Intelligent detection method and device for purity of food-grade carbon dioxide

Abstract

The present invention discloses a method and device for intelligent detection of the purity of food-grade carbon dioxide, and relates to the field of intelligent detection technology. The method comprises: disassembling the carbon dioxide purification process flow, obtaining the first process node to the Nth process node; traversing the first process node to the Nth process node to train the carbon dioxide purity prediction model, and obtaining the carbon dioxide purity prediction pipeline; obtaining the carbon dioxide purity prediction value through the carbon dioxide purity prediction pipeline; when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold, the carbon dioxide purification control parameter of the first process node, the carbon dioxide purification control parameter of the second process node to the carbon dioxide purification control parameter of the Nth process node are marked as abnormal in purity. The technical problem that the traditional process cannot predict the purity of carbon dioxide, resulting in the lack of quantitative control means, is solved, and the technical effect of realizing quantitative control of carbon dioxide purity is achieved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent detection technology, and in particular to a method and device for intelligent detection of the purity of food-grade carbon dioxide. Background Art

[0002] In the process of food production and processing, carbon dioxide, as an important food additive, is widely used in many fields such as carbonated beverages and food preservation. In order to ensure food safety, there are strict requirements for the purity of food-grade carbon dioxide. However, the existing carbon dioxide purification process usually relies on experience and manual detection. Not only can it not grasp the purity changes in real time, but there are also problems such as low detection efficiency and large errors. Especially in the purification process, due to the lack of effective prediction and quantitative control methods, process control is difficult to refine, and purification parameters cannot be quickly adjusted to ensure the target purity, thereby increasing the risk of unstable quality of food-grade carbon dioxide. Summary of the invention

[0003] The present application provides a method and device for intelligent detection of the purity of food-grade carbon dioxide, which solves the technical problem that traditional processes are unable to predict the purity of carbon dioxide, resulting in a lack of quantitative control means.

[0004] In view of the above problems, the present application provides a method and device for intelligent detection of the purity of food-grade carbon dioxide.

[0005] In a first aspect of the present application, a method for intelligently detecting the purity of food-grade carbon dioxide is provided, the method comprising:

[0006] A carbon dioxide purification process flow and a set of raw material component types are obtained; the carbon dioxide purification process flow is disassembled to obtain a first process node, a second process node, and up to an Nth process node; according to the carbon dioxide purification process flow and the set of raw material component types, the first process node, the second process node, and up to the Nth process node are traversed to train a carbon dioxide purity prediction model to obtain a carbon dioxide purity prediction pipeline; a raw material component weight ratio of the raw material component type set, and a first process node carbon dioxide purification control parameter, a second process node carbon dioxide purification control parameter, and up to an Nth process node carbon dioxide purification control parameter are received; through the carbon dioxide purity prediction pipeline, the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter, and up to the Nth process node carbon dioxide purification control parameter are traversed, and the carbon dioxide purification control parameter of the first process node, the second process node carbon dioxide purification control parameter, and up to the Nth process node carbon dioxide purification control parameter are processed in combination with the raw material component weight ratio to obtain a carbon dioxide purity prediction value; when the carbon dioxide purity prediction value is less than or equal to a carbon dioxide purity threshold, the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter, and up to the Nth process node carbon dioxide purification control parameter are marked as abnormal in purity.

[0007] The second aspect of the present application provides an intelligent detection device for the purity of food-grade carbon dioxide, the device comprising:

