubble two-dimensional image;

[0041] The two-dimensional information entropy of the debubbled two-dimensional image is used as the material uniformity index of the hyperspectral image.

[0042] The present invention has the following beneficial effects: the present application constructs the flatness of the spectral curve by analyzing the curve characteristics of the spectral curve of the pixels in the visible light and near-infrared bands in the hyperspectral image, and constructs the spectral similarity of the pixels in combination with the spectral angle value, which has the beneficial effect of reducing the difference in spectral characteristics between different pixels belonging to the same bubble in the hyperspectral image, and improving the distinction between pixels of different categories in the hyperspectral image; the circularity of the clustered area constructed by analyzing the various types of pixels in the hyperspectral image, and constructing the bubble pixel suspicion degree in combination with the spectral curve flatness, which has the beneficial effect of improving the distinction between the particle cluster area and the bubble area in the reaction solution area in the hyperspectral image; based on the bubble pixel suspicion degree A debubble hyperspectral image is obtained after bubble pixels are removed from the hyperspectral image, and a debubble material uniformity index is constructed, which has the beneficial effect of reducing the influence of bubbles in the reaction solution on the evaluation result of the material uniformity in the reaction solution; based on the change of the debubble material uniformity index over time, a neural network model is used to obtain the corrected purity of apigenin in the reaction solution sample at each sampling moment, which has the beneficial effect of improving the accuracy of the purity of apigenin in the solution sample and the overall purity of apigenin in the reaction solution; based on the change of the corrected purity of apigenin in the reaction solution sample over time, a neural network model is used to realize adaptive adjustment of the heating temperature of the solution in the reaction container in the synthesis step, thereby improving the purity of the synthesized apigenin. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 A synthetic process flow chart of apigenin provided by one embodiment of the present invention;

[0045] Figure 2 A flow chart of a reflux process in a synthesis process of apigenin provided by one embodiment of the present invention;

[0046] Figure 3 It is a schematic diagram of the hyperspectral image in the 500-550nm band;

[0047] Figure 4 Schematic diagram of the distribution of spectral curve flatness in the area where bubbles are located in the hyperspectral image. DETAILED DESCRIPTION

[0048] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the synthesis process of apigenin proposed according to the present invention, its specific implementation, structure, characteristics and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0050] A kind of synthesis process embodiment of apigenin:

[0051] The specific scheme of the synthesis process of apigenin provided by the present invention is described in detail below with reference to the accompanying drawings.

[0052] See also Figure 1 , which shows a synthetic process flow chart of apigenin provided by one embodiment of the present invention, the process comprises the following steps:

[0053] Step S001, apigenin raw material synthesis step.

[0054] Naringenin and pyridine were added to a reaction container, and iodine was added after being stirred and dissolved completely. Then, the temperature of the solution in the reaction container was raised to above 116° C., and the solution in the reaction container was refluxed.

[0055] Preferably, in one embodiment of the present application, the mass ratio of naringenin, pyridine and iodine is 1:6:0.8.

[0056] The process of reflux treatment in the reaction vessel is as follows:

[0057] S1, collects the hyperspectral image of the reaction liquid and the heating temperature.

[0058] During the reflux process, the solution in the reaction container is collected and recorded as a solution sample, and the solution samples are collected at different times; for each solution sample, its hyperspectral image and heating temperature are obtained;

[0059] Preferably, in one embodiment of the present application, the step of collecting the solution sample is:

[0060] The collection of solution samples is a continuous process. In this embodiment, the collection time is 3 seconds. After each collection, the solution is collected again after 3 minutes. In this embodiment, the number of collections is 10 times, and 2 ml of solution is collected each time.

[0061] Preferably, in another embodiment of the present application, the step of collecting the solution sample is:

[0062] In this embodiment, the collection time is 4 seconds, and the solution is collected again after 5 minutes after each collection. In this embodiment, the number of collections is 20 times, and 3 ml of solution is collected each time.

