CN115960144A - Intelligent production system and method of isomaltooligosaccharide iron complex - Google Patents

Intelligent production system and method of isomaltooligosaccharide iron complex Download PDF

Info

Publication number
CN115960144A
CN115960144A CN202310009378.7A CN202310009378A CN115960144A CN 115960144 A CN115960144 A CN 115960144A CN 202310009378 A CN202310009378 A CN 202310009378A CN 115960144 A CN115960144 A CN 115960144A
Authority
CN
China
Prior art keywords
feature
reaction
reaction liquid
liquid state
temperature
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202310009378.7A
Other languages
Chinese (zh)
Inventor
魏哲
杨春兰
徐善增
谢峰
周斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Huamao Pharmaceutical Co ltd
Original Assignee
Shanghai Huamao Pharmaceutical Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Huamao Pharmaceutical Co ltd filed Critical Shanghai Huamao Pharmaceutical Co ltd
Priority to CN202310009378.7A priority Critical patent/CN115960144A/en
Publication of CN115960144A publication Critical patent/CN115960144A/en
Withdrawn legal-status Critical Current

Links

Images

Landscapes

  • Feedback Control In General (AREA)

Abstract

The application relates to the technical field of intelligent production, and particularly discloses an intelligent production system and method of an isomaltooligosaccharide iron complex.

