Automatic preparation method and system of glycerophosphatidylcholine

Abstract

The invention discloses an automatic preparation method and system of glyceryl phosphatidylcholine. By constructing a three-dimensional by-product matrix of time points, temperatures, by-products and by-product volumes, and by constructing a special three-dimensional detection network, the change of by-products and by-product volumes generated at different temperatures at adjacent time points is detected. And adjusting the first feature vector. Influences of time, temperature increase and temperature interval length increase on by-products are detected. Therefore, a predicted stop time point capable of reducing by-products and corresponding by-product volumes is found. The reasonable temperature is adjusted at multiple time points in real time according to the historical time points and the preparation condition of the glycerophosphatidylcholine at the corresponding temperature. The reaction rate is improved, and the purity and yield of the product are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of automated production, and more particularly, to an automated preparation method and system for glycerophosphatidylcholine. Background Art

[0002] Currently, glycerophosphatidylcholine (GPC) is a common phospholipid compound, widely used in fields such as drug delivery, nutritional health care, cosmetics, and the food industry. The synthesis reaction process of glycerophosphatidylcholine (GPC, Glycerophosphatidylcholine) mainly combines triglyceride with choline compounds (such as phosphocholine) through a phosphatidylation reaction to generate glycerophospholipids containing phospholipid groups. This process can generally be carried out by chemical synthesis or enzymatic catalysis. The method for automated preparation of glycerophosphatidylcholine usually involves the optimization of the lipid synthesis process and automated operation. The reaction temperature is generally between 50 - 100 °C, and the reaction time depends on the use of the catalyst and the requirements of the reaction. Phospholipase D can catalyze the phosphatidylation reaction between triglyceride and phosphocholine to generate glycerophosphatidylcholine, accompanied by various by-products.

[0003] Generally, increasing the temperature can accelerate chemical reactions because it increases the collision frequency and energy between molecules, thereby increasing the reaction rate. However, too high a temperature may lead to side reactions or degradation reactions, especially when using solvents and catalysts. When synthesizing glycerophosphatidylcholine, too high a temperature may cause other unwanted esterification reactions or incomplete reactions, thereby reducing the purity of the product. Controlling the appropriate temperature helps to improve the selectivity of the reaction and avoid the formation of unnecessary by-products.

[0004] At the same time, the reaction time is another very important parameter in the synthesis process. Controlling the time can help optimize the purity and yield of the product and reduce the formation of by-products.

[0005] The currently adopted method usually involves finding a fixed temperature for a fixed duration to prepare glycerophosphatidylcholine. It cannot adjust the reasonable temperature at multiple time points in real time according to the preparation situation of glycerophosphatidylcholine at historical time points and corresponding temperatures, and stop at an appropriate duration. Summary of the Invention

[0006] The purpose of the present invention is to provide an automated preparation method and system for glycerophosphatidylcholine to solve the above problems existing in the prior art.

[0007] In a first aspect, an embodiment of the present invention provides an automated preparation method for glycerophosphatidylcholine, including:

[0008] Obtain the temperatures, by-products, and the corresponding by-product volumes at multiple time points, and construct a by-product matrix;

[0009] Based on the by-product matrix, through a three-dimensional detection network, obtain a first feature vector; the first feature vector is used to judge the changes in the volumes of products and by-products at different temperatures at adjacent time points;

[0010] Based on the by-product matrix and the first feature vector, obtain multiple time-temperature feature sequences;

[0011] Based on the by-product matrix, the time-temperature feature sequences, and the first feature vector, obtain a predicted stop time point and a predicted temperature set; the predicted temperature set represents the predicted temperatures corresponding to multiple time points between the current time point and the stop time point; the predicted temperature represents the temperature predicted for preparing glycerophosphatidylcholine at a future time point.

