Automatic monitoring method and monitoring platform for chemical dissolution process

Abstract

The invention discloses an automatic monitoring method and monitoring platform for a chemical dissolution process, and belongs to the field of computer vision, and the method comprises the steps: obtaining real-time image data of the chemical dissolution process; calculating a DCT coefficient and a DCT difference of the real-time image data, a gray scale difference value between an average gray scale value of the current frame and an average gray scale value of the reference value, and a maximum change rate of the gray scale; storing the data and a pre-acquired classification label as a data set; training a data set through a CatBoost model, and predicting the probability that a classification label is completely dissolved; when the probability that the classification label is completely dissolved is lower than a set threshold value, repeating the steps to re-train the CatBoost model until the probability that the classification label is completely dissolved reaches the set threshold value, and judging that the chemical medicine is completely dissolved; according to the invention, the efficiency and precision of chemical experiments can be improved, the difficulty and risk of experiment operation can be reduced, and the intelligent and automatic development of the chemical experiment technology can be promoted.

Description

Technical Field

[0001] The invention relates to an automatic monitoring method and a monitoring platform for a chemical dissolution process, and belongs to the interdisciplinary field of robotics, computer vision and chemistry. Background Art

[0002] In chemical laboratories, dissolution is a common experimental operation, which is widely used in the fields of solution preparation, solution preparation and sample processing. The dissolution process involves the interaction between solute and solvent, including the dispersion, diffusion and dissolution of solute molecules in the solvent. Effective monitoring of the dissolution process is crucial to ensure the accuracy of experimental data, improve experimental efficiency and ensure experimental safety. However, in chemical laboratories, the monitoring and operation of the dissolution process often face some challenges. Traditional experimental operations usually rely on the experience and skills of the experimenters, and there are problems such as unstable operation and low degree of operation standardization. In addition, since the experimenters need to perform other experimental operations or record experimental data during the operation, they may not be able to monitor every detail of the dissolution process in a timely and accurate manner. These problems may lead to uncertainty in experimental results, reduced experimental efficiency and the existence of experimental safety hazards. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies in the prior art and provide an automated monitoring method and monitoring platform for a chemical dissolution process, which can not only improve the efficiency and accuracy of chemical experiments, but also reduce the difficulty and risk of experimental operations, and promote the intelligent and automated development of chemical experimental technology.

[0004] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a method for automated monitoring of a chemical dissolution process, comprising: Obtain real-time image data of chemical dissolution process; Processing real-time image data based on DCT transformation, calculating DCT coefficients and DCT differences of real-time image data; Calculate the average grayscale value of the real-time image data, the grayscale difference between the average grayscale value of the current frame and the average grayscale value of the reference value, and the maximum grayscale change rate, wherein the maximum grayscale change rate is the maximum value of the difference ratio between the average grayscale value of the current frame and the average grayscale value of the previous N frames; The DCT coefficient, the DCT difference, the grayscale difference value, the maximum grayscale change rate and the pre-acquired classification label are stored as a data set, wherein the classification label is obtained by selecting a reference image and classifying the reference image, and the classification label includes label 1 and label 0, label 1 indicates complete dissolution, and label 0 indicates incomplete dissolution; The dataset is trained using a pre-built CatBoost model and the probability of the classification label being completely dissolved is predicted. When the probability of the classification label being completely dissolved is lower than the set threshold, repeat the above steps to retrain the CatBoost model until the probability of the classification label being completely dissolved reaches the set threshold, and it is determined that the chemical is completely dissolved.

[0005] Furthermore, the method also includes: if the maximum grayscale change rate is lower than a set threshold, calculating a new DCT coefficient based on the quotient of the DCT coefficient and the grayscale difference.

[0006] Further, the calculating of DCT coefficients of real-time image data includes: Dividing the real-time image data into a plurality of image blocks; Apply DCT transform to each image block to obtain DCT coefficient matrix; According to the DCT coefficient matrix, the average DCT coefficient of each DCT coefficient matrix is ​​calculated.

