CN115998752A

CN115998752A - Preparation method of vitamin D preparation containing linoleic acid vegetable oil

Info

Publication number: CN115998752A
Application number: CN202310058966.XA
Authority: CN
Inventors: 张现涛; 褚海彬; 蔡意悦
Original assignee: Nanjing Haijing Pharmaceutical Co ltd
Current assignee: Nanjing Haijing Pharmaceutical Co ltd
Priority date: 2023-01-18
Filing date: 2023-01-18
Publication date: 2023-04-25
Anticipated expiration: 2043-01-18
Also published as: CN115998752B

Abstract

The invention discloses a preparation method of a vitamin D preparation containing linoleic acid vegetable oil, which comprises the steps of representing a thermal distribution state of a stirring material by a thermal infrared image of the stirring material, enhancing position information in the thermal distribution state by a spatial attention mechanism, excavating change characteristics of the thermal distribution state in a time sequence dimension by using a convolution neural network model of a three-dimensional convolution kernel, capturing dynamic change characteristics of a stirring speed value under different time period spans, calculating response estimation between the dynamic change characteristics and the dynamic change characteristics to establish hidden association between the stirring speed and the spatial thermal distribution state characteristics of the stirring material, and finally obtaining a classification result that the stirring speed value at the current time point should be increased or decreased by a classifier. In this way, the thermal distribution characteristics of the stirred material are represented based on the thermal infrared image thereof, and the stirring speed value is adaptively adjusted based on the variation pattern of the thermal distribution characteristics.

Description

Preparation method of vitamin D preparation containing linoleic acid vegetable oil

Technical Field

The invention relates to the field of vitamin D preparation, and in particular relates to a preparation method of a vitamin D preparation containing linoleic acid vegetable oil.

Background

Vegetable oils include olive oil, sunflower oil, corn oil, soybean oil, peanut oil, etc., and various vegetable oils contain linoleic acid. Linoleic acid is used as essential fatty acid, and is synthesized with phospholipid in human body, is a constituent of tissue cells, and is particularly important for cytoplasmic membrane and mitochondrial structure; has cholesterol reducing and atherosclerosis preventing effects. If the infant lacks linoleic acid, the skin permeability to water increases, resulting in water metabolism disorders, skin diseases, hematuria, growth retardation, etc. The linoleic acid is properly supplemented, so that the tea has great benefits on the growth and intelligence development of children and skeleton growth, and has the effects of strengthening brain and improving intelligence.

Vitamin D preparation belongs to one kind of enzyme preparation, and vitamin A and vitamin D are essential matters for human growth and development, and especially have important functions on fetal and infant development, integrity of epithelial tissue, constancy of vision, genital organ, blood calcium and phosphorus, bone and tooth growth and development, etc. The proportion of linoleic acid contained in olive oil and vitamin in vitamin D preparation is balanced, so that the olive oil is favorable for human body absorption and utilization, and the digestibility in human body is 96.5%.

Chinese patent application number CN202011300944 discloses a vitamin D preparation containing linoleic acid vegetable oil, which discloses a preparation method comprising: under the heating condition, the active ingredients of the vitamin D preparation are added into the vegetable oil rich in linoleic acid, and the vegetable oil rich in linoleic acid and the active ingredients are heated together by heating and stirring, and the materials can be continuously mixed by external force in the heating process, so that the materials are uniformly dispersed and heated uniformly. And (5) passing through a colloid membrane when the temperature is reduced to 25 ℃ to prepare the capsule content.

However, in the actual preparation process of vitamin D preparations of linoleic acid vegetable oil by adopting the method, the yield of the capsule content obtained by each preparation is lower. This is because when stirring the materials by external force during heating, constant external force is usually applied to continuously stir, and the synergy between temperature and stirring speed is not concerned in such a treatment mode, and when the temperature rises, conditions such as uneven material dispersion and uneven heating may exist due to slower stirring speed, so that the yield of the prepared capsule contents is lower.

Thus, an optimized process for preparing vitamin D formulations containing linoleic acid vegetable oils is desired.

Disclosure of Invention

The present application has been made in order to solve the above technical problems. The embodiment of the application provides a preparation method of a vitamin D preparation containing linoleic acid vegetable oil, which is characterized in that a thermal infrared image of a stirred material is used for representing a thermal distribution state of the stirred material, a spatial attention mechanism is used for enhancing position information in the thermal distribution state, a convolution neural network model of a three-dimensional convolution kernel is used for mining change characteristics of the thermal distribution state in a time sequence dimension, dynamic change characteristics of stirring speed values under different time period spans are captured at the same time, and then response estimation between the dynamic change characteristics is calculated to establish hidden association between the stirring speed and the spatial thermal distribution state characteristics of the stirred material, and finally a classification result that the stirring speed value at a current time point is increased or reduced is obtained through a classifier. In this way, the thermal distribution characteristics of the stirred material are represented based on the thermal infrared image thereof, and the stirring speed value is adaptively adjusted based on the variation pattern of the thermal distribution characteristics.

According to one aspect of the present application, there is provided a method for preparing a vitamin D formulation comprising linoleic acid vegetable oil, comprising:

s110: acquiring stirring speed values of a plurality of preset time points in a preset time period and thermal infrared images of the stirred materials of the preset time points;

s120: respectively passing the thermal infrared images of the stirring materials at a plurality of preset time points through a first convolution neural network model using a spatial attention mechanism to obtain a plurality of temperature distribution feature matrixes;

s130: aggregating the plurality of temperature distribution feature matrixes into a three-dimensional input tensor along the channel dimension, and obtaining a temperature distribution time sequence association feature diagram through a second convolution neural network model using a three-dimensional convolution kernel;

s140: performing dimension reduction processing on the temperature distribution time sequence association characteristic diagram to obtain a temperature distribution time sequence association vector;

s150: arranging the stirring speed values of the plurality of preset time points into stirring speed input vectors according to a time dimension, and then obtaining stirring speed feature vectors through a multi-scale neighborhood feature extraction module;

s160: calculating the response estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector to obtain a classification feature matrix;

S170: optimizing the classification characteristic matrix to obtain an optimized classification characteristic matrix; and

s180: and the optimized classification characteristic matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the stirring speed value at the current time point should be increased or decreased.