[0063] Preferably, in one embodiment of the present application, the step of obtaining the hyperspectral image and heating temperature of the reaction solution is:

[0064] A hyperspectral camera is placed directly above the reaction container, and the solution in the reaction container is recorded as the reaction solution. The hyperspectral camera is used to collect a hyperspectral image of the solution in the reaction container. In this embodiment, N frames of hyperspectral images are collected during each sampling process. In this embodiment, N is 30, which can be set by the implementer. The start time of each solution sample collection is recorded as the sampling time; the temperature of the reaction solution at the sampling time is collected using a temperature sensor and recorded as the heating temperature of the reaction solution. The pixels corresponding to the solvent or solid particles in the reaction solution in the hyperspectral image are called material pixels, and the pixels corresponding to the bubbles in the reaction solution in the hyperspectral image are called bubble pixels. The hyperspectral image is as follows: Figure 3 shown.

[0065] At this point, the hyperspectral image and heating temperature of the reaction solution were acquired.

[0066] S2, construct a spectral curve according to the reflectivity and wavelength of the hyperspectral image band, and obtain the flatness of the spectral curve of the pixel.

[0067] Since the bubbles in the reaction solution are essentially composed of air, which is composed of a variety of gas components, and gas molecules will cause light scattering, the reflectivity of the bubbles in the visible light and near-infrared light bands of the hyperspectral image is relatively consistent, without significant reflection peaks, and the bubbles usually have a relatively regular, approximately circular shape; while the solvent or solid particles in the reaction solution usually have obvious reflection peaks in the visible light and near-infrared light bands of the hyperspectral image, and the shape of the aggregation area of ​​the solid particles in the reaction solution is usually irregular; based on the different properties of the bubbles and materials, the characteristics of different pixels can be obtained.

[0068] Preferably, in one embodiment of the present application, a method for obtaining the flatness of a spectral curve of a pixel includes:

[0069] For any frame of hyperspectral image of the reaction solution, a coordinate system is constructed with the reflectance of each band in the visible light and near-infrared bands of any pixel in the hyperspectral image as the ordinate and the minimum wavelength of the band as the abscissa. The data points in the coordinate system are fitted with a cubic B-spline curve fitting algorithm to obtain the spectral curve of each pixel in the visible light and near-infrared bands.

[0070] It is worth noting that the cubic B-spline curve fitting algorithm is only a method for fitting a curve, and implementers can choose different curve fitting methods to obtain spectral curves.

[0071] For the spectral curve, the spectral curve is used as the input of the peak and trough second-order difference identification algorithm, all the peak points and trough points in the spectral curve are output, and the peak height and peak width values ​​of the peak where each peak point and trough point in the spectral curve is located are obtained.

[0072] It is worth noting that the peak and trough second-order difference identification algorithm is only a method to obtain the peaks and troughs of the curve. The implementer can obtain the peaks and troughs of the curve through different methods.

[0073] The expression of spectral curve flatness is:

[0074]

[0075] Where n a represents the sum of the number of peaks and troughs in the spectral curve corresponding to the a-th pixel; h ′ a It represents the mean of the peak heights of all the peaks and valleys of the spectrum curve corresponding to the ath pixel, h ′ a ′ represents the standard deviation of the peak height values ​​of all peaks and troughs of the spectral curve corresponding to the ath pixel; w a ′ represents the mean of the peak widths of all peaks and valleys of the spectral curve corresponding to the ath pixel, w a ″ Represents the standard deviation of the peak width values ​​of all peaks and valleys of the spectral curve corresponding to the ath pixel, G a It represents the flatness of the spectral curve of the a-th pixel; β represents the parameter adjustment factor, which is used to prevent the denominator from being 0, where the empirical value of β is 0.01, and norm() represents the linear normalization function.