Description

Intelligent production system and method of isomaltooligosaccharide iron complex
Technical Field
The application relates to the technical field of intelligent production, and more specifically relates to an intelligent production system and method of an isomaltooligosaccharide iron complex.
Background
Iron is one of the essential trace elements in the human body, and iron deficiency in the body can cause iron-deficiency anemia. Ferrous salts have been selected for treating iron deficiency anemia for a long time, but the ferrous salts have unstable chemical properties, poor bioavailability and obvious stimulation effect on the digestive tract, and are easy to generate endogenous free radicals in vivo to cause cell membrane damage. Therefore, ferrous salts are not ideal iron-supplementing agents.
Research shows that the iron polysaccharide complex as a novel iron supplement has proper stability, has no or little stimulation to gastrointestinal tract, and the ligand polysaccharide has various biological activities, can be absorbed and utilized by organisms after being released, and has no toxic or side effect. Therefore, polysaccharide iron as a more ideal iron supplement has been a hot research in recent years.
Isomaltooligosaccharides, also known as isomaltooligosaccharides, and branched oligosaccharides, are functional oligosaccharides, and have the characteristics of low sweetness, difficult fermentation, heat resistance, acid resistance, good moisture retention, good flavor, etc. Because the isomaltooligosaccharide has been industrially produced and has good physiological functions, the polysaccharide iron prepared by using the isomaltooligosaccharide as a raw material can simultaneously play a role in supplementing iron and a plurality of functions of the polysaccharide.
However, the current production line for the iron isomaltooligosaccharide complex does not meet the expected requirements on the preparation efficiency and the preparation cost of the iron isomaltooligosaccharide complex. The existing production line for the isomaltooligosaccharide iron complex which is not formed has relatively high preparation efficiency and relatively low preparation cost.
Therefore, an optimized intelligent production scheme of the iron isomaltooligosaccharide complex is desired.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides an intelligent production system and method of an isomaltooligosaccharide iron complex, the system and method comprises the steps of firstly obtaining reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period, then simulating and establishing a complex mapping relation between state change and reaction temperature change of a reaction liquid through a deep neural network model to obtain an optimized classification characteristic matrix, and finally decoding the optimized classification characteristic matrix to obtain a classification result which is used for representing that the reaction temperature value of the current time point is increased or decreased.
According to one aspect of the present application, there is provided an intelligent production method of an isomalto-oligosaccharide iron complex, comprising: dissolving isomaltooligosaccharide in distilled water to obtain isomaltooligosaccharide solution; adding Na into the isomaltooligosaccharide solution 2 CO 3 Solution and FeCl 3 The solution is used for obtaining reaction mixed solution; adding a NaOH solution into the reaction mixed solution to adjust the pH value of the reaction mixed solution to 12; under a preset temperature control strategy, reacting the reaction mixed solution to obtain a reacted solution; filtering the reacted solution, performing rotary evaporation concentration on the filtered filtrate, and cooling to room temperature to obtain a concentrated reacted solution; adding anhydrous ethanol into the solution after the concentrated reaction for alcohol precipitation to obtain a crude product of the isomaltooligosaccharide iron; and purifying the crude product of the isomaltooligosaccharide iron to obtain the isomaltooligosaccharide iron complex.
In the above method for intelligently producing an isomaltooligosaccharide iron complex, the reacting the reaction mixture under a predetermined temperature control strategy to obtain a reacted solution includes: obtaining reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period; extracting reaction liquid state monitoring key frames of the plurality of preset time points from the reaction liquid state monitoring video; respectively enabling the reaction liquid state monitoring key frames of the plurality of preset time points to pass through a first convolution neural network model serving as a filter to obtain a plurality of reaction liquid state feature vectors; passing the plurality of reaction liquid state feature vectors through a context encoder based on a converter to obtain reaction liquid state context semantic feature vectors; arranging the reaction temperature values of the plurality of preset time points into a temperature input vector according to a time dimension, and then obtaining a temperature characteristic vector by using a second convolution neural network model of a one-dimensional convolution kernel; calculating the responsiveness estimation of the reaction liquid state context semantic feature vector relative to the temperature feature vector based on a Gaussian density map to obtain a classification feature matrix; based on the reaction liquid state context semantic feature vector and the temperature feature vector, carrying out vector granularity feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix; and passing the optimized classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased.
In the above method for intelligently producing an isomaltooligosaccharide iron complex, the passing the plurality of reaction solution state feature vectors through a context encoder based on a converter to obtain a reaction solution state context semantic feature vector includes: passing the plurality of reaction solution state feature vectors through the converter-based context encoder to obtain a plurality of context reaction solution state feature vectors; and cascading the plurality of context reaction liquid state feature vectors to obtain the context semantic feature vector of the reaction liquid state.
In the above method for intelligently producing an isomaltooligosaccharide iron complex, calculating the responsiveness estimation of the reaction solution state context semantic feature vector relative to the temperature feature vector based on a gaussian density map to obtain a classification feature matrix, including:
constructing a Gaussian density map of the reaction liquid state context semantic feature vector and the temperature feature vector by using the Gaussian density map according to the following formula to obtain a responsiveness estimation Gaussian density map;
wherein the formula is:
f 1 ＝N(f|μ(f 11 ,f 12 ),∑(f 11 ,f 12 ))
wherein, mu (f) 11 ,f 12 ) The mean vector, Σ (f), representing the responsiveness estimation gaussian density map 11 ,f 12 ) A covariance matrix, f, representing said responsivity-estimated Gaussian density map 11 Representing a semantic feature vector of the context of the state of the reaction liquid, f 12 Representing the temperature eigenvector;
and carrying out Gaussian discretization on the Gaussian distribution of each position in the responsiveness estimation Gaussian density map to obtain the classification feature matrix.
In the above method for intelligently producing an isomaltooligosaccharide iron complex, the optimizing the classification feature matrix by vector particle size feature distribution based on the reaction solution state context semantic feature vector and the temperature feature vector to obtain an optimized classification feature matrix includes: calculating a graph core wandering node distribution fusion characteristic matrix between the reaction liquid state context semantic characteristic vector and the temperature characteristic vector according to the following formula;
wherein the formula is:
Figure BDA0004035219980000031
wherein, V 1 Is the context semantic feature vector, V, of the reaction solution state 2 Is the temperature eigenvector, D (V) 1 ,V 2 ) Is a distance matrix between the context semantic feature vector of the reaction solution state and the temperature feature vector, and V 1 And V 2 Are column vectors, exp (-) represents the exponential operation of a matrix, the exponential operation table of said matrixA function value of a natural exponent expressed as a power of a characteristic value of each position in the matrix,
Figure BDA0004035219980000032
representing a matrix multiplication, M c Representing the distribution and fusion feature matrix of the graph core wandering node; and matrix multiplication is carried out on the graph core wandering node distribution fusion characteristic matrix and the classification characteristic matrix to obtain the optimized classification characteristic matrix.
In the above method for intelligently producing an isomaltooligosaccharide iron complex, the step of passing the classification feature matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased, and comprises the following steps: projecting the classification feature matrix into a classification feature vector; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; passing the encoded classification feature vector through a Softmax classification function of the classifier to obtain a first probability that a reaction temperature value attributed to a current time point should increase and a second probability that the reaction temperature value attributed to the current time point should decrease; and determining the classification result based on a comparison between the first probability and the second probability.
According to another aspect of the present application, there is provided an intelligent production system of an isomalto-oligosaccharide iron complex, comprising: the data acquisition module is used for acquiring reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period; a key frame extraction module for extracting the reaction liquid state monitoring key frames of the plurality of predetermined time points from the reaction liquid state monitoring video; the first convolution coding module is used for enabling the reaction liquid state monitoring key frames at the plurality of preset time points to pass through a first convolution neural network model serving as a filter respectively so as to obtain a plurality of reaction liquid state feature vectors; the context coding module is used for enabling the reaction liquid state feature vectors to pass through a context coder based on a converter so as to obtain reaction liquid state context semantic feature vectors; the second convolution coding module is used for arranging the reaction temperature values of the plurality of preset time points into a temperature input vector according to the time dimension and then obtaining a temperature characteristic vector by using a second convolution neural network model of the one-dimensional convolution kernel; the classification characteristic matrix acquisition module is used for calculating the responsiveness estimation of the reaction liquid state context semantic characteristic vector relative to the temperature characteristic vector based on a Gaussian density map to obtain a classification characteristic matrix; the characteristic distribution optimization module is used for carrying out vector granularity characteristic distribution optimization on the classification characteristic matrix based on the reaction liquid state context semantic characteristic vector and the temperature characteristic vector to obtain an optimized classification characteristic matrix; and the classification module is used for enabling the optimized classification characteristic matrix to pass through a classifier to obtain a classification result, and the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased.