In the above method for preparing vitamin D preparation containing linoleic acid vegetable oil, the steps of obtaining a plurality of temperature distribution feature matrices by respectively passing the thermal infrared images of the stirred materials at a plurality of predetermined time points through a first convolutional neural network model using a spatial attention mechanism comprise: performing depth convolution encoding on the thermal infrared image by using a convolution encoding part of the first convolution neural network model to obtain an initial convolution feature map; inputting the initial convolution feature map into a spatial attention portion of the first convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; calculating the position-by-position point multiplication of the space attention characteristic diagram and the initial convolution characteristic diagram to obtain a temperature distribution characteristic diagram; and carrying out global average pooling treatment along the channel dimension on the temperature distribution characteristic map to obtain the temperature distribution characteristic matrix.

In the above method for preparing vitamin D preparation containing linoleic acid vegetable oil, the step of aggregating the plurality of temperature distribution feature matrices into a three-dimensional input tensor along the channel dimension and obtaining a temperature distribution time sequence association feature map by using a second convolution neural network model of a three-dimensional convolution kernel comprises the following steps: input data are respectively subjected to forward transfer of layers by using the second convolution neural network model using the three-dimensional convolution kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution check to obtain a convolution characteristic diagram; carrying out mean pooling treatment based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network model is the temperature distribution time sequence correlation characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional input tensor.

In the above method for preparing vitamin D preparation containing linoleic acid vegetable oil, the step of performing dimension reduction treatment on the temperature distribution time sequence correlation characteristic map to obtain a temperature distribution time sequence correlation vector comprises the following steps: and carrying out global averaging pooling processing along the channel dimension on the temperature distribution time sequence association characteristic diagram to obtain the temperature distribution time sequence association vector.

In the above preparation method of the vitamin D preparation containing linoleic acid vegetable oil, the multi-scale neighborhood feature extraction module comprises: the system comprises a first convolution layer, a second convolution layer and a multi-scale fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer and the second convolution layer are parallel to each other, and one-dimensional convolution kernels with different scales are respectively used for the first convolution layer and the second convolution layer.

In the above preparation method of vitamin D preparation containing linoleic acid vegetable oil, the steps of arranging the stirring speed values at the plurality of preset time points into a stirring speed input vector according to a time dimension, and then obtaining a stirring speed feature vector by a multi-scale neighborhood feature extraction module comprise: performing one-dimensional convolution coding on the stirring speed input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first-scale stirring speed feature vector; wherein, the formula is:

wherein ,ais the first convolution kernelxWidth in the direction,

For the first convolution kernel parameter vector, +.>

For a local vector matrix that operates with a convolution kernel,wfor the size of the first convolution kernel,Xrepresenting the input vector of the stirring speed,

Representing the first scale stirring speed feature vector; performing one-dimensional convolution coding on the stirring speed input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second-scale stirring speed feature vector; wherein, the formula is:

wherein ,bis the second convolution kernelxWidth in the direction,

For a second convolution kernel parameter vector, +.>

For a local vector matrix that operates with a convolution kernel,mfor the size of the second convolution kernel,Xrepresenting the input vector of the stirring speed,

representing the second scale stirring speed feature vector; and cascading the first-scale stirring speed feature vector and the second-scale stirring speed feature vector by using a multi-scale fusion layer of the multi-scale neighborhood feature extraction module to obtain the stirring speed feature vector.

In the above preparation method of the vitamin D preparation containing linoleic acid vegetable oil, calculating the response estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector to obtain a classification feature matrix comprises the following steps: calculating a response estimate of the temperature distribution timing correlation vector relative to the agitation speed feature vector in the following formula to obtain a classification feature matrix; wherein, the formula is:

wherein

Representing the stirring speed feature vector, +.>

Representing the temperature distribution time sequence correlation vector, +.>

Representing the classification feature matrix,/->

Representing matrix multiplication.

In the above method for preparing vitamin D preparation containing linoleic acid vegetable oil, optimizing the classification feature matrix to obtain an optimized classification feature matrix comprises: performing matrix ordered Hilbert complete optimization on the classification feature matrix by using the following formula to obtain the optimized classification feature matrix; wherein, the formula is:

wherein

and

The classification characteristic matrix and the optimized classification characteristic matrix are respectively +.>

Representing the square of the two norms of the classification feature matrix,/->

Is an ordered feature matrix in which the respective row vectors of the classification feature matrix are arranged in order of magnitude as ordered vectors,/for>

Is the transpose of the classification feature matrix, < >>

Representing matrix multiplication +.>

Representing multiplication by location.

In the above method for preparing vitamin D preparation containing linoleic acid vegetable oil, the classifying feature matrix after optimization is passed through a classifier to obtain a classifying result, wherein the classifying result is used for indicating that the stirring speed value at the current time point should be increased or decreased, and the method comprises the following steps: expanding the optimized classification feature matrix into classification feature vectors according to row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

According to another aspect of the present application, there is provided a system for preparing a vitamin D formulation comprising a linoleic acid containing vegetable oil, comprising:

the stirring data acquisition module is used for acquiring stirring speed values at a plurality of preset time points in a preset time period and thermal infrared images of the stirring materials at the preset time points;

the temperature distribution space enhancement module is used for respectively passing the thermal infrared images of the stirring materials at a plurality of preset time points through a first convolution neural network model using a space attention mechanism to obtain a plurality of temperature distribution feature matrixes;

the time sequence correlation extraction module is used for acquiring a temperature distribution time sequence correlation characteristic diagram through a second convolution neural network model using a three-dimensional convolution kernel after the plurality of temperature distribution characteristic matrixes are aggregated into a three-dimensional input tensor along a channel dimension;

the dimension reduction module is used for carrying out dimension reduction processing on the temperature distribution time sequence association characteristic diagram to obtain a temperature distribution time sequence association vector;

the multi-scale coding module is used for arranging the stirring speed values of the plurality of preset time points into stirring speed input vectors according to the time dimension and then obtaining stirring speed feature vectors through the multi-scale neighborhood feature extraction module;

The responsiveness estimation module is used for calculating responsiveness estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector so as to obtain a classification feature matrix;

the optimizing module is used for optimizing the classification characteristic matrix to obtain an optimized classification characteristic matrix; and

and the result generation module is used for enabling the optimized classification characteristic matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the stirring speed value at the current time point should be increased or decreased.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the temperature distribution space enhancing module is further configured to: performing depth convolution encoding on the thermal infrared image by using a convolution encoding part of the first convolution neural network model to obtain an initial convolution feature map; inputting the initial convolution feature map into a spatial attention portion of the first convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; calculating the position-by-position point multiplication of the space attention characteristic diagram and the initial convolution characteristic diagram to obtain a temperature distribution characteristic diagram; and carrying out global average pooling treatment along the channel dimension on the temperature distribution characteristic map to obtain the temperature distribution characteristic matrix.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the time sequence correlation extraction module is further configured to: input data are respectively subjected to forward transfer of layers by using the second convolution neural network model using the three-dimensional convolution kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution check to obtain a convolution characteristic diagram; carrying out mean pooling treatment based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network model is the temperature distribution time sequence correlation characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional input tensor.