[0007] Furthermore, the calculation formula of the DCT coefficient matrix is ​​as follows: ,in, is the input image block matrix, is the coefficient matrix after DCT transformation, is the transformation matrix of DCT transform, Represents a transpose operation; The transformation matrix of DCT transform The calculation formula is as follows: ; in, and is the index of the matrix, is the size of the matrix, Represents pi.

[0008] Furthermore, the average DCT coefficient of each DCT coefficient matrix is ​​calculated as follows: ; in represents the average DCT coefficient of each DCT coefficient matrix, Indicates the DCT coefficient matrix Row, No. The element value of the column, and are the number of rows and columns of the DCT coefficient matrix, respectively.

[0009] Furthermore, the average grayscale value of the real-time image data is calculated by the following formula: ; in, Represents the average gray value of real-time image data, and Represent the height and width of the region of interest, respectively. = * Represents the total number of pixels, Indicates that the coordinates in the grayscale image are The pixel value of .

[0010] In a second aspect, the present invention provides an automated monitoring platform for a chemical dissolution process, comprising: The dissolution platform module is used to provide the basic conditions required for the dissolution experiment, including a reaction vessel and a stirrer; A robot module, including a robot arm, which is an automatic component for performing dissolution experiment operations and can perform the overall experimental process according to a preset program; The visual analysis module includes at least one high-definition camera for collecting real-time image data of the chemical dissolution process and executing the automatic monitoring method for the chemical dissolution process described in any one of the first aspects.

[0011] In a third aspect, the present invention provides an experimental method of the automated monitoring platform for chemical dissolution process according to the second aspect, comprising: Place the reaction vessel on the stirrer and place a high-definition camera in front of the stirrer for real-time monitoring; A high-definition camera is placed at the end of the robot module's mechanical arm. Based on the target detection algorithm, the target chemicals and containers are searched for. The mechanical arm approaches and confirms the target before picking and placing. The solvent is sucked by a robotic arm and placed into a reaction container through a control program; Turn on the stirrer and record the image of the solvent in the reaction container at this time as a reference value, and then turn off the stirrer; The solid solute is grabbed by the robotic arm and placed into the target beaker through the control program; Turn on the stirrer and monitor the dissolution state in the beaker in real time. If the dissolution time is too long, the stirring speed will be automatically adjusted. The method for automatically monitoring the chemical dissolution process described in any one of the first aspects is executed through a visual analysis module to determine whether the chemical has been completely dissolved. When the chemical is completely dissolved, the agitator is automatically turned off.

[0012] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0013] In a fifth aspect, the present invention provides a computer device, comprising: Memory, for storing computer programs / instructions; A processor, configured to execute the computer program / instructions to implement the steps of any one of the methods described in the first aspect.

[0014] In a sixth aspect, the present invention provides a computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides an automated monitoring method and monitoring platform for a chemical dissolution process. By acquiring real-time image data of a chemical dissolution process, processing the real-time image data based on a DCT transform, training a constructed CatBoost model, and predicting the dissolution probability, the efficiency and accuracy of chemical experiments can be improved, and the difficulty and risk of experimental operations can be reduced. At the same time, the present invention utilizes a method combining computer vision technology, robotics technology, and chemical dissolution experiments to provide an automated visual assistance solution for the dissolution process, realize unmanned automatic chemical experiments, and thus provide case support for further research on unmanned chemical laboratories. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of an automated monitoring platform for chemical dissolution processes that integrates robotics and visual analysis; Figure 2 It is a block diagram of the automated monitoring platform for chemical dissolution process integrating robotics and visual analysis; Figure 3 is a flow chart of a method for automated monitoring of a chemical dissolution process provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of different stages of the dissolution process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0018] Embodiment 1: This embodiment introduces a method for automatic monitoring of a chemical dissolution process, comprising: Obtain real-time image data of chemical dissolution process; Process the real-time image data based on DCT transform (discrete cosine transform) to calculate the DCT coefficients and DCT differences of the real-time image data; Calculate the average grayscale value of the real-time image data, the grayscale difference between the average grayscale value of the current frame and the average grayscale value of the reference value, and the maximum grayscale change rate, wherein the maximum grayscale change rate is the maximum value of the difference ratio between the average grayscale value of the current frame and the average grayscale value of the previous N frames; The DCT coefficient, the DCT difference, the grayscale difference value, the maximum grayscale change rate and the pre-acquired classification label are stored as a data set, wherein the classification label is obtained by selecting a reference image and classifying the reference image, and the classification label includes label 1 and label 0, label 1 indicates complete dissolution, and label 0 indicates incomplete dissolution; Train the dataset using a pre-built CatBoost model (a gradient boosted decision tree-based model) and predict the probability of the classification label being complete dissolution. When the probability of the classification label being completely dissolved is lower than the set threshold, repeat the above steps to retrain the CatBoost model until the probability of the classification label being completely dissolved reaches the set threshold, and it is determined that the chemical is completely dissolved.