Compared with the prior art, the intelligent production system and the intelligent production method for the isomaltooligosaccharide iron complex, provided by the application, are characterized in that reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period are firstly obtained, then, a complex mapping relation between state change and reaction temperature change of a reaction liquid is simulated and established through a deep neural network model so as to obtain an optimized classification characteristic matrix, and finally, the optimized classification characteristic matrix is decoded so as to obtain a classification result which is used for representing that the reaction temperature value of the current time point is increased or decreased.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
FIG. 1 is a flow chart of an intelligent production method of an iron isomaltooligosaccharide complex according to an embodiment of the present application.
Fig. 2 is a flow chart of the method for intelligently producing the iron isomaltooligosaccharide complex according to the embodiment of the application, under a predetermined temperature control strategy, allowing the reaction mixture to react to obtain a reacted solution.
Fig. 3 is a schematic diagram of a system architecture for reacting the reaction mixture to obtain a reacted solution under a predetermined temperature control strategy in the intelligent production method of the iron isomaltooligosaccharide complex according to the embodiment of the application.
FIG. 4 is a schematic block diagram of an intelligent production system for iron isomaltooligosaccharide complexes according to embodiments of the 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 understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of a scene
As mentioned above, the current production line for the iron isomaltooligosaccharide complex does not meet the expected requirements on the preparation efficiency and the preparation cost of the iron isomaltooligosaccharide complex. The existing production line for the iron isomaltooligosaccharide complex which is not formed has relatively high preparation efficiency and relatively low preparation cost. Therefore, an optimized intelligent production scheme of the iron isomaltooligosaccharide complex is expected.
Specifically, in the technical scheme of the application, an intelligent production method of an isomaltooligosaccharide iron complex is provided, which comprises the following steps: dissolving isomaltooligosaccharide in distilled water to obtain an isomaltooligosaccharide solution; adding Na into the isomaltooligosaccharide solution 2 CO 3 Solution and FeCl 3 The solution is used for obtaining reaction mixed solution; adding a NaOH solution into the reaction mixed solution to adjust the pH value of the reaction mixed solution to 12; at a predetermined temperatureUnder a control strategy, reacting the reaction mixed solution to obtain a reacted solution; filtering the reacted solution, performing rotary evaporation concentration on the filtered filtrate, and cooling to room temperature to obtain a concentrated reacted solution; adding anhydrous ethanol into the solution after the concentrated reaction for alcohol precipitation to obtain a crude product of the isomaltooligosaccharide iron; and purifying the crude product of the isomaltooligosaccharide iron to obtain the isomaltooligosaccharide iron complex.
Accordingly, it has been found that the effect of the prepared iron isomaltooligosaccharide complex is not ideal in the actual process of preparing the iron isomaltooligosaccharide complex, because the reaction effect of the prepared solution after the reaction is poor under the predetermined temperature control strategy, and the expected effect of the reaction efficiency and the reaction quality are difficult to achieve. That is, in the process of actually using the reaction mixture to prepare the solution after the reaction, the temperature control should be adapted to the reaction state of the reaction solution, that is, the change in the reaction temperature should be adaptively adjusted based on the reaction state change characteristics of the reaction solution. The difficulty is how to establish a mapping relationship between the state change of the reaction liquid and the reaction temperature so that the reaction temperature is adaptively adjusted based on the state change of the reaction liquid for the purpose of improving the reaction efficiency and the preparation quality.
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 also exhibit a level close to or even exceeding that of 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 a new solution for excavating a complex mapping relation between state change and reaction temperature change of the reaction liquid. Those skilled in the art will appreciate that the deep neural network model based on deep learning can adjust its parameters through a suitable training strategy, for example, through a back propagation algorithm of gradient descent to enable it to simulate a complex nonlinear correlation between things, which is obviously suitable for simulating and establishing a complex mapping relationship between a state change of a reaction liquid and a reaction temperature change.
Specifically, in the technical scheme of the application, firstly, reaction temperature values of a plurality of predetermined time points in a predetermined time period and a reaction liquid state monitoring video of the predetermined time period are acquired. Then, it is considered that in the reaction liquid state monitoring video, the state change characteristics of the reaction liquid can be represented by the difference between adjacent monitoring frames in the reaction liquid state monitoring video, that is, the change condition of the reaction liquid state is represented by the image representation of the adjacent image frames. However, considering that the difference between adjacent frames in the monitoring video is small and a large amount of data redundancy exists, in order to reduce the amount of calculation and avoid adverse effects on detection caused by the data redundancy, the reaction liquid state monitoring video is subjected to key frame sampling at a predetermined sampling frequency to extract the reaction liquid state monitoring key frames at the plurality of predetermined time points from the reaction liquid state monitoring video. Here, it is worth mentioning that the sampling frequency may be adjusted based on the application requirements of the actual scene, rather than a default value.
Then, feature mining is performed on the reaction liquid state monitoring key frames at the plurality of preset time points by using a first convolution neural network model which is used as a filter and has excellent performance in local implicit feature extraction of the image, so that implicit feature distribution information about the reaction liquid state in each reaction liquid state monitoring key frame is extracted respectively, and a plurality of reaction liquid state feature vectors are obtained.
Next, it is considered that, in the reaction liquid state monitoring video, the state change characteristics regarding the reaction liquid are correlated at each predetermined time point in the time dimension, that is, the state of the reaction liquid has change characteristic distribution information which is dynamic in time series. Therefore, in the technical solution of the present application, the plurality of reaction liquid state feature vectors are passed through a context encoder based on a converter to obtain a reaction liquid state context semantic feature vector. That is, based on the transformer idea, with the converter being able to capture the characteristic of long-distance context dependence, the plurality of reaction liquid state feature vectors are subjected to global-based context semantic coding to obtain a context semantic association feature representation with the overall semantic association of the plurality of reaction liquid state feature vectors as context background, that is, the reaction liquid state context semantic feature vectors. It should be understood that, in the technical solution of the present application, the context encoder based on the converter may capture the context semantic dynamic association feature representation of the state implicit feature about the reaction liquid in each key frame with respect to the whole context within the predetermined time period.
Furthermore, considering that the reaction temperature values also have a dynamic rule in the time dimension, in order to be able to sufficiently and accurately extract the dynamic association feature distribution information of the reaction temperature values in the time sequence, the reaction temperature values at the plurality of predetermined time points are arranged as a temperature input vector according to the time dimension, and then are processed in a second convolution neural network model using a one-dimensional convolution kernel so as to extract the dynamic association features of the reaction temperatures in the time sequence, thereby obtaining a temperature feature vector.
Then, calculating the responsiveness estimation of the context semantic feature vector of the reaction liquid state relative to the temperature feature vector so as to fuse the relevance feature distribution information representing the state change of the reaction liquid and the temperature change in a high-dimensional space, thereby obtaining a classification feature matrix. In particular, considering that the reaction solution state context semantic feature vector and the temperature feature vector each correspond to a feature distribution manifold in a high-dimensional feature space, and the feature distribution manifolds are due to their irregular shapes and scattering positions, if the relevance feature representations of the reaction solution state context semantic feature vector and the temperature feature vector are represented only by cascading the two, it is equivalent to simply superimposing the feature distribution manifolds in the original positions and shapes, so that the boundaries of the newly obtained feature distribution manifolds become very irregular and complex, and when an optimal point is found by gradient descent, it is very easy to fall into a local extreme point and a global optimal point cannot be obtained. Therefore, it is further necessary to appropriately fuse the reaction liquid state context semantic feature vector and the temperature feature vector so that the respective feature distributions can be topographically converged with respect to each other.
It should be understood that the gaussian density map is widely used for estimation based on a priori target posteriori in deep learning, and thus can be used for correcting data distribution, thereby achieving the above purpose. Specifically, in the technical solution of the present application, first, a gaussian density map of the reaction solution state context semantic feature vector and the temperature feature vector is constructed based on gaussian distribution. And then, carrying out Gaussian discretization processing on the Gaussian density map so as not to generate information loss when the data features are expanded, thereby obtaining a classification feature matrix.
Then, the optimized classification feature matrix is further processed by a classifier to obtain a classification result indicating that the reaction temperature 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 reaction temperature value at the current time point should be increased or decreased, wherein the classifier determines to which classification label the classification feature matrix belongs by 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 reaction temperature value control strategy label, and therefore, after the classification result is obtained, the reaction temperature value can be adaptively adjusted based on the classification result, so as to achieve the purpose of improving the reaction efficiency and the preparation quality.