[0012] Optionally, the obtaining of the predicted stop time point and the predicted temperature set based on the by-product matrix, the time-temperature feature sequences, and the first feature vector includes:

[0013] Obtain a first time point and a second time point; the first time point and the second time point are adjacent in the by-product matrix; the first time point is earlier than the second time point;

[0014] Take the increment of the by-product volume between the first time point and the second time point as the first by-product increment; multiple first by-product increments are obtained corresponding to multiple time points;

[0015] Arrange the temperatures corresponding to the first time point in ascending order of the first by-product increment to obtain a first temperature sequence;

[0016] Recombine the values in the first feature vector into a second feature vector according to the first temperature sequence;

[0017] Arrange the multiple time-temperature feature sequences in ascending order of the time-temperature change value to obtain a set of feature sequences;

[0018] Based on the first feature vector, the second feature vector, and the set of feature sequences, obtain the predicted stop time point and the predicted temperature set.

[0019] Optionally, the obtaining of the first feature vector based on the by-product matrix through a three-dimensional detection network includes:

[0020] Input the by-product matrix into the three-dimensional detection network, extract the changes in products and by-products at different temperatures among adjacent time points, and obtain the first feature vector;

[0021] The convolution kernel of the three-dimensional detection network is a three-dimensional convolution kernel of n*m*2; 2 corresponds to the time point; n represents the temperature; m represents the number of by-products;

[0022] Among them, n is the number of rows of the by-product matrix, and m is the number of columns of the by-product matrix;

[0023] Convolve the three-dimensional convolution kernel on the by-product matrix in chronological order from early to late with a step size of 1.

[0024] Optionally, obtaining multiple time-temperature feature sequences based on the by-product matrix and the first eigenvector includes:

[0025] Based on the by-product matrix, obtain multiple time-temperature change values; the time-temperature change value represents the change in temperature;

[0026] Retain the features with the same time-temperature change value in the first eigenvector to obtain multiple time-temperature features; the time-temperature feature represents the feature of the change in the by-product volume when the temperature changes are the same for two adjacent time points;

[0027] Sort the multiple time-temperature features in chronological order from early to late to obtain a time-temperature feature sequence;

[0028] Multiple time-temperature change values correspond to obtaining multiple time-temperature feature sequences.

[0029] Optionally, obtaining the predicted stop time point and the predicted temperature set based on the first eigenvector, the second eigenvector, and the feature pair sequence set includes:

[0030] Input the first eigenvector into the first neural network to detect the influence of time on by-products and obtain the first change feature;

[0031] Input the second eigenvector into the second neural network to detect the influence of temperature increase on by-products and obtain the second change feature;

[0032] Input the feature pair sequence set into the third neural network to detect the influence of the increase in the interval length of temperature on by-products and obtain the third change feature;

[0033] Through the first prediction network and the second prediction network, based on the first change feature, the second change feature, and the third change feature, obtain the predicted stop time point and the predicted temperature set.

[0034] Optionally, obtaining multiple time-temperature change values based on the by-product matrix includes:

[0035] Construct a temperature curve based on the temperatures corresponding to multiple time points of the by-product matrix from the earliest to the latest time point; the temperature curve represents the temperatures at historical time points.

[0036] Derive the positions of the temperature curve at the time points to obtain multiple time-temperature change values.

[0037] Optionally, the obtaining of the predicted stop time point and the predicted temperature set based on the first change feature, the second change feature, and the third change feature through the first prediction network and the second prediction network includes:

[0038] Input the first change feature, the second change feature, and the third change feature into the first prediction network to obtain the predicted stop time point.

[0039] Subtract the current time point from the predicted stop time point to obtain the predicted time length.

[0040] Input the first change feature, the second change feature, the third change feature, and the predicted time length into the second prediction network to obtain multiple predicted temperatures.

[0041] Construct a predicted temperature set from the multiple predicted temperatures from the earliest to the latest time point.

[0042] Optionally, the obtaining of the temperatures, by-products, and corresponding by-product volumes at multiple time points and constructing a by-product matrix includes:

[0043] Use the time points as the rows of the by-product matrix.

[0044] Use the temperatures as the columns of the by-product matrix.

[0045] Use the by-products as the heights of the by-product matrix.

[0046] Among them, different heights in the by-product matrix represent different by-products.