[0076] The less significant changes or fluctuations there are on the spectral curve as a whole, the smaller the peak heights of all the peaks and valleys of the spectral curve, and the smaller the differences between all the peak heights, h ′ a ×h ′ a ′ The smaller the value is; the less sharp rise or fall changes are observed in the horizontal direction between the peaks and valleys on the spectrum curve, the larger the peak width values ​​of all the peak points and valley points of the spectrum curve are, and the smaller the difference between all the peak width values ​​is; at the same time, the fewer the number of peak points and valley points of the spectrum curve, the smaller the peak width value is. aThe smaller the value of G, the flatter the overall change of the spectrum curve. a The larger the value of is, the less significant absorption line or reflection peak the spectral curve of the a-th pixel has, and the more uniform absorption and reflection the spectral characteristics of the a-th pixel have in the visible and near-infrared bands. The flatness of the spectral curve in the area where the bubble is located in the hyperspectral image is as follows: Figure 4 shown.

[0077] At this point, the flatness of the spectral curve of each pixel in the hyperspectral image is obtained.

[0078] S3, obtaining the pixel spectral similarity between the pixels according to the spectral curves of two pixels of the hyperspectral image and the flatness of the spectral curves.

[0079] Since different elements and their compounds have their own unique spectral characteristics in hyperspectral images, the same substance in the reaction solution has similar spectral characteristics in different pixels corresponding to the hyperspectral image. However, the gas composition in the bubbles generated in the reaction solution may be different, such as the gas evaporated from the solvent, the mixed gas in the air, or the gas generated by the reaction liquid during the chemical reaction, which will lead to large differences in the spectral characteristics between the bubble pixels in the hyperspectral image. The spectral characteristics refer to the characteristics of a substance reflecting, absorbing or emitting light under electromagnetic radiation of different wavelengths. These characteristics can be represented by spectral curves. Therefore, it is necessary to calculate the similarity between different pixels and perform clustering based on the similarity.

[0080] Preferably, in one embodiment of the present application, the method for obtaining pixel spectral similarity includes:

[0081] The spectral curves of all pixels in the hyperspectral image are used as input, and the spectral angle values ​​between the pixels in the hyperspectral image are calculated using the spectral angle mapping algorithm, wherein the spectral angle mapping algorithm is a well-known technology and will not be described in detail in this embodiment. The pixel spectral similarity between two pixels is calculated by the spectral angle value between two pixels and the flatness of the spectral curve, and the formula is as follows:

[0082]

[0083] In the formula, G a Indicates the flatness of the spectral curve of the a-th pixel, G a′ Indicates the first ′ The flatness of the spectral curve of each pixel, θ a,a′ Indicates the a-th pixel and the a-th ′ The spectral angle value between pixels, β represents the parameter adjustment factor, S a,a′ Indicates the a-th pixel and the a-th ′ The pixel spectral similarity between pixels.

[0084] The a-th pixel and the a-th ′ The greater the possibility that the a-th pixel belongs to the same material pixel, the ′ The more similar the spectral characteristics between pixels are, the a,a′ The smaller the value of ′ The greater the possibility that the a-th pixel belongs to the bubble pixel, the ′ The flatter the spectral curve of a pixel in the visible and near-infrared bands, the a +G a′ ) value is larger, the a-th pixel and the a-th pixel are ′ The closer the values ​​of the spectral curve flatness of the pixels in the visible light and near-infrared light bands are, the more similar the two pixels are. a -G a′ The smaller the value of |, the higher the S a,a′ The larger the value of ′ The greater the possibility of pixels belonging to the same class.

[0085] At this point, the pixel spectral similarity between any two pixels in the hyperspectral image is obtained.

[0086] S4, clustering is performed through pixel spectral similarity to obtain the bubble pixel suspicion degree of each type of pixel.

[0087] Since material pixels and bubble pixels have different characteristics, different pixels can be classified according to the similarity of pixel spectra between pixels, and then the pixels can be distinguished.