In particular, in the technical solution of the present application, when the classification feature matrix is obtained by calculating the responsiveness of the reaction liquid state context semantic feature vector with respect to the temperature feature vector based on the gaussian density map, the classification feature matrix is obtained via the gaussian density map based on the feature values of the respective positions of the reaction liquid state context semantic feature vector and the temperature feature vector, so that the classification feature matrix can express the responsiveness estimation features of the feature value granularities of the reaction liquid state context semantic feature vector and the temperature feature vector, but at the same time, it is still desirable that the classification feature matrix can express the responsiveness estimation features of the vector granularities of the reaction liquid state context semantic feature vector and the temperature feature vector.
Therefore, preferably, the reaction liquid state context semantic feature vector V is further calculated 1 And said temperature eigenvector V 2 The distribution and fusion feature matrix of the graph core wandering nodes between the two is expressed as follows:
Figure BDA0004035219980000081
D(V 1 ,V 2 ) For the reaction solution state context semantic feature vector V 1 And said temperature eigenvector V 2 A matrix of distances between, i.e. d i,j ＝d(v 1i ,v 2j ) And V is 1 And V 2 Are column vectors.
The idea of simulating the graph core by the graph core migration node distribution fusion characteristic matrix is to use the reaction liquid state context semantic feature vector V 1 And the temperature eigenvector V 2 Respectively viewed as nodes in the graph, based on the semantic feature vector V of the context of the reaction liquid state 1 And said temperature eigenvector V 2 Is run on the distance topological graph to generalize the topological nodes to semantic feature vectors V relative to the context of the reaction liquid state 1 And said temperature eigenvector V 2 Under the scene that the classification regression feature distribution has continuous high-dimensional regression space attributes, thereby representing the reaction liquid state context semantic feature vector V serving as a topological node 1 And the temperature eigenvector V 2 Local distribution information in a high-dimensional feature space of a fusion feature to express the reaction liquid state context semantic feature vector V 1 And the temperature eigenvector V 2 Vector granularity of (d) between.
Further, matrix multiplication is carried out on the graph core walking node distribution fusion feature matrix and the classification feature matrix, so that the classification feature matrix is mapped into a relevance response feature space, the classification feature matrix further expresses the response estimation features of the vector granularity of the reaction liquid state context semantic feature vector and the temperature feature vector, and the classification feature matrix is optimized. And then, carrying out classification control on the reaction temperature value of the current time point by the optimized classification characteristic matrix through a classifier. Therefore, the reaction temperature can be adaptively adjusted in real time based on the state change of the reaction liquid, so that the reaction efficiency and the preparation quality are improved.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
As mentioned above, the current production line for the iron isomaltooligosaccharide complex does not meet the expected requirements on the preparation efficiency and the preparation cost of the iron isomaltooligosaccharide complex. The existing production line for the iron isomaltooligosaccharide complex which is not formed has relatively high preparation efficiency and relatively low preparation cost. Therefore, an optimized intelligent production scheme of the iron isomaltooligosaccharide complex is desired. Specifically, in the technical scheme of the application, an intelligent production method of the isomaltooligosaccharide iron complex is provided. FIG. 1 is a flow chart of a method for the intelligent production of an iron isomaltooligosaccharide complex according to an embodiment of the present application. As shown in fig. 1, the intelligent production method of the iron isomaltooligosaccharide complex according to the embodiment of the application comprises: s110, dissolving isomaltooligosaccharide in distilled water to obtain an isomaltooligosaccharide solution; s120, adding Na into the isomaltooligosaccharide solution 2 CO 3 Solution and FeCl 3 The solution is used for obtaining reaction mixed solution; s130, adding a NaOH solution into the reaction mixed solution to adjust the pH value of the reaction mixed solution to 12; s140, under a preset temperature control strategy, reacting the reaction mixed solution to obtain a reacted solution; s150, filtering the reacted solution, performing rotary evaporation concentration on the filtered filtrate, and cooling to room temperature to obtain a concentrated reacted solution; s160, adding the concentrated reacted solutionCarrying out alcohol precipitation by using anhydrous ethanol to obtain a crude product of the isomaltooligosaccharide iron; and S170, purifying the crude product of the isomaltooligosaccharide iron to obtain the isomaltooligosaccharide iron complex.
Further, considering that the effect of the prepared iron isomaltooligosaccharide complex is not ideal when actually performing the preparation process of the iron isomaltooligosaccharide complex, the reaction effect of the prepared solution after the reaction is poor under the predetermined temperature control strategy, and the expected effect of the reaction efficiency and the reaction quality are difficult to achieve. That is, in the process of actually using the reaction mixture to prepare the solution after the reaction, the temperature control should be adapted to the reaction state of the reaction solution, that is, the change in the reaction temperature should be adaptively adjusted based on the reaction state change characteristics of the reaction solution. Therefore, if the change of the reaction temperature can be adaptively adjusted based on the reaction state change characteristics of the reaction solution, the reaction efficiency and the preparation quality can be certainly improved, and therefore, on the basis of the above-mentioned intelligent production method of the iron isomaltooligosaccharide complex, the step S140, more specifically, the temperature control strategy is improved such that the change of the reaction temperature is adaptively adjusted based on the reaction state change characteristics of the reaction solution, thereby improving the reaction efficiency and the preparation quality.
Fig. 2 is a flow chart of the method for intelligently producing the iron isomaltooligosaccharide complex according to the embodiment of the application, under a predetermined temperature control strategy, allowing the reaction mixture to react to obtain a reacted solution. As shown in fig. 2, the reacting the reaction mixture to obtain a reacted solution under the predetermined temperature control strategy according to the embodiment of the application includes: s210, obtaining reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period; s220, extracting the reaction liquid state monitoring key frames of the plurality of preset time points from the reaction liquid state monitoring video; s230, enabling the reaction liquid state monitoring key frames of the plurality of preset time points to pass through a first convolution neural network model serving as a filter respectively to obtain a plurality of reaction liquid state feature vectors; s240, enabling the plurality of reaction liquid state feature vectors to pass through a context encoder based on a converter to obtain reaction liquid state context semantic feature vectors; s250, arranging the reaction temperature values of the plurality of preset time points into a temperature input vector according to a time dimension, and then obtaining a temperature characteristic vector by using a second convolution neural network model of a one-dimensional convolution kernel; s260, calculating the responsiveness estimation of the reaction liquid state context semantic feature vector relative to the temperature feature vector based on a Gaussian density map to obtain a classification feature matrix; s270, based on the reaction liquid state context semantic feature vector and the temperature feature vector, performing vector granularity feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix; and S280, passing the optimized classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased.
Fig. 3 is a schematic diagram of a system architecture of reacting the reaction mixture to obtain a reacted solution under a predetermined temperature control strategy in the intelligent production method of the iron isomaltooligosaccharide complex according to the embodiment of the application. As shown in fig. 3, in the system architecture of the embodiment of the present application, in which the reaction mixed solution is reacted under the predetermined temperature control strategy to obtain a reacted solution, first, reaction temperature values at a plurality of predetermined time points in a predetermined time period and a reaction solution state monitoring video in the predetermined time period are obtained. And then, extracting the reaction liquid state monitoring key frames of the plurality of preset time points from the reaction liquid state monitoring video, and respectively enabling the reaction liquid state monitoring key frames of the plurality of preset time points to pass through a first convolution neural network model serving as a filter to obtain a plurality of reaction liquid state feature vectors. Then, the plurality of reaction liquid state feature vectors are passed through a context encoder based on a converter to obtain reaction liquid state context semantic feature vectors. Meanwhile, the reaction temperature values of the plurality of preset time points are arranged into a temperature input vector according to the time dimension, and then a second convolution neural network model of a one-dimensional convolution kernel is used for obtaining a temperature characteristic vector. And then calculating the responsiveness estimation of the reaction liquid state context semantic feature vector relative to the temperature feature vector based on a Gaussian density map to obtain a classification feature matrix. And then, based on the reaction liquid state context semantic feature vector and the temperature feature vector, carrying out vector granularity feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix. And finally, passing the optimized classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased.
In step S210, reaction temperature values at a plurality of predetermined time points within a predetermined time period and a reaction liquid state monitoring video of the predetermined time period are obtained. It will be appreciated that the difficulty in adaptively adjusting the change in reaction temperature based on the characteristic of the change in reaction state of the reaction liquid is how to establish a mapping relationship between the change in state of the reaction liquid and the reaction temperature such that the reaction temperature is adaptively adjusted based on the change in state of the reaction liquid for the purpose of improving reaction efficiency and production quality. And deep learning and development of a neural network provide a new solution for excavating a complex mapping relation between the state change of the reaction liquid and the change of the reaction temperature. Those skilled in the art will appreciate that the deep neural network model based on deep learning can adjust its parameters through a suitable training strategy, for example, through a back propagation algorithm of gradient descent to enable it to simulate a complex nonlinear correlation between things, which is obviously suitable for simulating and establishing a complex mapping relationship between a state change of a reaction liquid and a reaction temperature change.