In the above preparation system of the vitamin D preparation containing linoleic acid vegetable oil, the dimension reduction module is configured to perform global average pooling processing along a channel dimension on the temperature distribution time sequence correlation characteristic map to obtain the temperature distribution time sequence correlation vector.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the multi-scale neighborhood feature extraction module comprises: the system comprises a first convolution layer, a second convolution layer and a multi-scale fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer and the second convolution layer are parallel to each other, and one-dimensional convolution kernels with different scales are respectively used for the first convolution layer and the second convolution layer.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the multi-scale coding module is further configured to: performing one-dimensional convolution coding on the stirring speed input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first-scale stirring speed feature vector; wherein, the formula is:

wherein ,ais the first convolution kernelxWidth in the direction,

For the first convolution kernel parameter vector, +.>

wherein ,bis the second convolution kernelxWidth in the direction,

For a second convolution kernel parameter vector, +.>

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the responsiveness estimation module is further configured to: calculating a response estimate of the temperature distribution timing correlation vector relative to the agitation speed feature vector in the following formula to obtain a classification feature matrix; wherein, the formula is:

wherein

Representing the saidStirring speed feature vector, ++>

Representing the temperature distribution time sequence correlation vector, +.>

Representing the classification feature matrix,/->

Representing matrix multiplication.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the optimizing module is further configured to: performing matrix ordered Hilbert complete optimization on the classification feature matrix by using the following formula to obtain the optimized classification feature matrix; wherein, the formula is:

wherein

and

Is the transpose of the classification feature matrix, < >>

Representing matrix multiplication +. >

Representing multiplication by location.

In the above preparation system of vitamin D preparation containing linoleic acid vegetable oil, the result generation module is further configured to: expanding the optimized classification feature matrix into classification feature vectors according to row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

Compared with the prior art, the preparation method of the vitamin D preparation containing linoleic acid vegetable oil provided by the application has the advantages that the thermal distribution state of the stirred materials is represented by thermal infrared images of the stirred materials, the position information in the thermal distribution state is enhanced by a spatial attention mechanism, the change characteristics of the thermal distribution state in a time sequence dimension are mined by using a convolution neural network model of a three-dimensional convolution kernel, meanwhile, the dynamic change characteristics of stirring speed values under different time period spans are captured, the response estimation between the dynamic change characteristics is calculated, so that hidden association between the stirring speed and the spatial thermal distribution state characteristics of the stirred materials is established, and finally, the classification result that the stirring speed value at the current time point is increased or reduced is obtained by a classifier. In this way, the thermal distribution characteristics of the stirred material are represented based on the thermal infrared image thereof, and the stirring speed value is adaptively adjusted based on the variation pattern of the thermal distribution characteristics.

Drawings

The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.

Fig. 1 is a flow chart of a method of preparing a vitamin D formulation comprising linoleic acid vegetable oil in accordance with an embodiment of the present application.

Fig. 2 is a schematic diagram of a method for preparing a vitamin D formulation comprising linoleic acid vegetable oil according to an embodiment of the present application.

Fig. 3 is a flowchart of a method for preparing a vitamin D preparation containing linoleic acid vegetable oil according to an embodiment of the present application, wherein the thermal infrared images of the stirred materials at a plurality of predetermined time points are respectively processed by a first convolutional neural network model using a spatial attention mechanism to obtain a plurality of temperature distribution feature matrices.

Fig. 4 is a block diagram of a system for preparing vitamin D formulations comprising linoleic acid plant oil according to an embodiment of the present application.

Detailed Description

Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.

Summary of the application

As described above, in the process of preparing the vitamin D preparation of linoleic acid vegetable oil by actually adopting the preparation method of the vitamin D preparation containing linoleic acid vegetable oil disclosed in chinese patent application No. CN202011300944, it was found that the yield of the capsule content prepared each time is low. This is because when stirring the materials by external force during heating, constant external force is usually applied to continuously stir, and the synergy between temperature and stirring speed is not concerned in such a treatment mode, and when the temperature rises, conditions such as uneven material dispersion and uneven heating may exist due to slower stirring speed, so that the yield of the prepared capsule contents is lower. Thus, an optimized process for preparing vitamin D formulations containing linoleic acid vegetable oils is desired.

Accordingly, considering that in the actual preparation process of the vitamin D preparation of the linoleic acid vegetable oil, when materials are stirred by external force in the heating process, the temperature and the stirring speed have cooperativity, the stirring speed value needs to be adaptively adjusted based on the real-time heat distribution condition of the stirred materials, so that the yield of the prepared capsule contents is improved. In this process, the extraction of the change characteristic of the thermal distribution state of the stirred material may be achieved by analysis of the thermal infrared image of the stirred material, that is, the thermal distribution characteristic of the stirred material is represented based on the thermal infrared image of the stirred material, and the stirring speed value is adaptively adjusted based on the change pattern of the thermal distribution characteristic. In the process, the difficulty is how to establish the mapping relation between the thermal distribution change mode of the stirred materials and the stirring speed, so that the stirring speed value can be adaptively adjusted based on the change mode of the thermal distribution characteristics accurately in real time, and the yield of the prepared capsule content is improved.

In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.

Deep learning and development of a neural network provide new solutions and schemes for mining complex mapping relations between the thermal distribution change pattern of the stirred materials and the stirring speed. Those of ordinary skill in the art will appreciate that a deep learning based deep neural network model may adjust parameters of the deep neural network model by appropriate training strategies, such as by a gradient descent back-propagation algorithm, to enable modeling of complex nonlinear correlations between things, which is obviously suitable for modeling and establishing complex mappings between the thermal profile change pattern of the stirred material and the stirring speed.