[0088] Preferably, in one embodiment of the present application, the method for obtaining the bubble pixel suspicion degree includes:

[0089] The inverse of the pixel spectral similarity between pixels in the hyperspectral image is used as the metric distance in the AP clustering algorithm. The AP clustering algorithm is used to divide all pixels in the hyperspectral image to obtain K pixel sets. Different pixel sets correspond to substances or bubbles in the reaction solution. In this embodiment, 10 pixel sets are obtained.

[0090] It is worth noting that the AP clustering algorithm is only an algorithm for clustering all pixels. Implementers can use clustering algorithms with the same function to replace them and cluster the pixels.

[0091] For each pixel set, in the hyperspectral image, the pixel corresponding to the pixel set is marked as 1, and the remaining pixels are marked as 0, thereby obtaining a two-dimensional image of the hyperspectral image, in which the value of each pixel is the label of each pixel. Connected domain analysis is used for the two-dimensional image to obtain several connected domains of the pixel set, and each connected domain is recorded as an aggregation area. For each pixel set, the same method is used to obtain the aggregation area of ​​the pixel set, and the circularity of each aggregation area is calculated. The calculation of the circularity is a well-known method and will not be described in detail in this embodiment.

[0092] The bubble pixel suspicion of the pixel set is obtained by the circularity of different clustering areas and the flatness of the spectral curve of the pixel. The formula is as follows:

[0093]

[0094] In the formula, G k,m represents the flatness of the spectral curve of the cth pixel in the mth cluster area in the kth pixel set, ω(G k,m,c ) represents the coefficient of variation of the spectral curve flatness of all pixels in the mth cluster area in the kth pixel set, U k,m represents the circularity of the mth clustered area in the kth pixel set, f k,m represents the characteristic value of the bubble pixel in the mth cluster area in the kth pixel set, ω(f k,m ) represents the coefficient of variation of the characteristic values ​​of the bubble pixels in all clustered areas in the k-th pixel set, F k Indicates the bubble pixel suspicion of the k-th pixel set.

[0095] The calculation method of the coefficient of variation is an existing method and will not be described in detail in this embodiment.

[0096] Among them, the closer the flatness of the spectral curve of each clustering area is, the greater the coefficient of variation of the clustering area. Since the reflectivity of bubbles in different light bands is similar, the flatness of the spectral curve corresponding to the bubble pixels is not much different. Therefore, the greater the coefficient of variation, the greater the possibility that the pixels in the clustering area are bubble pixels; and the greater the circularity of the clustering area, the more the set is like a circle, and bubbles are generally round, so the greater the circularity, the greater the possibility that the pixels in the clustering area are bubble pixels. The larger the characteristic value of the bubble pixel, the more likely the corresponding pixel is a bubble pixel. If the characteristic value of the bubble pixel of all clustering areas in the pixel set is larger and the similarity between the clustering areas is higher, then all the pixels in the pixel set are more likely to be bubble pixels. Therefore, the higher the bubble pixel suspicion of the pixel set, the more likely the pixel in the pixel set is to be a bubble pixel.

[0097] At this point, the bubble pixel suspicion degree of each pixel set is obtained.

[0098] S5, obtaining the material uniformity index of the hyperspectral image according to the bubble pixel suspicion.

[0099] The reaction solution needs to be temperature-regulated subsequently, and bubbles in the reaction solution will affect the accuracy of temperature regulation. Therefore, it is necessary to confirm the bubble part in the hyperspectral image based on the bubble pixel suspicion obtained above, and calculate the material uniformity index of the hyperspectral image after removing the bubble part. Through this material uniformity index, the current state of the reaction solution can be analyzed more accurately, so as to better regulate the temperature.

