Intelligent control system and method for production of sildenafil citrate orodisintegrating tablets

Through deep learning technology, the correlation between mannitol temperature and humidity data and the speed of the dry whole pelletizer is analyzed, and the rotation speed is adjusted in real time, which solves the misjudgment and time-consuming problem of manual speed control, and improves the oral collapse of sildenafil citrate. Whole particle effect and finished product quality.

CN117872967BActive Publication Date: 2024-11-22JINGSHI (HANGZHOU) PHARM CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311654750.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-05
Publication Date
2024-11-22
Estimated Expiration
2043-12-05

AI Technical Summary

Technical Problem

In the current production of sildenafil citrate oral collapse tablets, the control of the speed of the dry whole pellet machine depends on manual judgment, and there are problems of misjudgment and time-consuming, which affects the effect of the whole pellet and the quality of the particle.

Method used

Using artificial intelligence technology based on deep learning, we analyze the responsive synergistic relationship between the correlation changes of mannitol temperature and humidity data and the speed of the dry whole granulator, and adjust the speed value in real time.

Benefits of technology

The whole particle effect and particle quality of mannitol are improved, the quality of finished products of sildenafil citrate oral collapse tablets is enhanced, and the time of manual adjustment and the risk of misjudgment are reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117872967B_ABST
    Figure CN117872967B_ABST
Patent Text Reader

Abstract

The present application relates to the field of intelligent control technology, and specifically discloses an intelligent control system and method for the production of sildenafil citrate orodisintegrating tablets, which utilizes artificial intelligence technology based on deep learning to analyze the correlation changes of mannitol temperature data and mannitol humidity data during the granulation process, as well as the responsive synergistic correlation between the correlation changes and the rotation speed of the dry granulator, thereby controlling the real-time adjustment of the rotation speed value of the dry granulator. In this way, better mannitol granulation effect and particle quality can be obtained, thereby improving the finished product quality of sildenafil citrate orodisintegrating tablets.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent control technology, and more specifically, to an intelligent control system and method for producing sildenafil citrate orodisintegrating tablets. Background Art

[0002] Sildenafil citrate orodisintegrating tablets are male medicines, mainly used for erectile dysfunction (ED) in adult men. Sildenafil citrate orodisintegrating tablets were first developed by Pfizer Inc. and marketed in the United States. The original formulation ingredients are mainly: mannitol, microcrystalline cellulose, povidone, polyvinyl acetate, cross-linked povidone, cross-linked sodium carboxymethyl cellulose, silicon dioxide, magnesium stearate, sucralose, flavor, indigo aluminum lake and sildenafil citrate. The results of stability studies show that the original formulation has a significant increase in related substances under high temperature 60°C, high humidity 75% RH and light, which affects the safety of clinical use.

[0003] In this regard, patent CN115813866A discloses a sildenafil citrate orodisintegrating tablet and its production method, which omits polyvinyl acetate in the original formulation ingredients, and can solve the problems of disintegration time limit and related substances. In the actual production and preparation process, more specifically, in the process of granulating mannitol using a dry granulator, it is necessary to control the rotation speed of the dry granulator. Usually, the rotation speed of the dry granulator is controlled by manual judgment, and this implementation method has some problems. First, manual judgment needs to rely on the operator's experience and skills, and there may be misjudgment problems; secondly, manual judgment requires continuous observation and adjustment of the operating status of the dry granulator, which consumes time and manpower.

[0004] Therefore, an optimized control system and method for producing sildenafil citrate orodisintegrating tablets is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent control system and method for the production of sildenafil citrate orodisintegrating tablets, which uses artificial intelligence technology based on deep learning to analyze the correlation changes of mannitol temperature data and mannitol humidity data during the granulation process, and the responsive synergistic correlation between this correlation change and the rotation speed of the dry granulator, thereby controlling the real-time adjustment of the rotation speed value of the dry granulator. In this way, better mannitol granulation effect and particle quality can be obtained, thereby improving the finished product quality of sildenafil citrate orodisintegrating tablets.

[0006] Accordingly, according to one aspect of the present application, an intelligent control method for producing sildenafil citrate orodisintegrating tablets is provided, comprising: granulating mannitol using a dry granulator; adding cross-linked polyvinylpyrrolidone, and spraying the cross-linked polyvinylpyrrolidone and a binder prepared with purified water into the dry granulator for mixing, granulating and drying to obtain prefabricated granules, and granulating to obtain prefabricated granules after granulation; adding sucralose, indigo aluminum lake and silicon dioxide into a mixer for mixing and granulating to obtain mixed granules; adding the prefabricated granules after granulation, the mixed granules, essence, cross-linked sodium carboxymethyl cellulose, magnesium stearate, microcrystalline cellulose and sildenafil citrate into the mixer in sequence for mixing to obtain mixed granules; and tableting and inner packaging the mixed granules to obtain sildenafil citrate orodisintegrating tablets; wherein granulating mannitol using a dry granulator comprises:

[0007] Obtaining mannitol temperature values, mannitol humidity values, and dry granulator speed values ​​at a plurality of predetermined time points within a predetermined time period;

[0008] Performing correlation analysis on the mannitol temperature values ​​and the mannitol humidity values ​​at a plurality of predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors;

[0009] Performing time series analysis on the rotation speed values ​​of the dry granulator at the plurality of predetermined time points to obtain a rotation speed time series feature vector;

[0010] The speed control strategy at the current time point is determined based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors.