Specifically, in the technical scheme of the application, first, stirring speed values at a plurality of preset time points in a preset time period and thermal infrared images of stirring materials at the preset time points are acquired. Next, feature extraction of the thermal infrared image of the stirred material at the plurality of predetermined time points is performed using a convolutional neural network model having excellent performance in implicit feature extraction of the image. In particular, considering that due to the consistency between the temperature and the stirring speed, the materials are required to be stirred by external force in the heating process, so that the stirring speed value is adaptively adjusted based on the real-time heat distribution condition of the stirred materials, and the materials are uniformly dispersed and heated uniformly. Therefore, during feature extraction, uniformity feature information of temperature and heat distribution of the stirred materials in space position needs to be focused more, so that uniformity of material dispersion and heating is improved, and yield of the prepared capsule contents is improved. Based on this, in the technical solution of the present application, the thermal infrared images of the stirred materials at the plurality of predetermined time points are further processed in the first convolutional neural network model using the spatial attention mechanism, so as to extract thermal characteristic distribution information about the temperature characteristics of the stirred materials at the spatial positions in the thermal infrared images of the stirred materials at the respective predetermined time points, thereby obtaining a plurality of temperature distribution characteristic matrices.

Further, in order to extract the correlation characteristic between the thermal distribution change pattern of the stirred material and the stirring speed, it is also necessary to extract dynamic change characteristic distribution information of the thermal distribution characteristic of the stirred material within the predetermined period of time. Therefore, in the technical scheme of the application, after the plurality of temperature distribution feature matrixes are aggregated into a three-dimensional input tensor along the channel dimension, feature mining is performed in a second convolution neural network model by using a three-dimensional convolution kernel, so that dynamic change feature information of the spatial heat distribution state feature of the stirred material in the time dimension is extracted, and a temperature distribution time sequence association feature map is obtained. In particular, the convolution kernel of the second convolutional neural network model is a three-dimensional convolution kernel, which has W (width), H (height) and C (channel dimension), and in the technical solution of the present application, the channel dimension of the three-dimensional convolution kernel corresponds to a time dimension in which the plurality of temperature distribution feature matrices are aggregated into a three-dimensional input tensor, so that dynamic change features of spatial heat distribution state features of the stirred material along with the time dimension can be extracted when three-dimensional convolution encoding is performed. And then, performing dimension reduction processing on the temperature distribution time sequence associated feature map, and correspondingly, in a specific example of the application, performing global pooling processing on each feature matrix of the temperature distribution time sequence associated feature map along the channel dimension so as to keep dynamic feature information of the temperature distribution time sequence associated feature map in time sequence during dimension reduction, thereby obtaining a temperature distribution time sequence associated vector.

Then, considering that the stirring speed values have different mode state characteristics under different time period spans in the preset time period, in order to accurately explore dynamic change characteristics of the stirring speed values in time sequence, so as to accurately control the stirring speed values of the current time point in real time, in the technical scheme of the application, the stirring speed values of the preset time points are further arranged into stirring speed input vectors according to time dimensions and then encoded in a multi-scale neighborhood feature extraction module, so that dynamic multi-scale neighborhood associated feature distribution information of the stirring speed under different time period spans is extracted, and the stirring speed feature vectors are obtained.

And then, further calculating the response estimation of the temperature distribution time sequence correlation vector relative to the stirring speed characteristic vector to represent the hidden correlation characteristic information between the time sequence multi-scale dynamic change characteristic of the stirring speed and the dynamic change characteristic of the space heat distribution state characteristic of the stirring materials on the time sequence, and taking the hidden correlation characteristic information as a classification characteristic matrix to carry out classification processing in a classifier, thereby obtaining a classification result for representing that the stirring speed value at the current time point should be increased or decreased. That is, in the technical solution of the present application, the label of the classifier includes that the stirring speed value at the current time point should be increased or decreased, wherein the classifier determines to which classification label the classification feature matrix belongs through a soft maximum function. It should be understood that, in the technical solution of the present application, the classification label of the classifier is a control policy label of the stirring speed value, so after the classification result is obtained, the stirring speed value may be adaptively adjusted based on the classification result, so as to achieve the technical purpose of uniformity of dispersion and heat balance of the stirred materials. That is, in the technical solution of the present application, based on the classification result, a control instruction of the stirring speed value is generated, and then the stirring speed value is controlled in real time by the stirring device to adaptively adjust.

In particular, in the technical solution of the present application, the temperature distribution time sequence correlation vector is obtained by global pooling of each feature matrix along a channel dimension of the temperature distribution time sequence correlation feature map, which causes the temperature distribution time sequence correlation vector to lose a large amount of spatial structure feature information relative to the temperature distribution time sequence correlation feature map, so when the responsiveness estimation of the temperature distribution time sequence correlation vector relative to the stirring speed feature vector is calculated to obtain the classification feature matrix, the structure feature loss exists in the temperature distribution time sequence correlation vector and the feature depth difference exists between the temperature distribution time sequence correlation vector and the stirring speed feature vector, and therefore, the structure of the feature distribution of the classification feature matrix may be blurred, thereby reducing the expression accuracy of the classification feature matrix and affecting the accuracy of the classification result obtained by the classifier of the classification feature matrix.

Based on this, in the technical solution of the present application, the matrix-ordered hilbert completion is performed on the classification feature matrix, which is expressed as:

and

The classification characteristic matrix and the optimized classification characteristic matrix are respectively +. >

Representing the square of the two norms of the classification characteristic matrix, < >>

Is an ordered feature matrix in which respective row vectors of the classified feature matrix are arranged in order of magnitude as ordered vectors,/v>

Is a transpose of the classification feature matrix.

Here, by mapping the ordered matrix into the hilbert space defined by the self-inner product of the matrix, a meaningful measurement of the numerical relation of the feature set in the consistency space can be realized, based on the meaningful measurement, the feature space with an orthorhombic structure is built by embedding the relative position of the ordered matrix and the feature matrix, and the structure in the feature space is completed for the high-dimensional manifold of the feature matrix, so that the reduction of the expression certainty of the feature matrix due to the fuzzification structure can be avoided, and the accuracy of the classification result obtained by the classifier of the classification feature matrix is improved. Therefore, the stirring speed value can be adaptively adjusted in real time and accurately based on the change mode of the heat distribution characteristics, so that the dispersion uniformity and the heating balance of materials during stirring are realized, and the yield of the prepared capsule content is improved.