In a specific example of the application, reaction temperature values of a plurality of predetermined time points in a predetermined time period are acquired through a temperature sensor, and a reaction liquid state monitoring video of the predetermined time period is acquired through a camera. And the reaction temperature values of a plurality of preset time points in the preset time period comprise past temperature control strategies. The reaction liquid state monitoring video of the preset time period comprises state change characteristic information of the reaction liquid.
In step S220, the reaction liquid state monitoring key frames at the plurality of predetermined time points are extracted from the reaction liquid state monitoring video. It should be understood that, in the reaction liquid state monitoring video, the state change characteristics of the reaction liquid can be represented by the difference between adjacent monitoring frames in the reaction liquid state monitoring video, that is, the change condition of the reaction liquid state is represented by the image representation of the adjacent image frames. However, in consideration of the fact that the difference between adjacent frames in the monitoring video is small and a large amount of data redundancy exists, in order to reduce the calculation amount and avoid adverse effects on detection caused by the data redundancy, the reaction liquid state monitoring key frames at the plurality of predetermined time points are extracted from the reaction liquid state monitoring video. Here, the plurality of predetermined time points of the reaction liquid state monitoring key frame are the same as the plurality of predetermined time points of the reaction temperature value, and the plurality of predetermined time points may be set according to an actual application scenario.
In step S230, the reaction solution state monitoring key frames at the plurality of predetermined time points are respectively passed through a first convolution neural network model as a filter to obtain a plurality of reaction solution state feature vectors. It should be understood that, considering that the reaction liquid state monitoring key frames at the plurality of predetermined time points contain abundant implicit feature information, and the first convolution neural network model has excellent performance in local implicit feature extraction of the image, feature mining is performed on the reaction liquid state monitoring key frames at the plurality of predetermined time points through the first convolution neural network model serving as a filter to extract implicit feature distribution information about the reaction liquid state in each reaction liquid state monitoring key frame, so as to obtain a plurality of reaction liquid state feature vectors.
In a specific example of the present application, the passing the reaction liquid state monitoring key frames at the plurality of predetermined time points through a first convolution neural network model as a filter to obtain a plurality of reaction liquid state feature vectors includes: each layer of the first convolutional neural network model respectively performs the following operations on input data in forward transmission of the layer: performing convolution processing on the input data based on a convolution kernel to generate a convolution feature map; performing global mean pooling processing based on a feature matrix on the convolution feature map to generate a pooled feature map; and carrying out nonlinear activation on the feature values of all positions in the pooled feature map to generate an activated feature increasing map; the input of the first convolutional neural network model is each of the reaction liquid state monitoring key frames at the plurality of predetermined time points, the input of the second layer to the last layer of the first convolutional neural network model is the output of the previous layer, and the output of the last layer of the first convolutional neural network model is each of the reaction liquid state feature vectors in the plurality of reaction liquid state feature vectors.
In step S240, the plurality of reaction liquid state feature vectors are passed through a context encoder based on a converter to obtain reaction liquid state context semantic feature vectors. It should be understood that, in the reaction liquid state monitoring video, the state change characteristics of the reaction liquid are correlated at each predetermined time point in the time dimension, that is, the state of the reaction liquid has dynamic change characteristic distribution information in time sequence. Therefore, in the technical solution of the present application, the plurality of reaction liquid state feature vectors are passed through a context encoder based on a converter to obtain a reaction liquid state context semantic feature vector. That is, based on the transformer idea, with the converter being able to capture the characteristic of long-distance context dependence, the plurality of reaction liquid state feature vectors are subjected to global-based context semantic coding to obtain a context semantic association feature representation with the overall semantic association of the plurality of reaction liquid state feature vectors as context background, that is, the reaction liquid state context semantic feature vectors. It should be understood that, in the technical solution of the present application, the context encoder based on the converter may capture the context semantic dynamic association feature representation of the state implicit feature about the reaction liquid in each key frame with respect to the whole context within the predetermined time period.
In a specific example of the present application, the passing the plurality of reaction liquid state feature vectors through a context encoder based on a converter to obtain a reaction liquid state context semantic feature vector includes: passing the plurality of reaction solution state feature vectors through the converter-based context encoder to obtain a plurality of context reaction solution state feature vectors; and cascading the plurality of context reaction liquid state feature vectors to obtain the context semantic feature vector of the reaction liquid state.
In a specific example of the present application, the passing the plurality of reaction liquid state feature vectors through the converter-based context encoder to obtain a plurality of context reaction liquid state feature vectors includes: and performing global context semantic-based coding on each reaction liquid state feature vector in the plurality of reaction liquid state feature vectors by using a Bert model based on a converter of the context encoder to obtain a plurality of context reaction liquid state feature vectors taking overall semantic association of the plurality of reaction liquid state feature vectors as context.
In step S250, the reaction temperature values at the plurality of predetermined time points are arranged as a temperature input vector according to a time dimension, and then a temperature feature vector is obtained by using a second convolution neural network model of a one-dimensional convolution kernel. It should be understood that, considering that the reaction temperature value also has a dynamic rule in the time dimension, in order to be able to sufficiently and accurately extract the dynamic association feature distribution information of the reaction temperature value in the time sequence, the reaction temperature values of the plurality of predetermined time points are arranged as a temperature input vector according to the time dimension, and then are processed in the second convolution neural network model using a one-dimensional convolution kernel to extract the dynamic association features of the reaction temperature in the time sequence, so as to obtain a temperature feature vector.
In a specific example of the present application, after arranging the reaction temperature values of the plurality of predetermined time points into a temperature input vector according to a time dimension, obtaining a temperature feature vector by using a second convolutional neural network model of a one-dimensional convolutional kernel, the method includes: each layer of the second convolutional neural network model performs, in a layer forward pass, respectively: performing convolution processing on the input data based on a one-dimensional convolution kernel to generate a convolution characteristic diagram; performing global mean pooling processing based on a feature matrix on the convolution feature map to generate a pooled feature map; and carrying out nonlinear activation on the feature values of all positions in the pooled feature map to generate an activated feature increasing map; the input of the second convolutional neural network model is the temperature input vector, the input of the second layer to the last layer of the second convolutional neural network model is the output of the previous layer, and the output of the last layer of the second convolutional neural network model is the temperature feature vector.
In step S260, calculating a responsiveness estimate of the reaction liquid state context semantic feature vector with respect to the temperature feature vector based on a gaussian density map to obtain a classification feature matrix. It should be understood that the responsiveness estimation of the context semantic feature vector of the state of the reaction liquid relative to the temperature feature vector is calculated to fuse the feature distribution information representing the relevance between the state change of the reaction liquid and the temperature change in the high-dimensional space, so as to obtain the classification feature matrix. In particular, considering that the reaction solution state context semantic feature vector and the temperature feature vector each correspond to a feature distribution manifold in a high-dimensional feature space, and the feature distribution manifolds are due to their irregular shapes and scattering positions, if the relevance feature representations of the reaction solution state context semantic feature vector and the temperature feature vector are represented only by cascading the two, it is equivalent to simply superimposing the feature distribution manifolds in the original positions and shapes, so that the boundaries of the newly obtained feature distribution manifolds become very irregular and complex, and when an optimal point is found by gradient descent, it is very easy to fall into a local extreme point and a global optimal point cannot be obtained. Therefore, it is further necessary to appropriately fuse the reaction liquid state context semantic feature vector and the temperature feature vector so that the respective feature distributions can be topographically converged with respect to each other.
It should be understood that the gaussian density map is widely used for estimation based on a priori target posteriori in deep learning, and therefore can be used for correcting data distribution, thereby achieving the purpose. Specifically, in the technical solution of the present application, first, a gaussian density map of the reaction solution state context semantic feature vector and the temperature feature vector is constructed based on gaussian distribution. And then, carrying out Gaussian discretization processing on the Gaussian density map so as not to generate information loss when the data features are expanded, thereby obtaining a classification feature matrix.