[0011] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, the mannitol temperature values ​​and the mannitol humidity values ​​at multiple predetermined time points within the predetermined time period are subjected to correlation analysis to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors, including: performing data preprocessing on the mannitol temperature values ​​and the mannitol humidity values ​​at the multiple predetermined time points to obtain a mannitol temperature time series input vector and a mannitol humidity time series input vector; associating and encoding the mannitol temperature time series input vector and the mannitol humidity time series input vector into a mannitol temperature time series input vector. Mannitol temperature-humidity time series correlation matrix; using a deep learning network model to extract features from the mannitol temperature-humidity time series correlation matrix to obtain a mannitol temperature-humidity time series correlation feature graph; performing feature distribution correction on the mannitol temperature-humidity time series correlation feature graph to obtain a corrected mannitol temperature-humidity time series correlation feature graph; and flattening the features of each feature matrix along the channel dimension of the corrected mannitol temperature-humidity time series correlation feature graph to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors.

[0012] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, data preprocessing is performed on the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points to obtain the mannitol temperature timing input vector and the mannitol humidity timing input vector, including: arranging the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points according to the time dimension to obtain the mannitol temperature timing input vector and the mannitol humidity timing input vector.

[0013] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, the deep learning network model is a temperature-humidity time series correlation feature extractor based on a convolutional neural network model; wherein, the temperature-humidity time series correlation feature extractor based on a convolutional neural network model includes an input layer, a two-dimensional convolution layer, an activation function layer, a two-dimensional pooling layer and an output layer.

[0014] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, a deep learning network model is used to perform feature extraction on the mannitol temperature-humidity time series association matrix to obtain a mannitol temperature-humidity time series association feature graph, including: passing the mannitol temperature-humidity time series association matrix through the temperature-humidity time series association feature extractor based on the convolutional neural network model to obtain the mannitol temperature-humidity time series association feature graph.

[0015] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, a time series analysis is performed on the speed values ​​of the dry granulator at the plurality of predetermined time points to obtain a speed time series feature vector, including: arranging the speed values ​​of the dry granulator at the plurality of predetermined time points into a speed time series input vector according to the time dimension; and passing the speed time series input vector through a speed time series feature extractor based on a one-dimensional convolutional layer to obtain the speed time series feature vector.

[0016] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, the speed control strategy at the current time point is determined based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed timing feature vector, including: calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed timing feature vector to obtain a responsive fused speed timing feature vector; passing the responsive fused speed timing feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the speed value at the current time point should be increased, decreased or maintained.

[0017] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain the responsiveness fused speed time series feature vector, including: calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector using the following responsiveness information fusion formula to obtain the responsiveness fused speed time series feature vector; wherein the responsiveness information fusion formula is:

[0018]

[0019]

[0020] Where A is 1×N w The matrix, N w is the dimension of the speed time series feature vector, V is the speed time series feature vector, and B is 1×N h The matrix, N h is the dimension of each of the local feature vectors of the mannitol temperature-humidity time series association, N is the number of the local feature vectors of the mannitol temperature-humidity time series association, h i is the ith mannitol temperature-humidity time series associated local feature vector in the sequence of the mannitol temperature-humidity time series associated local feature vectors, σ is the Sigmoid function, V' is the responsive fusion speed time series feature vector, M w and M h Represents the convolution operation of a 1×1 convolution kernel, and S is the weight value.

[0021] According to another aspect of the present application, an intelligent control system for the production of sildenafil citrate orodisintegrating tablets is provided, comprising:

[0022] A mannitol granulation parameter monitoring module is used to obtain the mannitol temperature value, mannitol humidity value and dry granulation machine speed value at multiple predetermined time points within a predetermined time period;

[0023] A mannitol temperature and humidity correlation analysis module, used for performing correlation analysis on the mannitol temperature values ​​and mannitol humidity values ​​at a plurality of predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors;

[0024] A dry granulator speed time series analysis module, used for performing time series analysis on the dry granulator speed values ​​at the plurality of predetermined time points to obtain a speed time series feature vector;

[0025] The speed control strategy generation module is used to determine the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors.

[0026] Compared with the prior art, the intelligent control system and method for producing sildenafil citrate orodisintegrating tablets provided in the present application utilizes artificial intelligence technology based on deep learning to analyze the correlation changes of mannitol temperature data and mannitol humidity data during the granulation process, as well as the responsive synergistic correlation between such correlation changes and the rotation speed of the dry granulator, thereby controlling the real-time adjustment of the rotation speed value of the dry granulator. In this way, better mannitol granulation effect and particle quality can be obtained, thereby improving the finished product quality of sildenafil citrate orodisintegrating tablets. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 The present invention is a flow chart of an intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application.

[0029] Figure 2 The figure is a schematic diagram of the structure of the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application.