[0008] A data acquisition module, the data acquisition module is used to obtain a carbon dioxide purification process flow and a set of raw material component types; a disassembly module, the disassembly module is used to disassemble the carbon dioxide purification process flow to obtain a first process node, a second process node, and finally an Nth process node; a training module, the training module is used to traverse the first process node, the second process node, and finally the Nth process node according to the carbon dioxide purification process flow and the set of raw material component types to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline; a data receiving module, the data receiving module is used to receive the raw material component weight ratio of the raw material component type set, as well as the first process node carbon dioxide purification control parameter, the second process node carbon dioxide a carbon purification control parameter up to the Nth process node carbon dioxide purification control parameter; a data processing module, the data processing module is used to traverse the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter up to the Nth process node carbon dioxide purification control parameter through the carbon dioxide purity prediction pipeline, and process in combination with the raw material component weight ratio to obtain a carbon dioxide purity prediction value; an abnormality identification module, the abnormality identification module is used to mark the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter up to the Nth process node carbon dioxide purification control parameter for purity abnormality when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] First, obtain the carbon dioxide purification process flow and the set of raw material component types. Then, disassemble the carbon dioxide purification process flow to obtain the first process node, the second process node, and the Nth process node. Further, according to the carbon dioxide purification process flow and the set of raw material component types, traverse the first process node, the second process node, and the Nth process node to train the carbon dioxide purity prediction model, and obtain the carbon dioxide purity prediction pipeline. Next, receive the raw material component weight ratio of the raw material component type set, as well as the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node. Then, through the carbon dioxide purity prediction pipeline, traverse the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node, and process them in combination with the raw material component weight ratio to obtain the carbon dioxide purity prediction value. Finally, when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold, the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node are marked as abnormal in purity. It solves the technical problem that traditional processes cannot predict the purity of carbon dioxide, resulting in a lack of quantitative control methods, and achieves the technical effect of realizing quantitative control of carbon dioxide purity. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic diagram of a method for intelligently detecting the purity of food-grade carbon dioxide provided in an embodiment of the present application;

[0013] Figure 2 This is a schematic diagram of the structure of an intelligent detection device for the purity of food-grade carbon dioxide provided in an embodiment of the present application.

[0014] Explanation of the accompanying drawings: data acquisition module 11, disassembly module 12, training module 13, data receiving module 14, data processing module 15, abnormality identification module 16. DETAILED DESCRIPTION

[0015] The present application solves the technical problem that traditional processes are unable to predict the purity of carbon dioxide, resulting in a lack of quantitative control means, by providing an intelligent detection method and device for the purity of food-grade carbon dioxide.

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides a method for intelligent detection of the purity of food-grade carbon dioxide, wherein the method comprises:

[0019] Obtain a set of carbon dioxide purification process flows and raw material component types.

[0020] The carbon dioxide purification process flow is obtained through the database. The carbon dioxide purification process flow records the sequence of each process step, equipment configuration, operating parameters, etc. in the purification process. At the same time, the raw material component type set related to the carbon dioxide purification process is collected. The raw material component type set includes various types of raw materials that may be used for purification and their component ratios.

[0021] The carbon dioxide purification process flow is disassembled to obtain a first process node, a second process node, and finally an Nth process node.

[0022] The complete carbon dioxide purification process is gradually disassembled into multiple independent process nodes, each of which corresponds to a purification process, such as pretreatment, cooling, absorption, separation, etc. Through decomposition, the first process node, the second process node, and the Nth process node are obtained, forming a step-by-step process of the purification process, so that each process node can be analyzed and controlled separately.

[0023] According to the carbon dioxide purification process and the raw material component type set, the first process node, the second process node and finally the Nth process node are traversed to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline.

[0024] Based on the purification process and the set of raw material component types, data is collected and analyzed for each process node in the purification process (from the first process node to the Nth process node), and a carbon dioxide purity prediction model is constructed and trained. After the model training is completed, the prediction model of each node is connected in series to form a carbon dioxide purity prediction pipeline. The carbon dioxide purity prediction pipeline receives the control parameters and raw material component ratio of each process node in turn, and gradually generates and outputs the purity prediction value.