[0047] Find the positions corresponding to the time points, temperatures, and by-products in the by-product matrix and fill in the corresponding by-product volumes.

[0048] In a second aspect, an embodiment of the present invention provides an automated preparation system for glycerophosphatidylcholine, including:

[0049] An acquisition module for acquiring the temperatures, by-products, and corresponding by-product volumes at multiple time points and constructing a by-product matrix.

[0050] A three-dimensional detection module for obtaining a first feature vector based on the by-product matrix through a three-dimensional detection network; the first feature vector is used to judge the changes in the volumes of products and by-products at different temperatures between adjacent time points.

[0051] A temperature change detection module, configured to obtain a plurality of time-temperature feature sequences based on a by-product matrix and a first eigenvector;

[0052] A prediction module, configured to obtain a predicted stop time point and a predicted temperature set based on the by-product matrix, the time-temperature feature sequences, and the first eigenvector; the predicted temperature set represents predicted temperatures corresponding to a plurality of time points between the current time point and the stop time point; the predicted temperature represents the predicted temperature for preparing glycerophosphatidylcholine at a future time point.

[0053] Optionally, obtaining the predicted stop time point and the predicted temperature set based on the by-product matrix, the time-temperature feature sequences, and the first eigenvector includes:

[0054] Obtain a first time point and a second time point; the first time point and the second time point are adjacent in the by-product matrix; the first time point is earlier than the second time point;

[0055] Take the increment of the by-product volume between the first time point and the second time point as the first by-product increment; a plurality of first by-product increments are obtained corresponding to a plurality of time points;

[0056] Arrange the temperatures corresponding to the first time point in ascending order of the first by-product increment to obtain a first temperature sequence;

[0057] Recombine the values in the first eigenvector into a second eigenvector according to the first temperature sequence;

[0058] Arrange the plurality of time-temperature feature sequences in ascending order of the time-temperature change value to obtain a set of feature sequences;

[0059] Obtain the predicted stop time point and the predicted temperature set based on the first eigenvector, the second eigenvector, and the set of feature sequences.

[0060] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects:

[0061] The embodiments of the present invention also provide an automated method and system for preparing glycerophosphatidylcholine.

[0062] In the present invention, a three-dimensional by-product matrix of time points, temperature, by-products, and by-product volumes is constructed, and a special three-dimensional detection network is constructed to detect the changes in by-products and by-product volumes generated at different temperatures at adjacent time points. By adjusting the first eigenvector, the influence of time on by-products, the influence of temperature increase on by-products, and the influence of the increase in the temperature interval length on by-products are detected. Thus, a predicted stop time point that can reduce by-products and the corresponding by-product volumes is found. And according to the preparation situation of glycerophosphocholine at historical time points and the corresponding temperatures, the reasonable temperature is adjusted in real time at multiple time points. The technical effects of improving the reaction rate, optimizing the purity and yield of the product are achieved. Description of the Drawings

[0063] Figure 1 is a flowchart of an automated preparation method of glycerophosphocholine provided by an embodiment of the present invention. Detailed Embodiments

[0064] The present invention will be described in detail below with reference to the accompanying drawings.

[0065] Example 1

[0066] As Figure 1 shown, an embodiment of the present invention provides an automated preparation method of glycerophosphocholine, and the method includes:

[0067] S101: Obtain the temperature, products and corresponding product volumes, by-products and corresponding by-product volumes at multiple time points, and construct a by-product matrix.

[0068] Among them, one time point corresponds to one temperature.

[0069] Among them, in this embodiment, the by-products and the corresponding by-product volumes are identified by the target detection model yolov5.

[0070] S102: Based on the by-product matrix, obtain a first eigenvector through a three-dimensional detection network; the first eigenvector is used to judge the volume changes of products and by-products at different temperatures at adjacent time points.

[0071] S103: Based on the by-product matrix and the first eigenvector, obtain multiple time-temperature feature sequences.