[0030] Figure 3 The present invention is a flowchart of performing correlation analysis on the mannitol temperature value and the mannitol humidity value at multiple predetermined time points within the predetermined time period in the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors.

[0031] Figure 4 The present invention is a flow chart of performing time series analysis on the rotation speed values ​​of the dry granulator at the plurality of predetermined time points in the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application to obtain a rotation speed time series characteristic vector.

[0032] Figure 5 The present invention is a flowchart of determining the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors in the intelligent control method for the production of sildenafil citrate orodisintegrating tablets according to an embodiment of the present application.

[0033] Figure 6 The block diagram is a smart control system for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0035] Figure 1 The present invention is a flow chart of an intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the structure of the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the intelligent control method for the production of sildenafil citrate orodisintegrating tablets includes the following steps: S1, obtaining the mannitol temperature value, mannitol humidity value and dry granulator speed value at multiple predetermined time points within a predetermined time period; S2, performing correlation analysis on the mannitol temperature value and the mannitol humidity value at multiple predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors; S3, performing time series analysis on the dry granulator speed values ​​at the multiple predetermined time points to obtain a speed time series feature vector; S4, determining the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series correlation local feature vectors and the speed time series feature vector.

[0036] As mentioned in the background technology, patent CN115813866A discloses a method for producing a sildenafil citrate orodisintegrating tablet, and the specific steps include: granulating mannitol with a dry granulator; adding cross-linked polyvinylpyrrolidone, and spraying the cross-linked polyvinylpyrrolidone and a binder prepared with purified water into the dry granulator for mixing, granulating and drying to obtain prefabricated granules, and granulating to obtain prefabricated granules after granulation; adding sucralose, indigo aluminum lake and silicon dioxide into a mixer for mixing and granulating to obtain mixed granules; adding the prefabricated granules after granulation, the mixed granules, essence, cross-linked sodium carboxymethyl cellulose, magnesium stearate, microcrystalline cellulose and sildenafil citrate into the mixer in sequence for mixing to obtain mixed granules; and tableting and inner packaging the mixed granules to obtain sildenafil citrate orodisintegrating tablets.

[0037] In the actual production and preparation process, more specifically, in the process of granulating mannitol using a dry granulator, the rotation speed of the dry granulator needs to be controlled. Because the rotation speed of the dry granulator will affect the quality and efficiency of the granulation. If the rotation speed is too high, the mannitol particles will be too small, which is not conducive to subsequent granulation and tableting; if the rotation speed is too low, the mannitol particles will be too large and uneven, which is easy to cause agglomeration and dust. Therefore, it is necessary to select a suitable rotation speed value during the granulation of the dry granulator to ensure the quality and efficiency of the granulation.

[0038] Usually, the speed of the dry granulator is controlled by manual judgment, and there are some problems with this implementation method. First, manual judgment needs to rely on the experience and skills of the operator, and there may be misjudgment problems; secondly, manual judgment requires continuous observation and adjustment of the operating status of the dry granulator, which consumes time and manpower. In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to analyze the correlation changes of mannitol temperature data and mannitol humidity data during the granulation process, as well as the responsive synergistic correlation between this correlation change and the speed of the dry granulator, so as to control the real-time adjustment of the speed value of the dry granulator. In this way, better mannitol granulation effect and particle quality can be obtained, thereby improving the finished product quality of sildenafil citrate orodisintegrating tablets.

[0039] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, the step S1 is to obtain the mannitol temperature value, mannitol humidity value and dry granulator speed value at multiple predetermined time points within a predetermined time period. Here, the mannitol temperature value and mannitol humidity value at multiple predetermined time points can reflect the real-time physical state change of mannitol during the granulation process. And the speed value at multiple predetermined time points can reflect the operating state change of the dry granulator. In an embodiment of the present application, the mannitol temperature value and mannitol humidity value at multiple predetermined time points within a predetermined time period are respectively collected by a temperature sensor and a humidity sensor deployed inside the granulation chamber of the dry granulator, and the speed value of the dry granulator speed at multiple predetermined time points within the predetermined time period is collected by a speed sensor.

[0040] In the above-mentioned intelligent control method for producing sildenafil citrate orodisintegrating tablets, the step S2 is to perform correlation analysis on the mannitol temperature value and mannitol humidity value at multiple predetermined time points in the predetermined time period to obtain a sequence of mannitol temperature-humidity time series associated local feature vectors. It should be understood that temperature and humidity are important factors affecting the morphology and size of mannitol particles. The mannitol particles of high temperature and high humidity have higher adhesion, and the physical state is softer and more viscous, while the mannitol particles of low temperature and low humidity become fragile and dry, making the particles easy to separate during the whole granulation process. That is, the temperature value and humidity value of mannitol are closely related to the physical state of mannitol. During the whole granulation process, the change of the temperature and humidity of mannitol can reflect the aggregation, dispersion or morphological change of particles, thereby expressing the whole granulation effect of the dry granulator. Therefore, it is very important to grasp the change of the temperature and humidity of mannitol for obtaining an ideal particle size. And, there is a correlation relationship of mutual influence and interaction in the time dimension of mannitol temperature and humidity data. Specifically, usually when the temperature of mannitol increases, the humidity of mannitol decreases, and vice versa. Therefore, in order to better understand the overall correlation pattern between temperature and humidity, the temperature and humidity data of mannitol are further correlated and analyzed to obtain the sequence of local feature vectors of mannitol temperature-humidity time series correlation, which is used to provide guidance when adjusting the speed of the dry granulator.