In a specific example of the present application, calculating a responsiveness estimate of the reaction solution state context semantic feature vector with respect to the temperature feature vector based on a gaussian density map to obtain a classification feature matrix, includes: constructing a Gaussian density map of the reaction liquid state context semantic feature vector and the temperature feature vector by using the Gaussian density map according to the following formula to obtain a responsiveness estimation Gaussian density map;
wherein the formula is:
f 1 ＝N(f|μ(f 11 ,f 12 ),∑(f 11 ,f 12 ))
wherein, μ (f) 11 ,f 12 ) The mean vector, Σ (f), representing the responsiveness estimated gaussian density map 11 ,f 12 ) Covariance matrix, f, representing the responsiveness estimated Gaussian density map 11 Representing the reaction liquid state context semantic feature vector, f 12 Representing the temperature eigenvector; and performing Gaussian discretization on the Gaussian distribution of each position in the responsiveness estimation Gaussian density map to obtain the classification feature matrix.
In step S270, based on the semantic feature vector of the context of the reaction solution state and the temperature feature vector, performing vector particle size feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix. Particularly, in the technical solution of the present application, when the classification feature matrix is obtained by calculating the responsiveness estimation of the reaction liquid state context semantic feature vector with respect to the temperature feature vector based on the gaussian density map, the classification feature matrix is obtained by obtaining the classification feature matrix through the gaussian density map based on the feature values of the reaction liquid state context semantic feature vector and the temperature feature vector at each positionA feature matrix, thus enabling the classification feature matrix to express responsiveness estimation features of the feature value granularities of the reaction liquid state context semantic feature vector and the temperature feature vector, but at the same time, it is still desirable that the classification feature matrix be able to express responsiveness estimation features of the vector granularities of the reaction liquid state context semantic feature vector and the temperature feature vector. Therefore, preferably, the reaction liquid state context semantic feature vector V is further calculated 1 And said temperature eigenvector V 2 And distributing and fusing the characteristic matrixes by the graph core wandering nodes.
In a specific example of the present application, the performing, based on the context semantic feature vector of the reaction liquid state and the temperature feature vector, vector granularity feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix includes: calculating a graph core wandering node distribution fusion characteristic matrix between the reaction liquid state context semantic characteristic vector and the temperature characteristic vector according to the following formula; wherein the formula is:
Figure BDA0004035219980000151
wherein, V 1 Is the context semantic feature vector, V, of the reaction solution state 2 Is the temperature eigenvector, D (V) 1 ,V 2 ) Is a distance matrix between the context semantic feature vector of the reaction solution state and the temperature feature vector, and V 1 And V 2 Are column vectors, exp (-) represents an exponential operation of a matrix representing a natural exponential function value raised to the eigenvalue of each position in the matrix,
Figure BDA0004035219980000152
representing a matrix multiplication, M c Representing the distribution and fusion feature matrix of the graph core wandering node;
the idea of simulating the graph core by the graph core migration node distribution fusion characteristic matrix is to use the reaction liquid state context semantic feature vector V 1 And the temperature eigenvector V 2 Respectively viewed as nodes in the graph, based on the semantic feature vector V of the context of the reaction liquid state 1 And the temperature eigenvector V 2 Is run on the distance topological graph to generalize the topological nodes to semantic feature vectors V relative to the context of the reaction liquid state 1 And the temperature eigenvector V 2 Under the scene that the classification regression feature distribution has continuous high-dimensional regression space attribute, thereby representing the reaction liquid state context semantic feature vector V as a topological node 1 And said temperature eigenvector V 2 Local distribution information in a high-dimensional feature space of a fusion feature to express the reaction liquid state context semantic feature vector V 1 And the temperature eigenvector V 2 Vector granularity of (d) between.
Further, matrix multiplication is carried out on the graph core wandering node distribution fusion characteristic matrix and the classification characteristic matrix to obtain the optimized classification characteristic matrix. That is, the classification feature matrix is mapped into a relevance response feature space such that the classification feature matrix further expresses the response estimation features of the vector granularity of the reaction liquid state context semantic feature vector and the temperature feature vector, thereby optimizing the classification feature matrix. And then, carrying out classification control on the reaction temperature value of the current time point by the optimized classification characteristic matrix through a classifier. Therefore, the reaction temperature can be adaptively adjusted in real time based on the state change of the reaction liquid, so that the reaction efficiency and the preparation quality are improved.
In step S280, the optimized classification feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used to indicate that the reaction temperature 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 reaction temperature value at the current time point should be increased or decreased, wherein the classifier determines to which classification label the classification feature matrix belongs by 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 reaction temperature value control strategy label, and therefore, after the classification result is obtained, the reaction temperature value can be adaptively adjusted based on the classification result, so as to achieve the purpose of improving the reaction efficiency and the preparation quality.
In a specific example of the present application, the passing the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the reaction temperature value at the current time point should be increased or decreased, includes: projecting the classification feature matrix into a classification feature vector; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; passing the encoded classification feature vector through a Softmax classification function of the classifier to obtain a first probability that a reaction temperature value attributed to a current time point should increase and a second probability that the reaction temperature value attributed to the current time point should decrease; and determining the classification result based on a comparison between the first probability and the second probability.
More specifically, a predetermined value for increasing or decreasing the reaction temperature may be set, for example, 0.5 degrees, and of course, the predetermined value for the actual change should be adjusted based on the application requirement of the actual scene, and when the first probability is greater than the second probability, the reaction temperature value control strategy labels that the reaction temperature value at the current time point should be increased, and at this time, the reaction temperature value is controlled to be increased by the predetermined value. Otherwise, when the first probability is smaller than the second probability, the reaction temperature value control strategy label is that the reaction temperature value at the current time point should be decreased, and at this time, the reaction temperature value is controlled to be decreased by a predetermined value. Then, the reaction temperature is adjusted in a self-adaptive manner in real time accurately based on the state change of the reaction liquid by cyclic reciprocation, so that the reaction efficiency and the preparation quality are improved.
In summary, the method for intelligently producing the isomaltooligosaccharide iron complex according to the embodiment of the present application has been elucidated, which includes first obtaining reaction temperature values at a plurality of predetermined time points within a predetermined time period and a reaction solution state monitoring video of the predetermined time period, then simulating and establishing a complex mapping relationship between a state change of a reaction solution and a reaction temperature change through a deep neural network model to obtain an optimized classification feature matrix, and finally decoding the optimized classification feature matrix to obtain a classification result indicating that the reaction temperature value at the current time point should be increased or decreased.
Exemplary System
FIG. 4 is a schematic block diagram of an intelligent production system for an iron isomaltooligosaccharide complex according to an embodiment of the present application. As shown in fig. 4, the system 100 for intelligent production of iron isomaltooligosaccharide complex according to the embodiment of the present application comprises: the data acquisition module 110 is configured to acquire reaction temperature values at a plurality of predetermined time points in a predetermined time period and a reaction liquid state monitoring video of the predetermined time period; a key frame extracting module 120, configured to extract the reaction liquid state monitoring key frames at the plurality of predetermined time points from the reaction liquid state monitoring video; the first convolution encoding module 130 is configured to pass the reaction solution state monitoring key frames at the plurality of predetermined time points through a first convolution neural network model serving as a filter, respectively, to obtain a plurality of reaction solution state feature vectors; a context encoding module 140, configured to pass the plurality of reaction liquid state feature vectors through a converter-based context encoder to obtain reaction liquid state context semantic feature vectors; the second convolution coding module 150 is configured to arrange the reaction temperature values of the plurality of predetermined time points into a temperature input vector according to a time dimension, and then obtain a temperature feature vector by using a second convolution neural network model of a one-dimensional convolution kernel; a classification feature matrix obtaining module 160, configured to calculate a responsiveness estimate of the reaction liquid state context semantic feature vector with respect to the temperature feature vector based on a gaussian density map to obtain a classification feature matrix; the feature distribution optimization module 170 is configured to perform vector granularity feature distribution optimization on the classification feature matrix based on the reaction liquid state context semantic feature vector and the temperature feature vector to obtain an optimized classification feature matrix; and a classification 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 reaction temperature value at the current time point should be increased or decreased.
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-mentioned intelligent production system of an iron isomaltooligosaccharide complex have been described in detail in the above description of the intelligent production method of an iron isomaltooligosaccharide complex with reference to FIGS. 1 to 3, and thus, the repetitive description thereof will be omitted.