[0019] like Figure 3 As shown, the application process of the automatic monitoring method of chemical dissolution process provided in this embodiment specifically involves the following steps: 1. Dissolution process monitoring based on DCT transformation; DCT transform is a mathematical transformation method that converts a signal or image from the time domain (or spatial domain) to the frequency domain and decomposes the image into a series of frequency components. In image processing, DCT transform is often used in signal compression, feature extraction, image coding and other fields. DCT transform decomposes a signal into a weighted sum of a set of cosine functions, which represent the frequency components of the signal in the frequency domain. DCT transform is similar to Fourier transform, but the coefficients generated by DCT transform are all real numbers, while the coefficients generated by Fourier transform can be complex numbers. In image processing, DCT transform is usually applied to block processing. The image is divided into many small blocks, each block is regarded as a signal, and then DCT transform is applied to each block. In the domain after DCT transform, the main information of the image (such as low-frequency components) is usually concentrated in the upper left corner, while the high-frequency part corresponds to the details and texture of the image.

[0020] During the dissolution process, when changes occur in the solid dissolution process, the solid particles in the solution gradually dissolve, causing changes in the optical properties of the solution. This change may affect the representation of the image in the frequency domain. For example, the gradual dissolution of solid particles may cause changes in the high-frequency part of the DCT transform. By calculating the average value of the DCT coefficients of each image block, the representation of the image block in the frequency domain can be obtained. These DCT coefficients reflect the frequency components in the image block and can be used to analyze the characteristics and changes of the image block. Therefore, by analyzing the frequency domain representation of the image block after the DCT transform, we can indirectly observe the changes in the dissolution process.

[0021] In order to be more suitable for digital image processing, it is necessary to use the matrix form of DCT transform: ,in, is the input image block matrix, is the coefficient matrix after DCT transformation, is the transformation matrix of DCT transform, Represents the transpose operation. The transformation matrix of DCT transformation is an orthogonal matrix. In the discrete case, The elements can be calculated by the following formula: ; in, and is the index of the matrix, is the size of the matrix, Represents pi.

[0022] For each image block coefficient matrix after DCT transformation , calculate the average value of all its elements and get the average DCT coefficient , this process can be expressed as: ,in Indicates the DCT coefficient matrix Row, No. The element value of the column, and are the number of rows and columns of the DCT coefficient matrix, respectively.

[0023] The average grayscale is the average grayscale value of all pixels in the image. When calculating the average grayscale, you first need to convert the color image into a grayscale image, then add up the grayscale value of each pixel, and finally divide it by the total number of pixels in the image. In this way, you get the average grayscale value of the image. .in, and Represent the height and width of the region of interest, respectively. = * Indicates the total number of pixels. Indicates that the coordinates in the grayscale image are The pixel value of .

[0024] During the monitoring process, extract a frame of image at a fixed time interval and repeat the following steps: Use the above method to calculate the average DCT coefficient of the image block in the rectangular frame of the new frame .examine Whether there is a significant change in the average DCT coefficient from the reference value. This can be judged by setting a threshold. The specific calculation process is as follows: First, the code tries to read a frame of image from the video capture device. After reading the image frame, the code extracts the region of interest (ROI) from it, which is defined by the rectangular coordinates (start_x, start_y) and (end_x, end_y). Next, the DCT coefficient and the average grayscale value in the region of interest are calculated. Then the difference between the DCT coefficient, the average grayscale of the current image and its reference value is compared. If the historical grayscale difference satisfies the maximum change rate of the historical grayscale for 6 consecutive times is less than 0.012, it is considered that the dissolution has entered a stable state. The code will adjust the new DCT coefficient according to the quotient of the average grayscale and the DCT coefficient to exclude the interference of color transformation during the dissolution process. Then the DCT coefficient, grayscale difference, DCT difference, and the maximum change rate of grayscale are stored in the data set. After completing the processing, the code will update the result index for the next processing.