[0041] Figure 3 The present invention is a flowchart of performing correlation analysis on the mannitol temperature value and the mannitol humidity value at multiple predetermined time points within the predetermined time period in the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to the embodiment of the present application to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors. Figure 3 As shown, the step S2 includes: S21, performing data preprocessing on the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points to obtain a mannitol temperature time series input vector and a mannitol humidity time series input vector; S22, associating and encoding the mannitol temperature time series input vector and the mannitol humidity time series input vector into a mannitol temperature-humidity time series association matrix; S23, using a deep learning network model to perform feature extraction on the mannitol temperature-humidity time series association matrix to obtain a mannitol temperature-humidity time series association feature graph; S24, performing feature distribution correction on the mannitol temperature-humidity time series association feature graph to obtain a corrected mannitol temperature-humidity time series association feature graph; S25, performing feature flattening on each feature matrix along the channel dimension of the corrected mannitol temperature-humidity time series association feature graph to obtain a sequence of the mannitol temperature-humidity time series association local feature vectors.

[0042] Specifically, step S21, data preprocessing is performed on the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points to obtain a mannitol temperature timing input vector and a mannitol humidity timing input vector. In a specific example of the present application, data preprocessing is performed on the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points to obtain a mannitol temperature timing input vector and a mannitol humidity timing input vector by arranging the mannitol temperature values ​​and mannitol humidity values ​​at the multiple predetermined time points according to the time dimension to obtain the mannitol temperature timing input vector and the mannitol humidity timing input vector. It should be understandable that arranging the temperature data and humidity data of mannitol according to the time dimension can retain the time sequence information in the data, ensure that the model takes into account the time sequence of data points when analyzing the data, and ensure that the time correlation of the temperature and humidity data of mannitol is reasonably analyzed, thereby more accurately capturing the temporal correlation between the temperature and humidity of mannitol, better understanding and predicting the correlation between the temperature and humidity data, and better describing the temporal correlation changes between the temperature and humidity of mannitol, which is conducive to the real-time adjustment of the rotation speed of the dry granulator, thereby improving the granulation effect and particle quality.

[0043] Particularly, described step S22, is associatedly coded as mannitol temperature-humidity time series association matrix with described mannitol temperature time series input vector and described mannitol humidity time series input vector.Should be appreciated that mannitol temperature and humidity data have the association relation of mutual influence, interaction in time dimension.Therefore, if mannitol temperature and humidity are processed as independent input vector respectively, may ignore the interaction between them, cause feature extraction insufficient.Therefore, further mannitol temperature and humidity are associatedly coded, so that two kinds of time series data are integrated into a unified structure, thereby obtain mannitol temperature-humidity time series association matrix.Like this, when retaining the time series relation between mannitol temperature and humidity, so that the relation between them is explicitly represented, make model can better understand the interaction and the change pattern between the two. The mannitol temperature-humidity time series association matrix provides a two-dimensional data representation, which can characterize the relationship between temperature and humidity in a global scope, rather than being limited to the value at a single time point. This is conducive to using deep learning models to extract spatial features from it, such as simultaneous changes in temperature and humidity, or sequential changes, etc., thereby helping to capture richer feature representations of temperature and humidity data and grasp the time series association pattern between temperature and humidity data from a holistic perspective.

[0044] Specifically, step S23, utilizes deep learning network model to carry out feature extraction to obtain mannitol temperature-humidity time series correlation matrix to obtain mannitol temperature-humidity time series correlation characteristic graph.In a specific example of the present application, the deep learning network model is the temperature-humidity time series correlation feature extractor based on convolutional neural network model.That is, the mannitol temperature-humidity time series correlation matrix is ​​passed through the temperature-humidity time series correlation feature extractor based on convolutional neural network model to obtain the mannitol temperature-humidity time series correlation characteristic graph.One of ordinary skill in the art should understand that convolutional neural network model is good at capturing spatial features in data, using the mannitol temperature-humidity time series correlation matrix of two-dimensional structure as input, convolutional neural network can automatically learn local neighborhood features in data, extract the spatial correlation features between temperature and humidity, and better capture the complex correlation between mannitol temperature and humidity data by multi-layer nonlinear transformation to realize the abstraction and combination of features. In the technical scheme of the present application, the convolutional neural network model can effectively capture the local features in the mannitol temperature-humidity temporal association matrix, such as the amplitude, frequency, direction etc. of temperature and humidity changes, thereby reflecting the trend and fluctuation mode of temperature and humidity changes, and reflecting the state change of mannitol in the whole granulation process. In addition, the convolutional neural network model reduces the number of parameters that need to be learned by means of parameter sharing, which makes the network more efficient and can process larger input data. In the temporal association feature extraction process of mannitol temperature and humidity, the parameter sharing mechanism can effectively utilize the local correlation of data, effectively obtain the characteristic information between temperature and humidity from the mannitol temperature-humidity temporal association matrix, and provide more abundant and specific feature representation for subsequent prediction and analysis. In a specific example, the temperature-humidity temporal association feature extractor based on the convolutional neural network model comprises an input layer, a two-dimensional convolutional layer, an activation function layer, a two-dimensional pooling layer and an output layer.