[0072] S104: Based on the time-temperature feature sequence and the first eigenvector, obtain a predicted stop time point and a predicted temperature set; the predicted temperature set represents the predicted temperatures corresponding to multiple time points between the current time point and the stop time point; the predicted temperature represents the temperature for preparing glycerophosphocholine at a future time point.

[0073] Optionally, obtaining the predicted stop time point and the predicted temperature set based on the by-product matrix, the time-temperature feature sequence, and the first feature vector includes:

[0074] Obtain a first time point and a second time point; the first time point and the second time point are adjacent in the by-product matrix; the first time point is earlier than the second time point;

[0075] Take the increment of the by-product volume between the first time point and the second time point as the first by-product increment; multiple first by-product increments are obtained corresponding to multiple time points;

[0076] Arrange the temperatures corresponding to the first time point in ascending order of the first by-product increment to obtain a first temperature sequence;

[0077] Recombine the values in the first feature vector into a second feature vector according to the first temperature sequence.

[0078] Wherein, the second feature vector represents the feature in the first feature vector that makes the first by-product increment increase from small to large at different temperatures;

[0079] Arrange the multiple time-temperature feature pair sequences in ascending order of the time-temperature change value to obtain a set of feature pair sequences;

[0080] Obtain the predicted stop time point and the predicted temperature set based on the first feature vector, the second feature vector, and the set of feature pair sequences.

[0081] Optionally, obtaining the first feature vector based on the by-product matrix through a three-dimensional detection network includes:

[0082] Input the by-product matrix into the three-dimensional detection network, and extract the changes of the products and by-products at different temperatures among adjacent time points to obtain the first feature vector.

[0083] Wherein, in this embodiment, the three-dimensional detection network is a three-dimensional convolutional neural network (3D Convolutional Neural Networks, CNN).

[0084] The convolutional kernel of the three-dimensional detection network is a three-dimensional convolutional kernel of n*m*2; 2 corresponds to the time point; n represents the temperature; m represents the number of by-products;

[0085] Wherein, n is the number of rows of the by-product matrix, and m is the number of columns of the by-product matrix.

[0086] Wherein, m and n are positive integers.

[0087] Convolve the three-dimensional convolutional kernel on the by-product matrix step by step from the earliest time point with a step size of 1.

[0088] Optionally, obtaining a plurality of time-temperature feature sequences based on the by-product matrix and the first eigenvector includes:

[0089] Obtaining a plurality of time-temperature change values based on the by-product matrix; the time-temperature change values represent temperature changes;

[0090] Retaining the features with the same time-temperature change value in the first eigenvector to obtain a plurality of time-temperature features; the time-temperature features represent the features of the by-product volume change when the temperature changes between adjacent two time points are the same;

[0091] Sorting the plurality of time-temperature features from the earliest time point to the latest time point to obtain a time-temperature feature sequence.

[0092] A plurality of time-temperature change values respectively obtain a plurality of time-temperature feature sequences.

[0093] Optionally, obtaining a predicted stop time point and a predicted temperature set based on the first eigenvector, the second eigenvector, and the feature pair sequence set includes:

[0094] Inputting the first eigenvector into a first neural network to detect the influence of time on by-products and obtaining a first change feature.

[0095] Wherein, the first neural network is a fully connected neural network (FCN).

[0096] Inputting the second eigenvector into a second neural network to detect the influence of temperature growth on by-products and obtaining a second change feature.

[0097] Wherein, the second neural network is a temporal convolutional network (TCN).

[0098] Inputting the feature pair sequence set into a third neural network to detect the influence of the growth of the temperature interval length on by-products and obtaining a third change feature.

[0099] Wherein, the third neural network is a recurrent neural network (RNN).

[0100] Through a first prediction network and a second prediction network, obtaining a predicted stop time point and a predicted temperature set based on the first change feature, the second change feature, and the third change feature.

[0101] Optionally, obtaining a plurality of time-temperature change values based on the by-product matrix includes:

[0102] Construct a temperature curve based on the temperatures corresponding to multiple time points of the by-product matrix from the earliest to the latest time point; the temperature curve represents the temperatures at historical time points.