Based on this, the application provides a preparation method of a vitamin D preparation containing linoleic acid vegetable oil, which comprises the following steps: acquiring stirring speed values of a plurality of preset time points in a preset time period and thermal infrared images of the stirred materials of the preset time points; respectively passing the thermal infrared images of the stirring materials at a plurality of preset time points through a first convolution neural network model using a spatial attention mechanism to obtain a plurality of temperature distribution feature matrixes; aggregating the plurality of temperature distribution feature matrixes into a three-dimensional input tensor along the channel dimension, and obtaining a temperature distribution time sequence association feature diagram through a second convolution neural network model using a three-dimensional convolution kernel; performing dimension reduction processing on the temperature distribution time sequence association characteristic diagram to obtain a temperature distribution time sequence association vector; arranging the stirring speed values of the plurality of preset time points into stirring speed input vectors according to a time dimension, and then obtaining stirring speed feature vectors through a multi-scale neighborhood feature extraction module; calculating the response estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector to obtain a classification feature matrix; optimizing the classification characteristic matrix to obtain an optimized classification characteristic matrix; and passing the optimized classification feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the stirring speed value at the current time point should be increased or decreased.

Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.

Exemplary method

Fig. 1 is a flow chart of a method of preparing a vitamin D formulation comprising linoleic acid vegetable oil in accordance with an embodiment of the present application. As shown in fig. 1, a method for preparing a vitamin D preparation containing linoleic acid vegetable oil according to an embodiment of the present application includes: s110, acquiring stirring speed values of a plurality of preset time points in a preset time period and thermal infrared images of the stirring materials of the preset time points; s120, respectively obtaining a plurality of temperature distribution feature matrixes by using a first convolution neural network model of a spatial attention mechanism according to the thermal infrared images of the stirring materials at a plurality of preset time points; s130, aggregating the plurality of temperature distribution feature matrixes into a three-dimensional input tensor along the channel dimension, and obtaining a temperature distribution time sequence association feature diagram through a second convolution neural network model using a three-dimensional convolution kernel; s140, performing dimension reduction processing on the temperature distribution time sequence association characteristic diagram to obtain a temperature distribution time sequence association vector; s150, arranging the stirring speed values of the plurality of preset time points into stirring speed input vectors according to a time dimension, and then obtaining stirring speed feature vectors through a multi-scale neighborhood feature extraction module; s160, calculating the response estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector to obtain a classification feature matrix; s170, optimizing the classification characteristic matrix to obtain an optimized classification characteristic matrix; and S180, the optimized classification feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the stirring speed value at the current time point should be increased or decreased.

Fig. 2 is a schematic diagram of a method for preparing a vitamin D formulation comprising linoleic acid vegetable oil according to an embodiment of the present application. In this architecture, as shown in fig. 2, first, stirring speed values at a plurality of predetermined time points within a predetermined period of time and thermal infrared images of the stirred material at the plurality of predetermined time points are acquired; then, respectively obtaining a plurality of temperature distribution feature matrixes by using a first convolution neural network model of a spatial attention mechanism through thermal infrared images of the stirring materials at a plurality of preset time points; then, the temperature distribution feature matrixes are aggregated into a three-dimensional input tensor along the channel dimension, and a temperature distribution time sequence association feature diagram is obtained through a second convolution neural network model using a three-dimensional convolution kernel; then, performing dimension reduction processing on the temperature distribution time sequence association feature map to obtain a temperature distribution time sequence association vector, and simultaneously, arranging stirring speed values of the plurality of preset time points into a stirring speed input vector according to a time dimension and then obtaining a stirring speed feature vector through a multi-scale neighborhood feature extraction module; then, calculating the response estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector to obtain a classification feature matrix; then, optimizing the classification characteristic matrix to obtain an optimized classification characteristic matrix; and finally, the optimized classification feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the stirring speed value at the current time point should be increased or decreased.

In step S110, stirring speed values at a plurality of predetermined time points within a predetermined period of time and thermal infrared images of the stirred material at the plurality of predetermined time points are acquired. Wherein the stirring speed value and the thermal infrared image of the stirring material can be obtained by a velocimeter and a thermal infrared sensor respectively. That is, the stirring speed values at a plurality of preset time points in the preset time period and the thermal infrared images of the stirring materials at the preset time points are used as the input of the deep neural network model.

In step S120, the thermal infrared images of the stirred materials at the predetermined time points are respectively passed through a first convolutional neural network model using a spatial attention mechanism to obtain a plurality of temperature distribution feature matrices. That is, feature extraction of the thermal infrared image of the stirred material at the plurality of predetermined time points is performed using a convolutional neural network model having excellent performance in implicit feature extraction of the image. In particular, considering that due to the consistency between the temperature and the stirring speed, the materials are required to be stirred by external force in the heating process, so that the stirring speed value is adaptively adjusted based on the real-time heat distribution condition of the stirred materials, and the materials are uniformly dispersed and heated uniformly. Therefore, during feature extraction, uniformity feature information of temperature and heat distribution of the stirred materials in space position needs to be focused more, so that uniformity of material dispersion and heating is improved, and yield of the prepared capsule contents is improved. Based on this, in the technical solution of the present application, the thermal infrared images of the stirred materials at the plurality of predetermined time points are further processed in the first convolutional neural network model using the spatial attention mechanism, so as to extract thermal characteristic distribution information about the temperature characteristics of the stirred materials at the spatial positions in the thermal infrared images of the stirred materials at the respective predetermined time points, thereby obtaining a plurality of temperature distribution characteristic matrices.

Fig. 3 is a flowchart of a method for preparing a vitamin D preparation containing linoleic acid vegetable oil according to an embodiment of the present application, wherein the thermal infrared images of the stirred materials at a plurality of predetermined time points are respectively processed by a first convolutional neural network model using a spatial attention mechanism to obtain a plurality of temperature distribution feature matrices. As shown in fig. 3, the steps of obtaining a plurality of temperature distribution feature matrices by using a first convolution neural network model of a spatial attention mechanism through thermal infrared images of the stirred materials at a plurality of preset time points respectively include: s210, performing depth convolution coding on the thermal infrared image by using a convolution coding part of the first convolution neural network model to obtain an initial convolution characteristic diagram; s220, inputting the initial convolution feature map into a spatial attention portion of the first convolution neural network model to obtain a spatial attention map; s230, the spatial attention is subjected to a Softmax activation function to obtain a spatial attention profile; s240, calculating the position-by-position point multiplication of the spatial attention characteristic map and the initial convolution characteristic map to obtain a temperature distribution characteristic map; and S250, carrying out global averaging pool processing on the temperature distribution characteristic diagram along the channel dimension to obtain the temperature distribution characteristic matrix.