[0025] 2. Prediction of dissolution probability based on CatBoost model; For the data set extracted in Invention Content 1, a CatBoost model is used to train it.

[0026] The CatBoost model uses a gradient boosting algorithm that improves the model's predictive power by iteratively adding decision trees. In each iteration, the algorithm builds a new decision tree to predict the residual of the current model (i.e., the difference between the model's predicted value and the actual value). This process is done using the principle of gradient descent, with the goal of minimizing the loss function.

[0027] The CatBoost model implements an efficient One-hot encoding method, which encodes during training rather than preprocessing, thereby reducing training time and memory usage. The CatBoost model uses a method called symmetric transformation, which processes categorical features by randomly permuting the dataset and then calculating the average label value for each category. This method helps reduce overfitting. The calculation formula for this method is as follows: ; in, It is a classification feature value, which is a dynamic coding value. The specific meaning is: sample The value generated by the kth classification feature after special encoding by the CatBoost model. Here, the classification feature value comes from the DCT coefficient, DCT difference, grayscale difference and maximum grayscale change rate mentioned above. After these features have been discretized into category features, the CatBoost model uses its dynamic encoding method to model these discretized features and finally generates the classification feature value. Represents the original value of the k-th categorical feature of the sample (after discretization into categorical features). It is a discrete category identifier and has not been dynamically encoded by CatBoost. is the category label, is a random sorting sequence, P represents the priority, α>0 represents the weight coefficient of the priority, To find the product symbol.

[0028] Any combination of several categorical features can be considered as a new feature. The numerical features calculated using the above formula may lose some information. Combination features can solve this problem and obtain a more effective feature. When constructing a new split in the current tree, the CatBoost model considers the combination of features in a greedy way. The combination of categorical features is not considered in the first split, and all feature combinations are considered in subsequent splits, and the combined features become numerical. The CatBoost model also uses all the two sets of values ​​after the split as categorical features to participate in the subsequent combination. Another way to replace categorical features is to count the number of times the category appears in the data set.

[0029] The CatBoost model is similar to other gradient boosting algorithms. Each time it builds a new tree, it approximates the gradient of the current model. However, the gradient boosting algorithm has the problem of overfitting caused by biased point-by-point gradient estimation, so the CatBoost model attempts to correct the gradient bias to alleviate overfitting. There are two stages in building a tree: first, selecting the tree structure; second, calculating the value of the leaf node after the tree structure is fixed. The CatBoost model uses the traditional GBDT method (gradient boosting decision tree method) in the second stage, while the unbiased gradient estimation is used in the first stage.

[0030] The CatBoost model uses a symmetric binary tree as a base predictor. This tree is balanced and not prone to overfitting. In a symmetric binary tree, each leaf node can be encoded as a binary vector with a length equal to the depth of the tree. All floating-point features, statistics, and One-Hot encoded features are first binarized, and then the binary features are used to calculate the prediction value. The binary feature values ​​of all samples are stored in a continuous vector that can be constructed in parallel at 3 times the speed, which leads to faster scoring.

[0031] Each test will get a new set of feature values ​​and labels, which will be added in. Each test will train the expanded data set and predict the feature values ​​extracted at that time. When the predicted probability of dissolution exceeds 95% for three consecutive times, the dissolution is considered complete.

[0032] Based on the above technical solution provided in this application, an automated monitoring platform solution for the chemical dissolution process integrating robots and visual analysis can be provided, which can be used in the future to assist chemical robots in autonomously performing dissolution operations.