[0025] Furthermore, according to the carbon dioxide purification process and the raw material component type set, the first process node, the second process node and the Nth process node are traversed to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline, including:

[0026] Taking the carbon dioxide purification process flow and the raw material component type set as constraints, collecting the purification log data of the first process node, wherein the purification log data of the first process node includes the raw material component weight record matrix, the carbon dioxide purification control record parameters and the first record data of the generated substance weight; taking the first record data of the generated substance weight as supervision, and taking the raw material component weight record matrix and the carbon dioxide purification control record parameters as input, training the carbon dioxide purity prediction model of the first process node; looping analysis until obtaining the carbon dioxide purity prediction model of the second process node until the carbon dioxide purity prediction model of the Nth process node; merging the output layer of the carbon dioxide purity prediction model of the first process node with an input node of the carbon dioxide purity prediction model of the second process node until merging the output layer of the carbon dioxide purity prediction model of the N-1th process node with an input node of the carbon dioxide purity prediction model of the Nth process node to generate the carbon dioxide purity prediction pipeline.

[0027] Specifically, the purification log data of the first process node is collected with the carbon dioxide purification process and the set of raw material component types as constraints. The purification log data of the first process node includes the raw material component weight record matrix, the carbon dioxide purification control record parameters and the first record data of the generated material weight, wherein the raw material component weight record matrix records the specific weight share of each component in the raw material used for purification, the carbon dioxide purification control record parameters include temperature, pressure, time and other control parameters, which are used to reflect the operating state of the process node, and the first record data of the generated material weight record records the weight share of the material generated after being processed by the first process node, as the supervision data for purity prediction. Using the first record data of the generated material weight as the supervision target, the raw material component weight record matrix and the carbon dioxide purification control record parameters as input variables, the carbon dioxide purity prediction model of the first process node is trained, and the model is used to predict the purity level of the carbon dioxide generated at the process node. Repeat the above process for each process node in the purification process, collect the log data of each node one by one, and train the carbon dioxide purity prediction model from the second process node to the Nth process node. The output layer of the purity prediction model of the first process node is merged with an input node of the second process node, so that the second process node can receive the prediction result of the previous node and its own input parameters; the prediction model output layer of each process node and the input layer of the next process node are connected in sequence until the output layer of the N-1th process node is merged with the input node of the Nth process node, thereby generating a carbon dioxide purity prediction pipeline.

[0028] Furthermore, the first recorded data of the weight of the generated substance is used as supervision, and the raw material component weight record matrix and the carbon dioxide purification control record parameters are used as input to train the first process node carbon dioxide purity prediction model, including:

[0029] Taking the raw material component weight record matrix and the carbon dioxide purification control record parameters as dynamic constraints, and the carbon dioxide purification process and the raw material component type set as static constraints, collect the first process node purification incremental log data, wherein the first process node purification incremental log data includes a plurality of generated substance weight second record data; perform centralized value evaluation on the generated substance weight first record data and the plurality of generated substance weight second record data to obtain the generated substance weight identification value; take the generated substance weight identification value as supervision, and take the raw material component weight record matrix and the carbon dioxide purification control record parameters as input to train the first process node carbon dioxide purity prediction model.

[0030] Specifically, the raw material component weight record matrix and carbon dioxide purification control record parameters are used as dynamic constraints, and their specific values ​​in the purification process may be adjusted as the operating conditions change; the carbon dioxide purification process and the raw material component type set are used as static constraints to ensure that the model training is based on the fixed conditions of the overall purification process and the basic framework of the raw material types. Based on the above constraints, the purification incremental log data of the first process node is collected. The incremental log data contains multiple weight records of generated substances, that is, the second record data of the weight of the generated substances, which reflect the purification results under different operating conditions. The initial first record data of the weight of the generated substances and the second record data of the weight of the generated substances in the incremental log are evaluated for the concentration value. By analyzing the average or weighted concentration trend of these data, the generated substance weight identification value representing the purity of the product of the process node is obtained as the supervision label of the model. The generated substance weight identification value is used as the supervision target, and the raw material component weight record matrix and carbon dioxide purification control record parameters are used as input variables to train the carbon dioxide purity prediction model of the first process node.