[0045] Specifically, described step S24, carries out characteristic distribution correction to described mannitol temperature-humidity time series correlation characteristic diagram to obtain the corrected mannitol temperature-humidity time series correlation characteristic diagram.In the above technical scheme, each characteristic matrix of described mannitol temperature-humidity time series correlation characteristic diagram expresses the high-order local time domain correlation feature of the full time domain source data association of described mannitol temperature value and mannitol humidity value, and each characteristic matrix follows the channel distribution of described convolutional neural network model, that is, each characteristic value of described mannitol temperature-humidity time series correlation characteristic diagram has the spatial information attribute based on predetermined dimension constraint.Like this, if can improve described mannitol temperature-humidity time series correlation characteristic diagram as the spatial information expression effect of high dimensional feature, then can improve described mannitol temperature-humidity time series correlation characteristic diagram expression effect, and then improve described mannitol temperature-humidity time series correlation local feature vector expression effect of sequence.Based on this, in the technical scheme of the present application, described mannitol temperature-humidity time series correlation characteristic diagram F is optimized.

[0046] In a specific example, the step S24 includes: performing characteristic distribution correction on the mannitol temperature-humidity time series correlation characteristic diagram using the following characteristic distribution correction formula to obtain a corrected mannitol temperature-humidity time series correlation characteristic diagram; wherein the characteristic distribution correction formula is:

[0047]

[0048]

[0049] Among them, f i,j,k ∈F,f i,j,k is the eigenvalue of the (i, j, k)th position in the mannitol temperature-humidity time series correlation feature map, N is the scale of the local neighborhood of the mannitol temperature-humidity time series correlation feature map, and α is the local space segmentation coefficient, f' i,j,k It is the (i, j, k)th eigenvalue in the corrected mannitol temperature-humidity time series correlation characteristic diagram.

[0050] Specifically, taking the local segmentation space in the Hilbert space after the mannitol temperature-humidity time series correlation feature graph F is expanded as a reference, the characteristic manifold of the mannitol temperature-humidity time series correlation feature graph F in the high-dimensional feature space is locally integrated on the surface, thereby correcting the phase change discontinuity of the characteristic manifold expressed by the non-stationary data sequence after the local space expansion of the mannitol temperature-humidity time series correlation feature graph F based on the local integration processing of the integral function, thereby obtaining a finer structure and geometric features of the characteristic manifold, improving the spatial information expression effect of the mannitol temperature-humidity time series correlation feature graph F in the high-dimensional feature space, and thus improving the accuracy of the subsequent dry granulator speed adjustment decision.

[0051] Specifically, described step S25, each characteristic matrix along the channel dimension of the mannitol temperature-humidity time series associated characteristic graph after described correction is carried out feature flattening to obtain the sequence of the local characteristic vector of the mannitol temperature-humidity time series associated local characteristic vector.Should be understood that the flattening operation is a kind of dimensionality reduction process, which can reduce the complexity of data, extract the most important characteristic information, thereby reduce the processing pressure of the model for a large amount of data, help simplify the complexity and the amount of calculation of the model, improve the training efficiency of the model.By carrying out feature flattening along each characteristic matrix of the channel dimension of the mannitol temperature-humidity time series associated characteristic graph after described correction, three-dimensional characteristic graph is converted into a one-dimensional vector sequence, be conducive to local feature being input in the next model in a more direct way, better match with subsequent model structure, realize the alignment of characteristic dimension.That is, dimensionality reduction process and dimension alignment process are carried out to the mannitol temperature-humidity time series associated characteristic graph after described correction, so that subsequent fusion and feature interaction.

[0052] In the above-mentioned intelligent control method for producing sildenafil citrate orodisintegrating tablets, in step S3, the speed values ​​of the dry granulator at the plurality of predetermined time points are subjected to time series analysis to obtain a speed time series feature vector. In order to capture the speed change pattern contained in the speed values ​​of the dry granulator at the plurality of predetermined time points to reflect the time series change of the operating state of the dry granulator, the speed values ​​of the dry granulator at the plurality of predetermined time points are further subjected to time series feature extraction.

[0053] Figure 4 The present invention is a flow chart of performing time series analysis on the rotation speed values ​​of the dry granulator at the plurality of predetermined time points in the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to the embodiment of the present application to obtain the rotation speed time series feature vector. Figure 4 As shown, the step S3 includes: S31, arranging the speed values ​​of the dry granulator at the multiple predetermined time points into a speed timing input vector according to the time dimension; S32, passing the speed timing input vector through a speed timing feature extractor based on a one-dimensional convolutional layer to obtain the speed timing feature vector.