[0033] Embodiment 2, as Figure 1 , Figure 2 As shown, this embodiment provides an automated monitoring platform for a chemical dissolution process, including: The dissolution platform module is used to provide the basic conditions required for the dissolution experiment, including a reaction vessel and a stirrer; A robot module, including a robot arm, which is an automatic component for performing dissolution experiment operations and can perform the overall experimental process according to a preset program; The visual analysis module includes at least one high-definition camera for collecting real-time image data of the chemical dissolution process and executing the automated monitoring method for the chemical dissolution process described in any one of Example 1.

[0034] The specific functional implementation of each of the above modules can be found in the relevant contents of the method in Example 1 and will not be elaborated here.

[0035] In order to better understand the contents of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. The described embodiments are only part of the embodiments of this application, not all of the embodiments, and are by no means intended to limit the present application and its application or use.

[0036] The first is the dissolution platform module, which uses a stirrer as the platform equipment basis for the experiment. The platform design needs to comply with automated operations. The platform areas include the drug area, chemical vessel area, dissolution area, etc. The robot module uses a robotic arm as the core component. Combined with the dissolution platform, it will set the working area of ​​the robotic arm, set the program, and perform the overall execution of the experiment. The visual assistance system module uses a high-definition camera installed at the end of the robotic arm and a fixed high-definition camera. Among them, the high-definition camera at the end of the robotic arm mainly uses the target detection network improved and constructed based on YOLOv8 (target detection and image segmentation model) to detect chemical vessels and drugs, and identify and determine the target object during the movement of the robotic arm. Another fixed high-definition camera is mainly used to monitor the dissolution state in the container. As a software driver module, the algorithm program driver calculates the DCT coefficients and average grayscale in the container in real time, and detects when it is completely dissolved by setting a threshold.

[0037] Preparation of medicines and chemical vessels: The embodiments of this application standardize the feed ratio, concentration and other aspects of the required materials. The materials for each robot experiment are quantitatively standardized to facilitate the use and process application of the robot module. The required reagents are 9-fluorenone and ethanol. Because 9-fluorenone dissolves in ethanol solvent and changes from a colorless and transparent solution to a yellow transparent solution, the function of the algorithm used in this application can be fully demonstrated. By querying the solubility table, the experiment initially sets ethanol to 90ml and 9-fluorenone to 18g. In addition, in the chemical vessel placement area, the required vessels including beakers, medicine bottles, etc. are placed.

[0038] The steps of visually assisted monitoring of the dissolution process are as follows: After the addition of ethanol is completed, the stirrer is turned on. At this time, the camera records the image in the beaker in real time, and the image detection algorithm identifies the position information of the beaker and automatically calculates the region of interest (ROI). The DCT coefficient, grayscale difference, DCT difference, and maximum grayscale change rate of the solvent image in the ROI area of ​​the beaker are recorded and stored in the data set, and then the stirrer is turned off. Add 9-fluorenone and turn on the stirrer, and then perform a dissolution test every two seconds. Each time the DCT coefficient and change rate of the current ROI area are calculated, and when the change rate is continuously lower than the threshold, the DCT coefficient is adjusted. The adjusted DCT coefficient, grayscale difference, DCT difference, and maximum grayscale change rate are stored in the data set. Starting from the tenth test, the CatBoost model is used to train the existing data set and predict the current time. When the prediction probability exceeds 95% for three consecutive times, the solution is completely dissolved and the stirring is stopped.

[0039] According to the experimental device and method disclosed in this application, the visually assisted monitoring and operation of the dissolution process in chemical experiments can be realized, and the unit operation of chemical automated synthesis can be explored. The experimental device and method disclosed in this application can be used to design a visual scheme to assist the chemical experiment - dissolution.

[0040] The contents involved in the above embodiment are described below in conjunction with a preferred embodiment.

[0041] The robot arm automatically monitors and operates the dissolution process in real time using 9-fluorenone and ethanol as raw materials: The DCT-CatBoost algorithm monitors the dissolution process execution steps as described above. A standard medicine bottle containing 18 g of 9-fluorenone and 90 ml of ethanol is placed in the robot feeding work area and fixed in place. All chemicals are attached with information labels. The robot module is automatically programmed according to the designed experimental process.