[0103] Among them, in this embodiment, a polynomial interpolation method is used to construct the temperature curve.

[0104] Derive the position of the temperature curve at the time point to obtain multiple time-temperature change values.

[0105] Optionally, the obtaining of the predicted stop time point and the predicted temperature set based on the first change feature, the second change feature, and the third change feature through the first prediction network and the second prediction network includes:

[0106] Input the first change feature, the second change feature, and the third change feature into the first prediction network to obtain the predicted stop time point.

[0107] Among them, in this embodiment, the first prediction network is a fully connected neural network (FullyConnectedNetural Network, FCN).

[0108] Subtract the current time point from the predicted stop time point to obtain the predicted time length.

[0109] Among them, the current time point represents the current time point except for future time points and historical time points.

[0110] Input the first change feature, the second change feature, the third change feature, and the predicted time length into the second prediction network to obtain multiple predicted temperatures.

[0111] Among them, in this embodiment, the second prediction network is a fully connected neural network (FullyConnectedNetural Network, FCN).

[0112] Construct a predicted temperature set from the multiple predicted temperatures from the earliest to the latest time point.

[0113] Optionally, the obtaining of the temperatures, by-products, and corresponding by-product volumes at multiple time points to construct a by-product matrix includes:

[0114] Use the time point as the row of the by-product matrix.

[0115] Among them, the time periods from small to large correspond to the subscripts of the rows of the by-product matrix from small to large.

[0116] Use the temperature as the column of the by-product matrix.

[0117] Among them, in this embodiment, the temperature ranges from 30 degrees Celsius to 100 degrees Celsius, with each increase of 5 degrees Celsius represented as a column. For example, the column with subscript 0 represents 30 degrees Celsius, and the column with subscript 1 represents 35 degrees Celsius.

[0118] Use the by-products as the height of the by-product matrix.

[0119] Among them, different columns in the by-product matrix represent different by-products.

[0120] Find the positions corresponding to the time point, temperature, and by-products in the by-product matrix, and fill in the corresponding by-product volumes.

[0121] Embodiment 2

[0122] Based on the above-mentioned automated preparation method of glycerophosphocholine, an embodiment of the present invention further provides an automated preparation system for glycerophosphocholine, which includes an acquisition module, a three-dimensional detection module, a temperature change detection module, and a prediction module.

[0123] The acquisition module is used to acquire the temperature, by-products, and corresponding by-product volumes at multiple time points, and construct a by-product matrix;

[0124] The three-dimensional detection module is used to obtain a first feature vector based on the by-product matrix through a three-dimensional detection network; the first feature vector is used to judge the change in the volumes of the product and by-products at different temperatures at adjacent time points;

[0125] The temperature change detection module is used to obtain multiple time-temperature feature sequences based on the by-product matrix and the first feature vector;

[0126] The prediction module is used to obtain a predicted stop time point and a predicted temperature set based on the by-product matrix, the time-temperature feature sequence, and the first feature vector; the predicted temperature set represents the predicted temperatures corresponding to multiple time points between the current time point and the stop time point; the predicted temperature represents the temperature predicted for preparing glycerophosphocholine at a future time point.

[0127] Optionally, obtaining the predicted stop time point and the predicted temperature set based on the by-product matrix, the time-temperature feature sequence, and the first feature vector includes:

[0128] Obtain a first time point and a second time point; the first time point and the second time point are adjacent in the by-product matrix; the first time point is earlier than the second time point;

[0129] Take the increment of the by-product volume between the first time point and the second time point as the first by-product increment; multiple first by-product increments are obtained corresponding to multiple time points;

[0130] Arrange the temperatures corresponding to the first time point in ascending order of the first by-product increment to obtain a first temperature sequence;

[0131] Recombine the values in the first feature vector into a second feature vector according to the first temperature sequence;

[0132] Arrange the multiple time-temperature feature sequences in ascending order of the time-temperature change value to obtain a set of feature sequences;

[0133] Based on the first feature vector, the second feature vector, and the set of feature sequences, obtain a predicted stop time point and a set of predicted temperatures.