[0054] Specifically, in step S31, the rotation speed values ​​of the dry granulator at the plurality of predetermined time points are arranged according to the time dimension as a rotation speed time series input vector. It should be understood that by arranging the rotation speed values ​​at the plurality of predetermined time points according to the time dimension to retain the time series information of the data, it is helpful for the model to better understand and capture the changing trend of the rotation speed value over time, reflecting the time series change of the operating state of the dry granulator. In addition, the rotation speed time series input vector retains the temporal correlation of the rotation speed value, so that the model can better capture the time dependency and pattern change between the rotation speed values, which helps to identify the changing characteristics of the rotation speed in different time periods, and provides a more time-series data form for subsequent feature extraction and pattern recognition.

[0055] Specifically, in step S32, the speed time series input vector is passed through a speed time series feature extractor based on a one-dimensional convolutional layer to obtain the speed time series feature vector. It should be understood that the one-dimensional convolutional layer is naturally adapted to time series data and can better process data with time series properties. It perceives features on different time scales with different receptive fields through a sliding window. Through the one-dimensional convolution operation, the time series data of the speed of the dry granulator can be reduced in dimension and a more abstract feature representation can be extracted, the local patterns and features in the time series data can be captured, and the speed change characteristics within different time ranges can be identified, which helps to reduce the complexity of the data, extract the key features that reflect the changes in the machine's operating state, and provide a more discriminative feature representation for subsequent state recognition and classification. In addition, the one-dimensional convolutional layer has the characteristic of parameter sharing, which means that the same convolution kernel can be used for feature extraction in different time intervals, so that the model can more efficiently utilize the characteristics of the data during the learning process, reduce the number of parameters, and reduce the risk of overfitting.

[0056] In the above-mentioned intelligent control method for the production of sildenafil citrate orodisintegrating tablets, the step S4 determines the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors. That is, the influence of the speed time series change on the associated change of mannitol temperature and humidity is characterized by the responsiveness information fusion method. In this way, the responsiveness and influence of the speed change mode of the dry granulator on the change of the physical state of mannitol are manifested and characterized, guiding the model to learn and judge whether the current speed value needs to be adjusted and how to adjust it.

[0057] Figure 5 The present invention is a flowchart of determining the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series correlation local feature vector and the speed time series feature vector in the intelligent control method for producing sildenafil citrate orodisintegrating tablets according to the embodiment of the present application. Figure 5 As shown, the step S4 includes: S41, calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain a responsive fused speed time series feature vector; S42, passing the responsive fused speed time series feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the speed value at the current time point should increase, decrease or maintain.

[0058] Specifically, the step S41 calculates the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain a responsive fused speed time series feature vector. It should be understood that by calculating the responsiveness information fusion of the mannitol temperature-humidity time series associated local feature vector and the speed time series feature vector, the synergistic correlation between the mannitol temperature-humidity time series associated changes and the speed of the dry granulator can be revealed, and the correlation pattern between the mannitol temperature-humidity changes and the speed response can be captured, which is helpful to discover the operating laws and characteristics under different working conditions, and enhance the model's prediction and control capabilities for the speed response, so as to achieve real-time adjustment of the speed value of the dry granulator.

[0059] In a specific example of the present application, the step S41 includes: calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector using the following responsiveness information fusion formula to obtain the responsive fused speed time series feature vector; wherein the responsiveness information fusion formula is:

[0060]

[0061]

[0062] Where A is 1×N w The matrix, N w is the dimension of the speed time series feature vector, V is the speed time series feature vector, and B is 1×N h The matrix, N h is the dimension of each of the local feature vectors of the mannitol temperature-humidity time series association, N is the number of the local feature vectors of the mannitol temperature-humidity time series association, h i is the ith mannitol temperature-humidity time series associated local feature vector in the sequence of the mannitol temperature-humidity time series associated local feature vectors, σ is the Sigmoid function, V' is the responsive fusion speed time series feature vector, M w and M h Represents the convolution operation of a 1×1 convolution kernel, and S is the weight value.

[0063] Here, the responsive information fusion can dynamically assign weights based on the correlation between each mannitol temperature-humidity time series associated local feature vector and the speed time series feature vector, so that the important feature distribution in the responsive fused speed time series feature vector can receive greater attention.

[0064] Specifically, in step S42, the responsive fusion speed time series feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the speed value at the current time point should be increased, decreased or maintained. It should be understood that the classifier is a machine learning model that can learn the characteristic pattern of the input data and map it to the corresponding label according to the learned characteristics. In the technical solution of the present application, the classifier is used to output the real-time adjustment decision of the speed of the dry granulator. By inputting the responsive fusion speed time series feature vector into the classifier, the classifier can learn the characteristic pattern in the responsive fusion speed time series feature vector and map it to the corresponding label, automatically judging whether the speed should be increased, decreased or kept unchanged in the current state, thereby realizing the automatic adjustment of the speed and reducing the burden of manual adjustment. In addition, the classifier can quickly judge the changing trend of the speed based on the input characteristics at the current moment, realize the rapid response to the speed change, support the real-time adjustment of the speed, so as to adapt to the requirements under the current working conditions, thereby optimizing the efficiency and accuracy of the speed adjustment of the granulator.