In step S130, the plurality of temperature distribution feature matrices are aggregated into a three-dimensional input tensor along the channel dimension, and then a temperature distribution time sequence correlation feature map is obtained by using a second convolution neural network model of the three-dimensional convolution kernel. In order to mine the correlation characteristic between the heat distribution change mode of the stirred material and the stirring speed, dynamic change characteristic distribution information of the heat distribution characteristic of the stirred material in the preset time period is also required to be mined. Therefore, in the technical scheme of the application, after the plurality of temperature distribution feature matrixes are aggregated into a three-dimensional input tensor along the channel dimension, feature mining is performed in a second convolution neural network model by using a three-dimensional convolution kernel, so that dynamic change feature information of the spatial heat distribution state feature of the stirred material in the time dimension is extracted, and a temperature distribution time sequence association feature map is obtained. In particular, the convolution kernel of the second convolutional neural network model is a three-dimensional convolution kernel, which has W (width), H (height) and C (channel dimension), and in the technical solution of the present application, the channel dimension of the three-dimensional convolution kernel corresponds to a time dimension in which the plurality of temperature distribution feature matrices are aggregated into a three-dimensional input tensor, so that dynamic change features of spatial heat distribution state features of the stirred material along with the time dimension can be extracted when three-dimensional convolution encoding is performed.

In the embodiment of the application, the input data are respectively performed in forward transfer of the layers by using the second convolution neural network model using the three-dimensional convolution kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution check to obtain a convolution characteristic diagram; carrying out mean pooling treatment based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network model is the temperature distribution time sequence correlation characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional input tensor.

In step S140, the temperature distribution time sequence correlation characteristic map is subjected to a dimension reduction process to obtain a temperature distribution time sequence correlation vector. Accordingly, in a specific example of the present application, global pooling processing may be performed on each feature matrix of the temperature distribution time sequence association feature map along the channel dimension, so as to retain dynamic feature information of the temperature distribution time sequence association feature map on time sequence during dimension reduction, thereby obtaining a temperature distribution time sequence association vector.

In step S150, the stirring speed values at the predetermined time points are arranged according to the time dimension to form a stirring speed input vector, and then the stirring speed input vector is passed through a multi-scale neighborhood feature extraction module to obtain a stirring speed feature vector. In the technical scheme of the application, stirring speed values at a plurality of preset time points are further arranged into stirring speed input vectors according to time dimensions and then encoded in a multi-scale neighborhood feature extraction module, so that dynamic multi-scale neighborhood associated feature distribution information of the stirring speed in different time period spans is extracted, and the stirring speed feature vectors are obtained. The multi-scale neighborhood feature extraction module comprises a first convolution layer and a second convolution layer which are parallel to each other, and a multi-scale fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer and the second convolution layer respectively use one-dimensional convolution kernels with different scales.

Specifically, in the embodiment of the present application, the arranging the stirring speed values at the plurality of predetermined time points according to the time dimension into the stirring speed input vector, and then passing through a multi-scale neighborhood feature extraction module to obtain a stirring speed feature vector includes: performing one-dimensional convolution coding on the stirring speed input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first-scale stirring speed feature vector; wherein, the formula is:

wherein ,ais the first convolution kernelxWidth in the direction,

For the first convolution kernel parameter vector, +.>

wherein ,bis the second convolution kernelxWidth in directionA degree of,

For a second convolution kernel parameter vector, +.>

For a local vector matrix that operates with a convolution kernel, mFor the size of the second convolution kernel,Xrepresenting the input vector of the stirring speed,

In step S160, a responsiveness estimate of the temperature distribution timing correlation vector with respect to the stirring speed feature vector is calculated to obtain a classification feature matrix. Namely, calculating the response estimation of the temperature distribution time sequence correlation vector relative to the stirring speed characteristic vector to represent the hidden correlation characteristic information between the time sequence multi-scale dynamic change characteristic of the stirring speed and the dynamic change characteristic of the space heat distribution state characteristic of the stirring material on the time sequence, and taking the hidden correlation characteristic information as a classification characteristic matrix.

Specifically, in the embodiment of the application, the responsiveness estimation of the temperature distribution time sequence association vector relative to the stirring speed feature vector is calculated by the following formula to obtain a classification feature matrix; wherein, the formula is:

wherein

Representing the stirring speed feature vector, +.>

Representing the temperature distribution time sequence correlation vector, +.>

Representing the classification feature matrix,/->

Representing matrix multiplication.

In step S170, the classification feature matrix is optimized to obtain an optimized classification feature matrix. In particular, in the technical solution of the present application, the temperature distribution time sequence correlation vector is obtained by global pooling of each feature matrix along a channel dimension of the temperature distribution time sequence correlation feature map, which causes the temperature distribution time sequence correlation vector to lose a large amount of spatial structure feature information relative to the temperature distribution time sequence correlation feature map, so when the responsiveness estimation of the temperature distribution time sequence correlation vector relative to the stirring speed feature vector is calculated to obtain the classification feature matrix, the structure feature loss exists in the temperature distribution time sequence correlation vector and the feature depth difference exists between the temperature distribution time sequence correlation vector and the stirring speed feature vector, and therefore, the structure of the feature distribution of the classification feature matrix may be blurred, thereby reducing the expression accuracy of the classification feature matrix and affecting the accuracy of the classification result obtained by the classifier of the classification feature matrix.

wherein

and

Is the transpose of the classification feature matrix, < >>

Representing matrix multiplication +.>

Representing multiplication by location.

In step S180, the optimized classification feature matrix is passed through a classifier to obtain a classification result, where the classification result is used to indicate that the stirring speed value at the current time point should be increased or decreased. That is, in the technical solution of the present application, the label of the classifier includes that the stirring speed value at the current time point should be increased or decreased, wherein the classifier determines to which classification label the classification feature matrix belongs through a soft maximum function. It should be understood that, in the technical solution of the present application, the classification label of the classifier is a control policy label of the stirring speed value, so after the classification result is obtained, the stirring speed value may be adaptively adjusted based on the classification result, so as to achieve the technical purpose of uniformity of dispersion and heat balance of the stirred materials. That is, in the technical solution of the present application, based on the classification result, a control instruction of the stirring speed value is generated, and then the stirring speed value is controlled in real time by the stirring device to adaptively adjust.