[0134] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The structure required to construct such systems will be apparent from the above description. In addition, the present invention is not directed to any particular programming language. It should be understood that the content of the present invention described herein can be implemented using various programming languages, and the description of the specific language above is to disclose the best mode of the present invention.

[0135] In the specification provided herein, a large number of specific details are set forth. However, it can be understood that the embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0136] Each component embodiment of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that a microprocessor or a digital signal processor (DSP) can be used in practice to implement some or all of the functions of some or all of the components in the device according to the embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and a computer program product) for performing part or all of the methods described herein. Such a program for implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

Claims

1. An automated preparation method for glycerophosphatidylcholine, characterized in that: include: Obtain the temperature, byproducts and corresponding byproduct volumes at multiple time points to construct a byproduct matrix; Based on the byproduct matrix, a first eigenvector is obtained through a three-dimensional detection network; the first eigenvector is used to determine the volume changes of products and byproducts at different temperatures at adjacent time points; Based on the byproduct matrix and the first eigenvector, multiple time-temperature characteristic sequences are obtained; Based on the byproduct matrix, the time-temperature feature sequence and the first feature vector, a predicted stop time point and a predicted temperature set are obtained; the predicted temperature set represents the predicted temperatures corresponding to multiple time points between the current time point and the stop time point; the predicted temperature represents the predicted temperature for preparing glycerophosphatidylcholine at a future time point.

2. The automated preparation method of glycerophosphatidylcholine according to claim 1, characterized in that: The step of obtaining a predicted stop time point and a predicted temperature set based on the byproduct matrix, the time-temperature characteristic sequence and the first characteristic vector includes: Acquire a first time point and a second time point; the first time point and the second time point are adjacent in the byproduct matrix; the first time point is earlier than the second time point; The increment of the byproduct volume between the first time point and the second time point is taken as the first byproduct increment; and a plurality of first byproduct increments are obtained corresponding to a plurality of time points; Arrange the temperatures corresponding to the first time point in ascending order according to the increment of the first byproduct to obtain a first temperature sequence; reorganizing the values ​​in the first eigenvector into a second eigenvector according to a first temperature sequence; Arrange the multiple time-temperature characteristic sequences from small to large according to the time-temperature change value to obtain a characteristic sequence set; Based on the first feature vector, the second feature vector and the feature sequence set, a predicted stop time point and a predicted temperature set are obtained.

3. The automated preparation method of glycerophosphatidylcholine according to claim 1, characterized in that: The method of obtaining a first eigenvector based on the byproduct matrix through a three-dimensional detection network includes: Inputting the byproduct matrix into a three-dimensional detection network, extracting changes in products and byproducts at different temperatures at adjacent time points, and obtaining a first eigenvector; The convolution kernel of the three-dimensional detection network is a three-dimensional convolution kernel of n*m*2; 2 corresponds to the time point; n represents the temperature; m represents the number of by-products; Where n is the number of rows of the byproduct matrix, and m is the number of columns of the byproduct matrix; According to the time points from early to late, with a step size of 1, the three-dimensional convolution kernel is convolved on the byproduct matrix.

4. The automated preparation method of glycerophosphatidylcholine according to claim 1, characterized in that: Based on the byproduct matrix and the first eigenvector, a plurality of time-temperature characteristic sequences are obtained, including: Based on the byproduct matrix, a plurality of time-temperature change values ​​are obtained; the time-temperature change values ​​represent temperature changes; The features with the same time-temperature change value in the first feature vector are retained to obtain multiple time-temperature features; the time-temperature features represent the features of the by-product volume change when the temperature changes corresponding to two adjacent time points are the same; According to the time points from early to late, multiple time-temperature features are sorted to obtain a time-temperature feature sequence; Multiple time-temperature characteristic sequences are obtained corresponding to multiple time-temperature change values.