[0100] Preferably, in one embodiment of the present application, the method for obtaining the material uniformity index includes:

[0101] 50 hyperspectral images containing bubbles and 50 hyperspectral images without bubbles are obtained by artificial means respectively. These known hyperspectral images are recorded as standard spectral images, where the number of obtained images is set by the implementer. In this embodiment, 50 images are selected just to make the subsequent results more convincing. The bubble pixel suspicion of each pixel set of each standard spectral image is obtained, and the bubble pixel suspicion of all pixel sets corresponding to all standard spectral images is used as the input of the maximum inter-class variance algorithm to obtain the bubble suspicion threshold, which is used to determine whether all pixels in the pixel set in the hyperspectral image are bubble pixels. The maximum inter-class variance algorithm is a well-known technology and will not be elaborated here.

[0102] Obtain a pixel set in the hyperspectral image whose bubble pixel suspicion is greater than the bubble suspicion threshold, record this pixel set as the bubble pixel set, and remove all pixels of the bubble pixel set from the hyperspectral image to obtain the hyperspectral image as the bubble-free spectrum map. Label the pixel set in the bubble-free spectrum map according to the number of pixels in it, convert the hyperspectral image into a two-dimensional image, and the value of each pixel in the two-dimensional image is the label of the pixel set corresponding to it, thereby obtaining the bubble-free two-dimensional image.

[0103] The two-dimensional information entropy of the debubble two-dimensional image is calculated, and this value is recorded as the material uniformity index of the hyperspectral image. The larger the value, the more uniform the material distribution in the non-bubble area of ​​the reaction solution at the sampling moment corresponding to the hyperspectral image.

[0104] At this point, the material uniformity index of the hyperspectral image has been obtained.

[0105] S6, obtaining a temperature adjustment value according to the uniformity index of the bubble material, and adaptively adjusting the heating temperature until no naringenin is detected in the thin layer, and then stopping the reflux to obtain the reaction solution.

[0106] Preferably, in another embodiment of the present application, the method for adaptively adjusting the heating temperature is:

[0107] Obtain the material uniformity index of all frames of hyperspectral images corresponding to each acquisition, and arrange the obtained material uniformity index in ascending order according to the acquisition frame number of the hyperspectral image to obtain the uniform distribution time series of each acquisition. At the same time, the purity of apigenin in the reaction solution at each acquisition is determined by high performance liquid chromatography, wherein high performance liquid chromatography is a known technology, and this embodiment will not be repeated. The uniform distribution time series and the purity of apigenin collected each time are used as the input of the LSTM (Long Short Term Memory) neural network model, wherein data annotation is performed by professionals, and the stochastic gradient descent method and the mean square error are used as the optimization algorithm and loss function of the LSTM neural network model respectively, and the corrected purity of apigenin in the reaction solution at each acquisition is output, wherein the training of the LSTM neural network model is a known technology, and the specific process will not be repeated.

[0108] The corrected purity corresponding to each collection is obtained in the above manner, wherein the number of collections closest to the current moment is selected each time, and the corrected purity is sorted in ascending order according to the sampling moments to obtain the apigenin purity change sequence of the reaction solution, and the heating temperature at the latest sampling moment and the apigenin purity change sequence are used as inputs to the LSTM neural network model, wherein data annotation is performed by professionals, and the stochastic gradient descent method and mean square error are used as the optimization algorithm and loss function of the LSTM neural network model respectively, and the output is the temperature adjustment value at the next sampling moment, and this temperature adjustment value is the heating temperature at the next sampling.

[0109] When the temperature of the reaction solution changes and refluxes, the solution sample is detected by thin layer chromatography simultaneously. When naringenin is not detected by thin layer chromatography, the heating and reflux are stopped, and the reaction solution at this time is recorded as the reaction solution. The flow chart of the reflux process is shown in FIG. Figure 2 shown.

[0110] Preferably, in another embodiment of the present application, the developing solvent of thin layer chromatography is toluene:ethyl acetate:formic acid=5:4:1, which is recorded as the thin layer condition of thin layer detection.