Specifically, in the embodiment of the present application, the optimized classification feature matrix is first expanded into classification feature vectors according to row vectors or column vectors; then, using a full-connection layer of the classifier to carry out full-connection coding on the classification feature vector so as to obtain a coded classification feature vector; and then, inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

In summary, a preparation method of a vitamin D preparation containing linoleic acid vegetable oil according to the embodiments of the present application is illustrated, which characterizes a thermal distribution state of a stirred material with a thermal infrared image of the stirred material, enhances position information in the thermal distribution state with a spatial attention mechanism, uses a convolutional neural network model of a three-dimensional convolutional kernel to mine a change feature of the thermal distribution state in a time sequence dimension, captures dynamic change features of a stirring speed value under different time period spans, calculates a response estimate between the two features to establish a hidden association between the stirring speed and the spatial thermal distribution state feature of the stirred material, and finally obtains a classification result that the stirring speed value at a current time point should be increased or decreased through a classifier. In this way, the thermal distribution characteristics of the stirred material are represented based on the thermal infrared image thereof, and the stirring speed value is adaptively adjusted based on the variation pattern of the thermal distribution characteristics.

Exemplary System

Fig. 4 is a block diagram of a system for preparing vitamin D formulations comprising linoleic acid plant oil according to an embodiment of the present application. As shown in fig. 4, a system 100 for preparing a vitamin D formulation comprising linoleic acid vegetable oil according to an embodiment of the present application comprises: a stirring data obtaining module 110, configured to obtain stirring speed values at a plurality of predetermined time points in a predetermined time period and thermal infrared images of a stirred material at the plurality of predetermined time points; the temperature distribution space enhancement module 120 is configured to obtain a plurality of temperature distribution feature matrices by respectively passing the thermal infrared images of the stirred materials at the plurality of predetermined time points through a first convolutional neural network model using a spatial attention mechanism; the time sequence correlation extraction module 130 is configured to aggregate the plurality of temperature distribution feature matrices into a three-dimensional input tensor along a channel dimension, and obtain a temperature distribution time sequence correlation feature map by using a second convolution neural network model of a three-dimensional convolution kernel; the dimension reduction module 140 is configured to perform dimension reduction processing on the temperature distribution time sequence association feature map to obtain a temperature distribution time sequence association vector; the multi-scale encoding module 150 is configured to arrange the stirring speed values at the plurality of predetermined time points into a stirring speed input vector according to a time dimension, and then obtain a stirring speed feature vector through the multi-scale neighborhood feature extraction module; a responsiveness estimation module 160, configured to calculate a responsiveness estimate of the temperature distribution timing correlation vector relative to the stirring speed feature vector to obtain a classification feature matrix; the optimizing module 170 is configured to optimize the classification feature matrix to obtain an optimized classification feature matrix; and a result generating module 180, configured to pass the optimized classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the stirring speed value at the current time point should be increased or decreased.

In one example, in the above-mentioned vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the temperature distribution space enhancement module 120 is further configured to: performing depth convolution encoding on the thermal infrared image by using a convolution encoding part of the first convolution neural network model to obtain an initial convolution feature map; inputting the initial convolution feature map into a spatial attention portion of the first convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; calculating the position-by-position point multiplication of the space attention characteristic diagram and the initial convolution characteristic diagram to obtain a temperature distribution characteristic diagram; and carrying out global average pooling treatment along the channel dimension on the temperature distribution characteristic map to obtain the temperature distribution characteristic matrix.

In one example, in the above-mentioned vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the timing-related extraction module 130 is further configured to: input data are respectively subjected to forward transfer of layers by using the second convolution neural network model using the three-dimensional convolution kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution check to obtain a convolution characteristic diagram; carrying out mean pooling treatment based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network model is the temperature distribution time sequence correlation characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional input tensor.

In one example, in the preparation system 100 of the vitamin D preparation containing linoleic acid plant oil, the dimension reduction module 140 is configured to perform global averaging along a channel dimension on the temperature distribution time-series correlation feature map to obtain the temperature distribution time-series correlation vector.

In one example, in the preparation system 100 of the vitamin D formulation containing linoleic acid vegetable oil, the multi-scale neighborhood feature extraction module comprises: the system comprises a first convolution layer, a second convolution layer and a multi-scale fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer and the second convolution layer are parallel to each other, and one-dimensional convolution kernels with different scales are respectively used for the first convolution layer and the second convolution layer.

In one example, in the above-described vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the multi-scale encoding module 150 is further configured to: performing one-dimensional convolution coding on the stirring speed input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first-scale stirring speed feature vector; wherein, the formula is:

wherein ,ais the first convolution kernel xWidth in the direction,

For the first convolution kernel parameter vector, +.>

wherein ,bis the second convolution kernelxWidth in the direction,

For a second convolution kernel parameter vector, +.>

representing the second scale stirring speed feature vector; and, using the multi-scale neighborhood feature extractionAnd the multi-scale fusion layer of the module cascades the first-scale stirring speed characteristic vector and the second-scale stirring speed characteristic vector to obtain the stirring speed characteristic vector.

In one example, in the above-described vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the responsiveness estimation module 160 is further configured to: calculating a response estimate of the temperature distribution timing correlation vector relative to the agitation speed feature vector in the following formula to obtain a classification feature matrix; wherein, the formula is:

wherein

Representing the stirring speed feature vector, +.>

Representing the temperature distribution time sequence correlation vector, +.>

Representing the classification feature matrix,/->

Representing matrix multiplication.

In one example, in the above-described vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the optimization module 170 is further configured to: performing matrix ordered Hilbert complete optimization on the classification feature matrix by using the following formula to obtain the optimized classification feature matrix; wherein, the formula is:

wherein

and

Is the transpose of the classification feature matrix, < >>

Representing matrix multiplication +.>

Representing multiplication by location.

In one example, in the above-described vitamin D formulation preparation system 100 containing linoleic acid vegetable oil, the result generation module 180 is further configured to: expanding the optimized classification feature matrix into classification feature vectors according to row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described preparation system 100 for a vitamin D preparation containing linoleic acid vegetable oil have been described in detail in the above description of the preparation method for a vitamin D preparation containing linoleic acid vegetable oil with reference to fig. 1 to 3, and thus, repetitive descriptions thereof will be omitted.