5. The automated preparation method of glycerophosphatidylcholine according to claim 2, characterized in that: The step of obtaining a predicted stop time point and a predicted temperature set based on the first feature vector, the second feature vector and the feature pair sequence set includes: Inputting the first feature vector into a first neural network, detecting the effect of time on the by-product, and obtaining a first change feature; Inputting the second feature vector into a second neural network, detecting the effect of temperature increase on the by-product, and obtaining a second change feature; Inputting the feature pair sequence set into a third neural network, detecting the effect of the increase in the interval length of the temperature on the by-product, and obtaining a third change feature; Through the first prediction network and the second prediction network, based on the first change characteristic, the second change characteristic and the third change characteristic, a predicted stop time point and a predicted temperature set are obtained.

6. The automated preparation method of glycerophosphatidylcholine according to claim 4, characterized in that: Based on the byproduct matrix, a plurality of time-temperature variation values ​​are obtained, including: According to the time points from early to late, a temperature curve is constructed with the temperatures corresponding to multiple time points of the byproduct matrix; the temperature curve represents the temperature at the historical time points; The temperature curve is derivated at a time point to obtain a plurality of time-temperature variation values.

7. The automated preparation method of glycerophosphatidylcholine according to claim 5, characterized in that: The step of obtaining a predicted stop time point and a predicted temperature set based on the first change feature, the second change feature, and the third change feature through the first prediction network and the second prediction network includes: Inputting the first change feature, the second change feature and the third change feature into a first prediction network to obtain a predicted stop time point; Subtract the current time point from the predicted stop time point to get the predicted time length; Inputting the first change characteristic, the second change characteristic, the third change characteristic, and the predicted time length into a second prediction network to obtain a plurality of predicted temperatures; According to time points from early to late, the multiple predicted temperatures are combined to form a predicted temperature set.

8. The automated preparation method of glycerophosphatidylcholine according to claim 1, characterized in that: The step of obtaining the temperature, byproducts and corresponding byproduct volumes at multiple time points and constructing a byproduct matrix includes: The time points are taken as rows of the byproduct matrix; Add temperature as a column of the byproduct matrix; The byproducts are high as byproduct matrix; Among them, different heights in the byproduct matrix represent different byproducts; Find the corresponding time point, temperature and position of the by-product in the by-product matrix, and fill in the corresponding by-product volume.

9. An automated preparation system for glycerol phosphatidylcholine, characterized in that: include: An acquisition module is used to obtain the temperature, by-products and corresponding by-product volumes at multiple time points to construct a by-product matrix; A three-dimensional detection module, used to obtain a first eigenvector based on the byproduct matrix through a three-dimensional detection network; The first eigenvector is used to determine the volume changes of products and by-products at different temperatures at adjacent time points; A temperature change detection module, used to obtain a plurality of time-temperature characteristic sequences based on the byproduct matrix and the first eigenvector; A prediction module is used to obtain a predicted stop time point and a predicted temperature set based on the byproduct matrix, the time-temperature feature sequence and the first feature vector; the predicted temperature set represents the predicted temperatures corresponding to multiple time points between the current time point and the stop time point; the predicted temperature represents the predicted temperature for preparing glycerophosphatidylcholine at a future time point.

10. The automated preparation system of glycerophosphatidylcholine according to claim 9, characterized in that: The step of obtaining a predicted stop time point and a predicted temperature set based on the byproduct matrix, the time-temperature characteristic sequence and the first characteristic vector includes: Acquire a first time point and a second time point; the first time point and the second time point are adjacent in the byproduct matrix; the first time point is earlier than the second time point; The increment of the byproduct volume between the first time point and the second time point is taken as the first byproduct increment; and a plurality of first byproduct increments are obtained corresponding to a plurality of time points; Arrange the temperatures corresponding to the first time point in ascending order according to the increment of the first byproduct to obtain a first temperature sequence; reorganizing the values ​​in the first eigenvector into a second eigenvector according to a first temperature sequence; Arrange the multiple time-temperature characteristic sequences from small to large according to the time-temperature change value to obtain a characteristic sequence set; Based on the first feature vector, the second feature vector and the feature sequence set, a predicted stop time point and a predicted temperature set are obtained.

Images