Claims (10)

1. An intelligent production method of an isomaltooligosaccharide iron complex is characterized by comprising the following steps: dissolving isomaltooligosaccharide in distilled water to obtain isomaltooligosaccharide solution; adding Na into the isomaltooligosaccharide solution 2 CO 3 Solution and FeCl 3 The solution is used for obtaining reaction mixed solution; adding a NaOH solution into the reaction mixed solution to adjust the pH value of the reaction mixed solution to 12; under a preset temperature control strategy, reacting the reaction mixed solution to obtain a reacted solution; filtering the reacted solution, performing rotary evaporation concentration on the filtered filtrate, and cooling to room temperature to obtain a concentrated reacted solution; adding anhydrous ethanol into the solution after the concentrated reaction for alcohol precipitation to obtain a crude product of the isomaltooligosaccharide iron; and purifying the crude product of the isomaltose hypgather iron to obtain the isomaltose hypgather iron complex.
2. The intelligent production method of iron isomaltooligosaccharide complex as claimed in claim 1, wherein the reacting the reaction mixture under a predetermined temperature control strategy to obtain a reacted solution comprises: obtaining reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period; extracting reaction liquid state monitoring key frames of the plurality of preset time points from the reaction liquid state monitoring video; respectively enabling the reaction liquid state monitoring key frames of the plurality of preset time points to pass through a first convolution neural network model serving as a filter to obtain a plurality of reaction liquid state feature vectors; passing the plurality of reaction liquid state feature vectors through a context encoder based on a converter to obtain reaction liquid state context semantic feature vectors; arranging the reaction temperature values of the plurality of preset time points into a temperature input vector according to a time dimension, and then obtaining a temperature characteristic vector by using a second convolution neural network model of a one-dimensional convolution kernel; calculating the responsiveness estimation of the reaction liquid state context semantic feature vector relative to the temperature feature vector based on a Gaussian density map to obtain a classification feature matrix; based on the reaction liquid state context semantic feature vector and the temperature feature vector, carrying out vector granularity feature distribution optimization on the classification feature matrix to obtain an optimized classification feature matrix; and passing the optimized classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the reaction temperature value at the current time point should be increased or decreased.
3. The intelligent production method of iron isomaltooligosaccharide complex as claimed in claim 2, wherein the step of passing the monitoring key frames of reaction solution status at the predetermined time points through the first convolution neural network model as a filter to obtain a plurality of feature vectors of reaction solution status comprises: each layer of the first convolutional neural network model respectively performs the following operations on input data in forward transmission of the layer: performing convolution processing on the input data based on a convolution kernel to generate a convolution feature map; performing global mean pooling processing based on a feature matrix on the convolution feature map to generate a pooled feature map; and carrying out nonlinear activation on the feature values of all positions in the pooled feature map to generate an activated feature increasing map; the input of the first convolutional neural network model is each of the reaction liquid state monitoring key frames at the plurality of predetermined time points, the input of the second layer to the last layer of the first convolutional neural network model is the output of the previous layer, and the output of the last layer of the first convolutional neural network model is each of the reaction liquid state feature vectors in the plurality of reaction liquid state feature vectors.
4. The intelligent method for producing an isomaltooligosaccharide iron complex as claimed in claim 3, wherein the step of passing the plurality of reaction solution state feature vectors through a context encoder based on a converter to obtain the reaction solution state context semantic feature vector comprises: passing the plurality of reaction solution state feature vectors through the converter-based context encoder to obtain a plurality of context reaction solution state feature vectors; and cascading the plurality of context reaction liquid state feature vectors to obtain the context semantic feature vector of the reaction liquid state.
5. The intelligent method of claim 4, wherein the passing the plurality of reaction solution state feature vectors through the converter-based context encoder to obtain a plurality of context reaction solution state feature vectors comprises: and carrying out global context semantic-based coding on each reaction liquid state feature vector in the plurality of reaction liquid state feature vectors by using a Bert model based on a converter of the context encoder to obtain the plurality of context reaction liquid state feature vectors taking overall semantic association of the plurality of reaction liquid state feature vectors as context.
6. The intelligent production method of iron isomaltooligosaccharide complex as claimed in claim 5, wherein the step of arranging the reaction temperature values of the predetermined time points into a temperature input vector according to the time dimension and then obtaining the temperature feature vector by using a second convolution neural network model of one-dimensional convolution kernel comprises: each layer of the second convolutional neural network model performs, in a layer forward pass, respectively: performing convolution processing on the input data based on a one-dimensional convolution kernel to generate a convolution characteristic diagram; performing global mean pooling processing based on a feature matrix on the convolution feature map to generate a pooled feature map; and carrying out nonlinear activation on the feature values of all positions in the pooled feature map to generate an activated feature increasing map; the input of the second convolutional neural network model is the temperature input vector, the input of the second layer to the last layer of the second convolutional neural network model is the output of the previous layer, and the output of the last layer of the second convolutional neural network model is the temperature feature vector.
7. The intelligent production method of iron isomaltooligosaccharide complex as claimed in claim 6, wherein calculating the responsiveness estimate of the reaction solution state context semantic feature vector relative to the temperature feature vector based on Gaussian density map to obtain the classification feature matrix comprises: constructing a Gaussian density map of the reaction liquid state context semantic feature vector and the temperature feature vector by using the Gaussian density map according to the following formula to obtain a responsiveness estimation Gaussian density map; wherein the formula is:
f 1 ＝N(f|μ(f 11 ,f 12 ),∑(f 11 ,f 12 ))
wherein, μ (f) 11 ,f 12 ) The mean vector, Σ (f), representing the responsiveness estimated gaussian density map 11 ,f 12 ) A covariance matrix, f, representing said responsivity-estimated Gaussian density map 11 Representing a semantic feature vector of the context of the state of the reaction liquid, f 12 Representing the temperature eigenvector; and performing Gaussian discretization on the Gaussian distribution of each position in the responsiveness estimation Gaussian density map to obtain the classification feature matrix.
8. The intelligent production method of iron isomaltooligosaccharide complex as claimed in claim 7, wherein the vector particle size feature distribution optimization of the classification feature matrix based on the context semantic feature vector of the reaction solution state and the temperature feature vector to obtain an optimized classification feature matrix comprises: calculating a graph core wandering node distribution fusion characteristic matrix between the reaction liquid state context semantic characteristic vector and the temperature characteristic vector according to the following formula; wherein the formula is:
Figure FDA0004035219970000031
wherein, V 1 Is the context semantic feature vector, V, of the reaction solution state 2 Is the temperature eigenvector, D (V) 1 ,V 2 ) Is a distance matrix between the context semantic feature vector of the reaction solution state and the temperature feature vector, and V 1 And V 2 Are column vectors, exp (-) represents an exponential operation of a matrix representing a natural exponential function value raised to the eigenvalue of each position in the matrix,
Figure FDA0004035219970000032
representing a matrix multiplication, M c Representing the distribution and fusion feature matrix of the graph core wandering node; and matrix multiplying the graph core wandering node distribution fusion characteristic matrix and the classification characteristic matrix to obtain the optimized classification characteristic matrix.
9. The intelligent method for producing an isomaltooligosaccharide iron complex as claimed in claim 8, wherein the step of passing the classification feature matrix through a classifier to obtain a classification result, wherein the classification result is used to indicate that the reaction temperature value at the current time point should be increased or decreased, comprises: projecting the classification feature matrix into a classification feature vector; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; passing the encoded classification feature vector through a Softmax classification function of the classifier to obtain a first probability that a reaction temperature value attributed to a current time point should increase and a second probability that the reaction temperature value attributed to the current time point should decrease; and determining the classification result based on a comparison between the first probability and the second probability.
10. The intelligent production system of the iron isomaltooligosaccharide complex is characterized by comprising the following components: the data acquisition module is used for acquiring reaction temperature values of a plurality of preset time points in a preset time period and a reaction liquid state monitoring video of the preset time period; a key frame extraction module, configured to extract the reaction liquid state monitoring key frames at the plurality of predetermined time points from the reaction liquid state monitoring video; the first convolution coding module is used for enabling the reaction liquid state monitoring key frames at the plurality of preset time points to pass through a first convolution neural network model serving as a filter respectively so as to obtain a plurality of reaction liquid state feature vectors; the context coding module is used for enabling the plurality of reaction liquid state feature vectors to pass through a context coder based on a converter so as to obtain reaction liquid state context semantic feature vectors; the second convolution coding module is used for arranging the reaction temperature values of the plurality of preset time points into a temperature input vector according to a time dimension and then obtaining a temperature characteristic vector by using a second convolution neural network model of a one-dimensional convolution kernel; the classification characteristic matrix acquisition module is used for calculating the responsiveness estimation of the reaction liquid state context semantic characteristic vector relative to the temperature characteristic vector based on a Gaussian density map to obtain a classification characteristic matrix; the characteristic distribution optimization module is used for carrying out vector granularity characteristic distribution optimization on the classification characteristic matrix based on the reaction liquid state context semantic characteristic vector and the temperature characteristic vector to obtain an optimized classification characteristic matrix; and the classification module is used for enabling the optimized classification characteristic matrix to pass through a classifier to obtain a classification result, and the classification result is used for indicating that the reaction temperature value of the current time point should be increased or decreased.
CN202310009378.7A 2023-01-03 2023-01-03 Intelligent production system and method of isomaltooligosaccharide iron complex Withdrawn CN115960144A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310009378.7A CN115960144A (en) 2023-01-03 2023-01-03 Intelligent production system and method of isomaltooligosaccharide iron complex