As described above, the vitamin D preparation system 100 including the linoleic acid-containing plant oil according to the embodiment of the present application may be implemented in various terminal devices, for example, a server or the like for the preparation of the vitamin D preparation including the linoleic acid-containing plant oil. In one example, the system 100 for preparing vitamin D formulations of linoleic acid containing vegetable oils according to embodiments of the present application may be integrated into the terminal device as a software module and/or hardware module. For example, the vitamin D formulation preparation system 100 containing linoleic acid plant oil may be a software module in the operating system of the terminal device or may be an application developed for the terminal device; of course, the system 100 for preparing the vitamin D formulation containing the linoleic acid plant oil can also be one of a plurality of hardware modules of the terminal device.

Alternatively, in another example, the preparation system 100 of the vitamin D formulation including linoleic acid vegetable oil and the terminal device may be separate devices, and the preparation system 100 of the vitamin D formulation including linoleic acid vegetable oil may be connected to the terminal device through a wired and/or wireless network and transmit interactive information according to a contracted data format.

The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.

The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.

It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to the present application.

The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.

Claims

1. A method for preparing a vitamin D preparation containing linoleic acid vegetable oil, which is characterized by comprising the following steps:

2. The method for preparing a vitamin D formulation containing linoleic acid plant oil according to claim 1, wherein the preparing the plurality of temperature distribution feature matrices by using the first convolutional neural network model of the spatial attention mechanism with the thermal infrared images of the stirred materials at the predetermined time points comprises:

s210: performing depth convolution encoding on the thermal infrared image by using a convolution encoding part of the first convolution neural network model to obtain an initial convolution feature map;

s220: inputting the initial convolution feature map into a spatial attention portion of the first convolution neural network model to obtain a spatial attention map;

s230: -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile;

s240: calculating the position-by-position point multiplication of the space attention characteristic diagram and the initial convolution characteristic diagram to obtain a temperature distribution characteristic diagram; and

s250: and carrying out global average pooling treatment along the channel dimension on the temperature distribution characteristic map to obtain the temperature distribution characteristic matrix.

3. The method for preparing a vitamin D formulation containing linoleic acid vegetable oil according to claim 2, wherein after aggregating the plurality of temperature distribution feature matrices into a three-dimensional input tensor along a channel dimension, obtaining a temperature distribution time sequence correlation feature map by using a second convolution neural network model of a three-dimensional convolution kernel, comprising:

input data are respectively subjected to forward transfer of layers by using the second convolution neural network model using the three-dimensional convolution kernel:

performing three-dimensional convolution processing on the input data based on the three-dimensional convolution check to obtain a convolution characteristic diagram;

carrying out mean pooling treatment based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and

non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map;

the output of the last layer of the second convolutional neural network model is the temperature distribution time sequence correlation characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional input tensor.

4. The method for preparing a vitamin D preparation containing linoleic acid plant oil according to claim 3, wherein the step of performing a dimension reduction treatment on the temperature distribution time sequence correlation characteristic map to obtain a temperature distribution time sequence correlation vector comprises the steps of: and carrying out global averaging pooling processing along the channel dimension on the temperature distribution time sequence association characteristic diagram to obtain the temperature distribution time sequence association vector.

5. The method for preparing a vitamin D formulation containing linoleic acid plant oil according to claim 4, wherein the multi-scale neighborhood feature extraction module comprises: the system comprises a first convolution layer, a second convolution layer and a multi-scale fusion layer connected with the first convolution layer and the second convolution layer, wherein the first convolution layer and the second convolution layer are parallel to each other, and one-dimensional convolution kernels with different scales are respectively used for the first convolution layer and the second convolution layer.

6. The method for preparing a vitamin D preparation containing linoleic acid plant oil according to claim 5, wherein the step of obtaining the stirring speed feature vector by the multi-scale neighborhood feature extraction module after arranging the stirring speed values of the plurality of predetermined time points into the stirring speed input vector according to the time dimension comprises the steps of:

performing one-dimensional convolution coding on the stirring speed input vector by using a first convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a first-scale stirring speed feature vector;

wherein, the formula is:

wherein ,ais the first convolution kernelxWidth in the direction,

For the first convolution kernel parameter vector, +.>

For a local vector matrix that operates with a convolution kernel, wFor the size of the first convolution kernel,Xrepresenting the input vector of the stirring speed,

representing the first scale stirring speed feature vector;

performing one-dimensional convolution coding on the stirring speed input vector by using a second convolution layer of the multi-scale neighborhood feature extraction module according to the following formula to obtain a second-scale stirring speed feature vector;

wherein, the formula is:

wherein ,bis the second convolution kernelxWidth in the direction,

For a second convolution kernel parameter vector, +.>

representing the second scale stirring speed feature vector;

and cascading the first-scale stirring speed feature vector and the second-scale stirring speed feature vector by using a multi-scale fusion layer of the multi-scale neighborhood feature extraction module to obtain the stirring speed feature vector.

7. The method of claim 6, wherein calculating a response estimate of the temperature profile timing correlation vector relative to the agitation speed profile vector to obtain a classification profile matrix comprises:

Calculating a response estimate of the temperature distribution timing correlation vector relative to the agitation speed feature vector in the following formula to obtain a classification feature matrix;

wherein, the formula is:

； wherein

Representing the stirring speed feature vector, +.>

Representation ofThe temperature distribution time sequence association vector, +.>

Representing the classification feature matrix,/->

Representing matrix multiplication.

8. The method of claim 7, wherein optimizing the classification matrix to obtain an optimized classification matrix comprises:

performing matrix ordered Hilbert complete optimization on the classification feature matrix by using the following formula to obtain the optimized classification feature matrix;

wherein, the formula is:

wherein

and

Is the transposed moment of the classification feature matrixArray (S)>

Representing matrix multiplication +.>

Representing multiplication by location.

9. The method for preparing a vitamin D formulation containing linoleic acid plant oil according to claim 8, wherein the classifying feature matrix after optimization is passed through a classifier to obtain a classification result, and the classification result is used for indicating that the stirring speed value at the current time point should be increased or decreased, and comprises:

expanding the optimized classification feature matrix into classification feature vectors according to row vectors or column vectors;

performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and

and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.