[0031] Further, the raw material component weight record matrix and the carbon dioxide purification control record parameters are used as dynamic constraints, and the carbon dioxide purification process and the raw material component type set are used as static constraints to collect the first process node purification incremental log data, wherein the first process node purification incremental log data includes a plurality of generated substance weight second record data, including:

[0032] Taking the carbon dioxide purification process flow and the raw material component type set as static constraints, the first-level purification incremental log data of the first process node is collected, wherein the first-level purification incremental log data of the first process node includes a first-level raw material component weight record matrix, a first-level carbon dioxide purification control record parameter and a first-level generated substance weight second record data; the matrix element deviation mean and matrix element deviation variance of the first-level raw material component weight record matrix and the raw material component weight record matrix are compared; the carbon dioxide purification control record parameter and the first-level carbon dioxide purification control record parameter are compared. The proportion of attributes whose control record parameter deviation is greater than or equal to the control parameter deviation threshold is set as the control record parameter deviation degree; when the matrix element deviation mean is less than or equal to the deviation mean threshold, and the matrix element deviation variance is less than or equal to the deviation variance threshold, and the control record parameter deviation degree is less than or equal to the deviation degree threshold, the first-level generated substance weight second record data of the first-level purification incremental log data of the first process node is added to the first process node purification incremental log data.

[0033] Specifically, under static constraint conditions, the first-level purification incremental log data of the first process node is collected. The first-level purification incremental log data of the first process node includes the first-level raw material component weight record matrix, the first-level carbon dioxide purification control record parameter and the first-level generated substance weight second record data. The first-level raw material component weight record matrix records the weight ratio of the raw material components in the first-level process node, the first-level carbon dioxide purification control record parameter reflects the control parameters of the first-level process node (such as temperature, pressure, time, etc.), and the first-level generated substance weight second record data records the weight of the generated substance under the purification operation of the node. Calculate and compare the element deviations in the first-level raw material component weight record matrix and the raw material component weight record matrix, that is, calculate the mean (deviation mean) and variance (deviation variance) of the two matrix element deviations; when the matrix element deviation mean is less than or equal to the preset deviation mean threshold and the matrix element deviation variance is less than or equal to the deviation variance threshold, it is considered that the deviations of the two matrices meet the requirements. The first-level carbon dioxide purification control record parameters are compared with the carbon dioxide purification control record parameters, and the deviation of the control record parameters is calculated. The deviation is determined by statistically analyzing the proportion of attributes in the control record parameters whose deviation is greater than or equal to the control parameter deviation threshold; when the control record parameter deviation is less than or equal to the set deviation threshold, the control parameters of the first-level data are considered to meet the requirements. Only when all the above comparison conditions are met, that is, the mean value of the matrix element deviation and the variance of the matrix element deviation are within the threshold range, and the control record parameter deviation is also within the allowable range, the second record data of the first-level generated material weight of the first-level purification incremental log data is added to the first process node purification incremental log data.

[0034] Furthermore, the first recorded data of the generated substance weight portion and the plurality of second recorded data of the generated substance weight portion are evaluated for a centralized value to obtain the generated substance weight portion identification value, including:

[0035] Arrange the first record data of the weight of the generated substance of any substance type and the second record data of the generated substance weight in order from small to large to obtain the weight of the generated substance with a serial number rounded up to one quarter and the weight of the generated substance with a serial number rounded down to three quarters; calculate the weight of the generated substance with the serial number rounded down to three quarters minus the weight of the generated substance with the serial number rounded up to one quarter, and obtain the range of the weight of the generated substance; calculate the weight of the generated substance with the serial number rounded down to three quarters plus 1.5 times the range of the generated substance weight to obtain the first boundary weight of the generated substance; calculate the weight of the generated substance with the serial number rounded up to one quarter minus 1.5 times the range of the generated substance weight to obtain the second boundary weight of the generated substance; count the average of the weight of the generated substances greater than or equal to the second boundary weight of the generated substance and less than or equal to the first boundary weight of the generated substance in the first record data of the generated substance weight and the second record data of the generated substance weight, and obtain the identification value of the generated substance weight.