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310009378.7A CN115960144A (en) 2023-01-03 2023-01-03 Intelligent production system and method of isomaltooligosaccharide iron complex

Publications (1)

Publication Number Publication Date
CN115960144A true CN115960144A (en) 2023-04-14

Family

ID=87359805

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310009378.7A Withdrawn CN115960144A (en) 2023-01-03 2023-01-03 Intelligent production system and method of isomaltooligosaccharide iron complex

Country Status (1)

Country Link
CN (1) CN115960144A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116285481A (en) * 2023-05-23 2023-06-23 佛山市时力涂料科技有限公司 Method and system for producing and processing paint
CN116933941A (en) * 2023-07-27 2023-10-24 郑州软通合力计算机技术有限公司 Intelligent supply chain logistics intelligent optimization method, system and storage medium
CN117037028A (en) * 2023-07-31 2023-11-10 邵阳阳光发品有限公司 Intelligent wig preparation method and system

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116285481A (en) * 2023-05-23 2023-06-23 佛山市时力涂料科技有限公司 Method and system for producing and processing paint
CN116933941A (en) * 2023-07-27 2023-10-24 郑州软通合力计算机技术有限公司 Intelligent supply chain logistics intelligent optimization method, system and storage medium
CN117037028A (en) * 2023-07-31 2023-11-10 邵阳阳光发品有限公司 Intelligent wig preparation method and system

Similar Documents

Publication Publication Date Title
CN115960144A (en) Intelligent production system and method of isomaltooligosaccharide iron complex
CN115896817B (en) Production method and system of fluorine-nitrogen mixed gas
CN116052050A (en) Intelligent production system and method for detergent
CN116048031B (en) Control system and method for petroleum auxiliary production
CN115983126A (en) Intelligent preparation method of iron isomaltose anhydride 1000
Sun et al. A discriminatively deep fusion approach with improved conditional GAN (im-cGAN) for facial expression recognition
CN107977661A (en) The region of interest area detecting method decomposed based on full convolutional neural networks and low-rank sparse
CN116434117A (en) Preparation method of composite polyacrylamide oil displacement agent
CN115723287B (en) Production equipment for controlling polyurethane foaming based on temperature
CN115998752B (en) Preparation method of vitamin D preparation containing linoleic acid vegetable oil
CN116229113A (en) Preparation method of selenium-enriched dendrobium flower wine
CN116385945B (en) Video interaction action detection method and system based on random frame complement and attention
CN109062811B (en) Test case generation method based on neural style migration
CN113420794A (en) Binaryzation Faster R-CNN citrus disease and pest identification method based on deep learning
CN116164497A (en) Rectification control system and method in liquid oxygen preparation process
Pan et al. A Novel Combinational Convolutional Neural Network for Automatic Food-Ingredient Classification.
CN116149280A (en) Intelligent production system of electronic-grade potassium hydroxide
CN115984339A (en) Double-pipeline point cloud completion method based on geometric feature refining and confrontation generation network
CN116649191A (en) Remote fertilization and irrigation control system and method based on PLC
CN116059681B (en) Ultrasonic control method and system for controlling crystallization nucleation
CN116837058A (en) Plum blossom small molecule polypeptide preparation method and intelligent system thereof
CN115019397A (en) Comparison self-monitoring human behavior recognition method and system based on temporal-spatial information aggregation
Wang et al. Research and implementation of image dehazing based on deep learning
CN109215057B (en) High-performance visual tracking method and device
CN113449259A (en) Channel state information feature extraction method and system based on deep learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20230414

WW01 Invention patent application withdrawn after publication