[0042] First, the high-definition camera at the end of the robotic arm is used to identify the experimental scene in real time, and the chemical vessel area is detected through the target detection algorithm to detect the two targets of the drug bottle containing 9-fluorenone and ethanol that need to be used. After confirming the target, the robotic arm begins to approach the drug, and the high-definition camera placed on the robotic arm is facing the position of the drug label to identify the information of the chemical drug label. After confirmation, the robotic arm grabs the target drug. Through the above-mentioned drug confirmation process, a certain amount of ethanol is first added to the beaker containing the rotor placed on the stirrer. A fixed high-definition camera is placed in front of the stirrer for real-time monitoring, recording the roi area and DCT coefficient and other parameters at this time, and the DCT coefficient, DCT difference, grayscale difference value, and maximum grayscale change rate are placed in the data set, and then the stirrer is turned off. The robotic arm grabs the drug bottle containing 9-fluorenone and pours 9-fluorenone into the target beaker through the control program. Turn on the stirrer and monitor the dissolution state in the beaker in real time. The DCT coefficient and DCT difference in the roi area are calculated every 2 seconds. At the same time, calculate the ratio of the grayscale difference between the current element and the previous five elements, and get the maximum value. If the maximum value is less than 1.2% for 6 consecutive times, calculate the new DCT coefficient, which is the original DCT coefficient divided by the grayscale difference. DCT coefficient, DCT difference, grayscale difference value, and maximum rate of change of grayscale. Starting from the tenth detection, the CatBoost model is used to train the existing data set each time and predict the current time. When the predicted probability exceeds 95% for 3 consecutive times, it is determined that the solution is completely dissolved and the stirring is ended. In addition, when the dissolution time is too long, the robotic arm automatically adjusts parameters such as the stirring speed by controlling the knob. According to the above process, the robot automatically monitors and operates the dissolution process. The scene diagram is shown as follows. Figure 1 The schematic diagram of different stages of the dissolution process is shown in Figure 4shown.

[0043] Embodiment 3, the present invention provides an experimental method of the automatic monitoring platform for chemical dissolution process described in embodiment 2, comprising: Place the reaction vessel on the stirrer and place a high-definition camera in front of the stirrer for real-time monitoring; A high-definition camera is placed at the end of the robot module's mechanical arm. Based on the target detection algorithm, the target chemicals and containers are searched for. The mechanical arm approaches and confirms the target before picking and placing. The solvent is sucked by a robotic arm and placed into a reaction container through a control program; Turn on the stirrer and record the image of the solvent in the reaction container at this time as a reference value, and then turn off the stirrer; The solid solute is grabbed by the robotic arm and placed into the target beaker through the control program; Turn on the stirrer and monitor the dissolution state in the beaker in real time. If the dissolution time is too long, the stirring speed will be automatically adjusted. The automatic monitoring method for the chemical dissolution process described in any one of Example 1 is executed by the visual analysis module to determine whether the chemical has been completely dissolved. When the chemical is completely dissolved, the agitator is automatically turned off.

[0044] Embodiment 4: This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of any method described in Embodiment 1 are implemented.

[0045] Embodiment 5: This embodiment provides a computer device, including: Memory, for storing computer programs / instructions; A processor, configured to execute the computer program / instructions to implement the steps of any one of the methods described in Example 1.

[0046] Embodiment 6: This embodiment provides a computer program product, including a computer program / instruction, which implements the steps of any method described in Embodiment 1 when executed by a processor.

[0047] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0048] It should be understood by those skilled in the art that the embodiments of the present disclosure may be provided as methods, systems or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0049] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram and the combination of processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0050] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0051] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure rather than to limit its protection scope. Although the present disclosure has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that after reading the present disclosure, those skilled in the art can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the disclosed claims to be approved.