[0111] Step S002, step of preparing crude apigenin.

[0112] Preferably, in another embodiment of the present application, the method for obtaining crude apigenin is:

[0113] The reaction liquid obtained in the product synthesis step is evaporated to obtain a solid, and pyridine is recovered after evaporation. Hydrochloric acid and water with the same mass as that of naringenin in the raw material are added to the solid. The mass ratio of hydrochloric acid and water added at this time is 1:1. After the solid is filtered out, water is added until it is washed to neutral. The solid obtained at this time is crude apigenin, and the content is about 95%.

[0114] At this point, crude apigenin was obtained.

[0115] Step S003, apigenin refining step.

[0116] Preferably, in another embodiment of the present application, the method for obtaining fine apigenin is:

[0117] For crude apigenin, add 50 times the mass of ethanol, heat and reflux for 1 hour, filter out the insoluble matter, add 50 times the mass of ethanol to the insoluble matter and heat and reflux, combine the two filtrates, concentrate to 1 / 6 of the volume, let the concentrated filtrate stand for 4 hours, filter out and dry to obtain fine apigenin with a content of more than 98%.

[0118] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0119] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A synthesis process of apigenin, characterized in that, The process includes the following steps: The synthesis steps of apigenin raw materials are as follows: after mixing naringenin and pyridine, iodine is added, the temperature of the solution in the reaction container is raised to 116°C, and the solution is refluxed; when no naringenin is detected by thin layer detection, the reflux is stopped to obtain the reaction solution; the reflux process includes the following steps: Collecting hyperspectral images of the reaction liquid and the heating temperature; Construct a spectral curve based on the reflectance and wavelength of the hyperspectral image band to obtain the flatness of the spectral curve of the pixel; Obtain the pixel spectral similarity between pixels according to the spectral curves and spectral curve flatness of two pixels of the hyperspectral image; Clustering is performed through pixel spectral similarity to obtain the bubble pixel suspicion of each type of pixel; The material uniformity index of the hyperspectral image is obtained according to the bubble pixel suspicion; Obtain the temperature adjustment value according to the material uniformity index and adjust the heating temperature adaptively; Preparation step of crude apigenin: evaporating the reaction liquid obtained by stopping reflux in the apigenin raw material synthesis step to obtain a solid, adding hydrochloric acid and water of preset mass to the solid until it is neutral, and obtaining crude apigenin; Apigenin refining step: for crude apigenin, add a preset mass of ethanol, heat and reflux, filter to obtain a filtrate, concentrate the filtrate, let it stand for a preset time, filter the filtrate after standing, dry it to obtain fine apigenin, and complete the synthesis of apigenin; The method of clustering the pixel spectrum similarity to obtain the bubble pixel suspicion of each type of pixel is: The inverse of the pixel spectral similarity between pixels in the hyperspectral image is used as the metric distance of the clustering algorithm to obtain several pixel sets; Obtain a two-dimensional image of the hyperspectral image. For each pixel set, mark it as 1 in the two-dimensional image, and mark the remaining pixels as 0. After connected domain analysis, obtain the clustering area of ​​each pixel set; calculate the circularity of each clustering area; Calculate the coefficient of variation of the flatness of the spectral curves of all pixels in each clustering area; let the product of the inverse of the coefficient of variation of the clustering area and the circularity of the clustering area be the characteristic value of the bubble pixel in the clustering area; Calculate the coefficient of variation of the bubble pixel characteristic values ​​of all clustered areas in each pixel set, and record the inverse of the coefficient of variation of the pixel set as the bubble pixel suspicion of the pixel set; The method for obtaining the material uniformity index of a hyperspectral image according to the bubble pixel suspicion is as follows: A preset number of hyperspectral images without bubbles and hyperspectral images containing bubbles are obtained as standard spectral images, the bubble pixel suspicion of all standard spectral images is calculated, the bubble pixel suspicion of all standard spectral images is taken as input, and the bubble suspicion threshold is obtained using the maximum inter-class variance; The pixel set whose bubble pixel suspicion degree in the hyperspectral image is greater than the bubble suspicion degree threshold is recorded as the bubble pixel set; The hyperspectral image after removing the bubble pixel set from the hyperspectral image is recorded as a de-bubble spectrum image; The pixel set in the de-bubble spectrum image is converted into a two-dimensional image according to the number of pixels, and the value of each pixel in the image is the label of its corresponding pixel set, so as to obtain a de-bubble two-dimensional image; The two-dimensional information entropy of the debubbled two-dimensional image is used as the material uniformity index of the hyperspectral image.