[0065] In summary, the intelligent control method for the production of sildenafil citrate orodisintegrating tablets according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to analyze the correlation changes of mannitol temperature data and mannitol humidity data during the granulation process, and the responsive synergistic correlation between this correlation change and the rotation speed of the dry granulator, thereby controlling the real-time adjustment of the rotation speed value of the dry granulator. In this way, better mannitol granulation effect and particle quality can be obtained, thereby improving the finished product quality of sildenafil citrate orodisintegrating tablets.

[0066] Figure 6 FIG. 1 is a block diagram of an intelligent control system for producing sildenafil citrate orodisintegrating tablets according to an embodiment of the present application. Figure 6 As shown, according to the embodiment of the present application, the intelligent control system 100 for the production of sildenafil citrate orodisintegrating tablets includes: a mannitol granulation parameter monitoring module 110, which is used to obtain the mannitol temperature value, mannitol humidity value and dry granulator speed value at multiple predetermined time points within a predetermined time period; a mannitol temperature and humidity correlation analysis module 120, which is used to perform correlation analysis on the mannitol temperature value and the mannitol humidity value at multiple predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors; a dry granulator speed timing analysis module 130, which is used to perform time series analysis on the dry granulator speed values ​​at the multiple predetermined time points to obtain a speed timing feature vector; a speed control strategy generation module 140, which is used to determine the speed control strategy at the current time point based on the responsiveness fusion feature between the sequence of the mannitol temperature-humidity time series correlation local feature vectors and the speed timing feature vector.

[0067] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned intelligent control system for the production of sildenafil citrate orodisintegrating tablets have been described in the above reference. Figures 1 to 5 The intelligent control method for producing sildenafil citrate orodisintegrating tablets has been described in detail, and therefore, its repeated description will be omitted.

[0068] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0069] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0070] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0071] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An intelligent control method for the production of sildenafil citrate orodisintegrating tablets, comprising: The mannitol is granulated by using a dry granulator; Adding cross-linked polyvinylpyrrolidone, spraying the cross-linked polyvinylpyrrolidone and the binder prepared with purified water into the dry granulator for mixing, granulating and drying to obtain prefabricated granules, and granulating to obtain prefabricated granules after granulation; adding sucralose, indigo aluminum lake and silicon dioxide into the mixer for mixing and granulating to obtain mixed granules; adding the prefabricated granules after granulation, the mixed granules, flavor, cross-linked sodium carboxymethyl cellulose, magnesium stearate, microcrystalline cellulose and sildenafil citrate into the mixer in sequence for mixing to obtain mixed granules; and tableting and inner packaging the mixed granules to obtain sildenafil citrate orally disintegrating tablets, characterized in that mannitol is granulated by using a dry granulator, comprising: Obtaining mannitol temperature values, mannitol humidity values, and dry granulator speed values ​​at a plurality of predetermined time points within a predetermined time period; Performing correlation analysis on the mannitol temperature values ​​and the mannitol humidity values ​​at a plurality of predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors; Performing time series analysis on the rotation speed values ​​of the dry granulator at the plurality of predetermined time points to obtain a rotation speed time series feature vector; Determine the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vector; Among them, the correlation analysis is performed on the mannitol temperature values ​​and the mannitol humidity values ​​at multiple predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors, including: Performing data preprocessing on the mannitol temperature values ​​and the mannitol humidity values ​​at the plurality of predetermined time points to obtain a mannitol temperature time series input vector and a mannitol humidity time series input vector; Associatively encoding the mannitol temperature time series input vector and the mannitol humidity time series input vector into a mannitol temperature-humidity time series association matrix; Using a deep learning network model to extract features from the mannitol temperature-humidity time series association matrix to obtain a mannitol temperature-humidity time series association feature graph; Performing characteristic distribution correction on the mannitol temperature-humidity time series correlation characteristic diagram to obtain a corrected mannitol temperature-humidity time series correlation characteristic diagram; Flatten the features of each feature matrix along the channel dimension of the corrected mannitol temperature-humidity time series correlation feature graph to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors; Wherein, determining the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors includes: Calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain a responsive fused speed time series feature vector; Passing the responsive fused speed time series feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the speed value at the current time point should be increased, decreased, or maintained; The step of calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain the responsive fused speed time series feature vector comprises: calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector using the following responsiveness information fusion formula to obtain the responsive fused speed time series feature vector; Among them, the responsiveness information fusion formula is: ;in, yes The matrix of is the dimension of the speed time series feature vector, is the speed time series characteristic vector, yes The matrix of is the dimension of each of the local feature vectors of the mannitol temperature-humidity time series association, is the number of local eigenvectors of the mannitol temperature-humidity time series association, is the number of local feature vectors in the sequence of mannitol temperature-humidity time series association Mannitol temperature-humidity time series correlation local feature vectors, is the Sigmoid function, is the responsive fusion speed time series feature vector, and represents the convolution operation of a 1×1 convolution kernel, is the weight value.