Claims

1. A method for automated monitoring of a chemical dissolution process, characterized in that: include: Obtain real-time image data of chemical dissolution process; Processing real-time image data based on DCT transformation, calculating DCT coefficients and DCT differences of real-time image data; Calculate the average grayscale value of the real-time image data, the grayscale difference between the average grayscale value of the current frame and the average grayscale value of the reference value, and the maximum grayscale change rate, wherein the maximum grayscale change rate is the maximum value of the difference ratio between the average grayscale value of the current frame and the average grayscale value of the previous N frames; The DCT coefficient, the DCT difference, the grayscale difference value, the maximum grayscale change rate and the pre-acquired classification label are stored as a data set, wherein the classification label is obtained by selecting a reference image and classifying the reference image, and the classification label includes label 1 and label 0, label 1 indicates complete dissolution, and label 0 indicates incomplete dissolution; The dataset is trained using a pre-built CatBoost model and the probability of the classification label being completely dissolved is predicted. When the probability of the classification label being completely dissolved is lower than the set threshold, repeat the above steps to retrain the CatBoost model until the probability of the classification label being completely dissolved reaches the set threshold, and it is determined that the chemical is completely dissolved.

2. The method for automated monitoring of a chemical dissolution process according to claim 1, characterized in that: The method further comprises: if the maximum grayscale change rate is lower than a set threshold, calculating a new DCT coefficient according to the quotient of the DCT coefficient and the grayscale difference.

3. The method for automated monitoring of a chemical dissolution process according to claim 1, characterized in that: The step of calculating the DCT coefficients of the real-time image data comprises: Dividing the real-time image data into a plurality of image blocks; Apply DCT transform to each image block to obtain DCT coefficient matrix; According to the DCT coefficient matrix, the average DCT coefficient of each DCT coefficient matrix is ​​calculated.

4. The method for automated monitoring of a chemical dissolution process according to claim 3, characterized in that: The calculation formula of the DCT coefficient matrix is ​​as follows: ,in, is the input image block matrix, is the coefficient matrix after DCT transformation, is the transformation matrix of DCT transform, Represents a transpose operation; The transformation matrix of DCT transform The calculation formula is as follows: ； in, and is the index of the matrix, is the size of the matrix, Represents pi.

5. The method for automated monitoring of a chemical dissolution process according to claim 4, characterized in that: The average DCT coefficient of each DCT coefficient matrix is ​​calculated as follows: ； in represents the average DCT coefficient of each DCT coefficient matrix, Indicates the DCT coefficient matrix Row, No. The element value of the column, and are the number of rows and columns of the DCT coefficient matrix, respectively.

6. The method for automated monitoring of a chemical dissolution process according to claim 5, characterized in that: The average grayscale value of the real-time image data is calculated by the following formula: ； in, Represents the average gray value of real-time image data, and Represent the height and width of the region of interest, respectively. = * Represents the total number of pixels, Indicates that the coordinates in the grayscale image are The pixel value of .

7. A chemical dissolution process automation monitoring platform, characterized in that: include: The dissolution platform module is used to provide the basic conditions required for the dissolution experiment, including a reaction vessel and a stirrer; A robot module, including a robot arm, which is an automatic component for performing dissolution experiment operations and can perform the overall experimental process according to a preset program; The visual analysis module includes at least one high-definition camera for collecting real-time image data of the chemical dissolution process and executing the automatic monitoring method of the chemical dissolution process described in any one of claims 1-6.

8. An experimental method of the automatic monitoring platform for chemical dissolution process according to claim 7, characterized in that: include: The reaction vessel is placed on the stirrer, and a high-definition camera is placed in front of the stirrer for real-time monitoring; A high-definition camera is placed at the end of the robot module's mechanical arm. Based on the target detection algorithm, the target chemicals and containers are searched for. The mechanical arm approaches and confirms the target before picking and placing. The solvent is sucked by a robotic arm and placed into a reaction container through a control program; Turn on the stirrer and record the image of the solvent in the reaction container at this time as a reference value, and then turn off the stirrer; The solid solute is grabbed by the robotic arm and placed into the target beaker through the control program; Turn on the stirrer and monitor the dissolution state in the beaker in real time. If the dissolution time is too long, the stirring speed will be automatically adjusted. The method for automatically monitoring the chemical dissolution process according to any one of claims 1 to 6 is executed through a visual analysis module to determine whether the chemical has been completely dissolved. When the chemical is completely dissolved, the agitator is automatically turned off.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are implemented.

10. A computer device, characterized in that: include: Memory, for storing computer programs / instructions; A processor, configured to execute the computer program / instructions to implement the steps of the method according to any one of claims 1 to 6.

Images