2. A synthesis process for apigenin as claimed in claim 1, characterized in that, During the synthesis, the mass ratio of the raw materials naringenin, pyridine and iodine is 1:6:0.

8.

3. A synthesis process for apigenin as claimed in claim 1, characterized in that, The thin layer conditions for the thin layer detection of naringenin-free are toluene: Ethyl acetate:formic acid=5:4:

1.

4. A synthesis process for apigenin as claimed in claim 1, characterized in that, In the step of preparing crude apigenin, the mass of hydrochloric acid added is the mass of raw material naringenin, and when hydrochloric acid and water are added at the same time, the mass ratio of the two is 1:

1.

5. A synthesis process for apigenin as claimed in claim 1, characterized in that, In the apigenin refining step, the mass of ethanol added to the crude apigenin is 50 times that of the crude apigenin.

6. A process for synthesizing apigenin as claimed in claim 1, characterized in that: In the apigenin refining step, a filtrate and insoluble matter are obtained after heating and reflux, the insoluble matter is added with the same mass of ethanol again and heated and refluxed again, and the two filtrates are combined to obtain a concentrated filtrate.

7. A process for synthesizing apigenin as claimed in claim 1, characterized in that: The method of constructing a spectral curve according to the reflectivity and wavelength of the hyperspectral image band and obtaining the flatness of the spectral curve of the pixel is: The reflectance of each band of the pixel in the hyperspectral image in the visible light and near-infrared bands is set as the ordinate, and the minimum wavelength of the band is set as the abscissa, and the spectral curve is obtained by fitting; Obtain the peak points and trough points of the spectrum curve, as well as the peak height and peak width of the peaks where the peak points and trough points are located; The expression of spectral curve flatness is: Where n a represents the sum of the number of peaks and troughs in the spectral curve corresponding to the a-th pixel; h ′ a It represents the mean of the peak heights of all the peaks and valleys of the spectrum curve corresponding to the ath pixel, h ′ a ′ represents the standard deviation of the peak height values ​​of all peaks and troughs of the spectral curve corresponding to the ath pixel; w a ′ represents the mean of the peak widths of all peaks and valleys of the spectral curve corresponding to the ath pixel, w a ″ Represents the standard deviation of the peak width values ​​of all peaks and valleys of the spectral curve corresponding to the ath pixel, G a It represents the flatness of the spectral curve of the a-th pixel; β represents the parameter adjustment factor, and norm() represents the linear normalization function.

8. A process for synthesizing apigenin as claimed in claim 1, characterized in that: The method for obtaining the pixel spectral similarity between pixels according to the spectral curves of two pixels of the hyperspectral image and the flatness of the spectral curves is: The spectral angle value between two pixels is obtained by using the spectral angle mapping algorithm; The expression of pixel spectral similarity is: In the formula, G a Indicates the flatness of the spectral curve of the a-th pixel, G a′ Indicates the first ′ The flatness of the spectral curve of each pixel, θ a,a′ Indicates the a-th pixel and the a-th ′ The spectral angle value between pixels, β represents the parameter adjustment factor, S a,a′ Indicates the a-th pixel and the a-th ′ The pixel spectral similarity between pixels.

Images