2. The intelligent control method for producing sildenafil citrate orodisintegrating tablets according to claim 1, characterized in that: Performing data preprocessing on the mannitol temperature values ​​and the mannitol humidity values ​​at the plurality of predetermined time points to obtain a mannitol temperature time series input vector and a mannitol humidity time series input vector, comprising: The mannitol temperature values ​​and the mannitol humidity values ​​at the plurality of predetermined time points are arranged respectively according to the time dimension to obtain the mannitol temperature time series input vector and the mannitol humidity time series input vector.

3. The intelligent control method for producing sildenafil citrate orodisintegrating tablets according to claim 2, characterized in that: The deep learning network model is a temperature-humidity time series correlation feature extractor based on a convolutional neural network model; wherein the temperature-humidity time series correlation feature extractor based on a convolutional neural network model includes an input layer, a two-dimensional convolution layer, an activation function layer, a two-dimensional pooling layer and an output layer.

4. The intelligent control method for producing sildenafil citrate orodisintegrating tablets according to claim 3, characterized in that: The mannitol temperature-humidity time series correlation matrix is ​​subjected to feature extraction using a deep learning network model to obtain a mannitol temperature-humidity time series correlation feature graph, including: The mannitol temperature-humidity time series correlation matrix is ​​passed through the temperature-humidity time series correlation feature extractor based on the convolutional neural network model to obtain the mannitol temperature-humidity time series correlation feature graph.

5. The intelligent control method for producing sildenafil citrate orodisintegrating tablets according to claim 4, characterized in that: Performing time series analysis on the rotation speed values ​​of the dry granulator at the plurality of predetermined time points to obtain a rotation speed time series feature vector includes: Arrange the rotation speed values ​​of the dry granulator at the plurality of predetermined time points into a rotation speed time series input vector according to the time dimension; The speed time series input vector is passed through a speed time series feature extractor based on a one-dimensional convolutional layer to obtain the speed time series feature vector.

6. An intelligent control system for the production of sildenafil citrate orodisintegrating tablets, characterized in that: include: A mannitol granulation parameter monitoring module is used to obtain the mannitol temperature value, mannitol humidity value and dry granulation machine speed value at multiple predetermined time points within a predetermined time period; A mannitol temperature and humidity correlation analysis module, used for performing correlation analysis on the mannitol temperature values ​​and mannitol humidity values ​​at a plurality of predetermined time points within the predetermined time period to obtain a sequence of mannitol temperature-humidity time series correlation local feature vectors; A dry granulator speed time series analysis module, used for performing time series analysis on the dry granulator speed values ​​at the plurality of predetermined time points to obtain a speed time series feature vector; A speed control strategy generation module, used to determine the speed control strategy at a current time point based on the responsiveness fusion feature between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors; Wherein, the mannitol temperature and humidity correlation analysis module includes: A data preprocessing unit, used for performing data preprocessing on the mannitol temperature values ​​and the mannitol humidity values ​​at the plurality of predetermined time points to obtain a mannitol temperature time series input vector and a mannitol humidity time series input vector; A temperature and humidity time series correlation coding unit, used for correlation coding the mannitol temperature time series input vector and the mannitol humidity time series input vector into a mannitol temperature-humidity time series correlation matrix; A temperature and humidity time series correlation feature extraction unit, used for extracting features from the mannitol temperature-humidity time series correlation matrix using a deep learning network model to obtain a mannitol temperature-humidity time series correlation feature graph; A characteristic distribution correction unit, used for performing characteristic distribution correction on the mannitol temperature-humidity time series correlation characteristic diagram to obtain a corrected mannitol temperature-humidity time series correlation characteristic diagram; A feature flattening unit, used for flattening each feature matrix along the channel dimension of the corrected mannitol temperature-humidity time series correlation feature graph to obtain a sequence of local feature vectors of the mannitol temperature-humidity time series correlation; Wherein, determining the speed control strategy at the current time point based on the responsiveness fusion characteristics between the sequence of the mannitol temperature-humidity time series associated local feature vectors and the speed time series feature vectors includes: Calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain a responsive fused speed time series feature vector; Passing the responsive fused speed time series feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the speed value at the current time point should be increased, decreased, or maintained; The step of calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector to obtain the responsive fused speed time series feature vector comprises: calculating the responsiveness information fusion of the sequence of the mannitol temperature-humidity time series associated local feature vectors relative to the speed time series feature vector using the following responsiveness information fusion formula to obtain the responsive fused speed time series feature vector; Among them, the responsiveness information fusion formula is: ;in, yes The matrix of is the dimension of the speed time series feature vector, is the speed time series characteristic vector, yes The matrix of is the dimension of each of the local feature vectors of the mannitol temperature-humidity time series association, is the number of local eigenvectors of the mannitol temperature-humidity time series association, is the number of local feature vectors in the sequence of mannitol temperature-humidity time series association Mannitol temperature-humidity time series correlation local feature vectors, is the Sigmoid function, is the responsive fusion speed time series feature vector, and represents the convolution operation of a 1×1 convolution kernel, is the weight value.