[0036] Specifically, the first record data of the weight of the generated substance and several second record data of the weight of the generated substance are arranged in ascending order, and these data record the weight share of a specific substance type under different purification operations; according to the arranged order, the serial number of the quarter position is rounded up to determine the weight of the generated substance at that position, which is called the weight of the generated substance with the quarter-rounded serial number; similarly, the serial number of the three-quarter position is rounded down to determine the weight of the generated substance at that position, which is called the weight of the generated substance with the three-quarters rounded down serial number; the range of the weight of the generated substance is calculated, that is, the weight of the generated substance with the three-quarters rounded down serial number minus the weight of the generated substance with the quarter-rounded up serial number. ; Calculate the weight of the substance generated by rounding down three quarters of the serial number plus 1.5 times the range of the weight of the substance generated, and obtain the first boundary weight of the substance generated (as the upper boundary); calculate the weight of the substance generated by rounding up one quarter of the serial number minus 1.5 times the range of the weight of the substance generated, and obtain the second boundary weight of the substance generated (as the lower boundary); count the data of the first record data of the weight of the substance generated and several second record data of the weight of the substance generated, that is, the data of the weight of the substance generated that is greater than or equal to the second boundary and less than or equal to the first boundary; calculate the mean of the data that meets the conditions to obtain the final identification value of the weight of the substance generated for subsequent model training and evaluation.

[0037] The raw material component weight ratio of the raw material component type set, and the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter, and so on to the Nth process node carbon dioxide purification control parameter are received.

[0038] Receive and confirm the set of raw material component types entering the purification process and their respective weight ratios. This information is used to establish the basic ratio of raw material components as one of the inputs for model prediction, so that the impact of raw material components on the final purity can be accurately reflected throughout the purification process. Receive carbon dioxide purification control parameters from the first process node to the Nth process node in sequence. The control parameters of each node usually include key process conditions, such as temperature, pressure, time, flow rate, etc.

[0039] Through the carbon dioxide purity prediction pipeline, the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter are traversed, and the raw material component weight ratio is combined for processing to obtain a carbon dioxide purity prediction value.

[0040] In the carbon dioxide purity prediction pipeline, the purification control parameters (such as temperature, pressure, time, flow rate, etc.) of each process node and the weight ratio of raw material components are passed to the model as input. The model processes the control parameters of the first process node to the Nth process node in turn, and the output result of each node is used to provide a purity prediction for the current node. The purity prediction value of each node is combined with the raw material component ratio as one of the inputs for the next node prediction to ensure the correlation between upstream and downstream nodes. After the prediction and processing of all process nodes, the pipeline finally generates a comprehensive carbon dioxide purity prediction value. This prediction value reflects the purity result of the entire purification process, and provides a reference for subsequent quality monitoring and process optimization based on the conditions of each process node and the raw material component ratio.

[0041] Furthermore, it also includes:

[0042] Through the carbon dioxide purity prediction pipeline, according to the first process node carbon dioxide purification control parameters, combined with the weight ratio of the raw material components, processing is performed to obtain a predicted value of the weight of the substance generated at the first process node, wherein the weight of the substance generated at the first process node includes the predicted value of the carbon dioxide purity at the first process node; when the predicted value of the carbon dioxide purity at the first process node is less than or equal to the carbon dioxide purity threshold of the first process node, the carbon dioxide purification control parameter of the first process node is marked as abnormal in purity; when the predicted value of the carbon dioxide purity at the first process node is greater than the carbon dioxide purity threshold of the first process node, processing is performed through the carbon dioxide purity prediction pipeline according to the predicted value of the weight of the substance generated at the first process node and the carbon dioxide purification control parameter of the first process node.

[0043] Through the carbon dioxide purity prediction pipeline, according to the carbon dioxide purification control parameters of the first process node, combined with the weight ratio of the raw material components, the predicted value of the weight of the substance generated by the first process node is generated. The predicted value of the weight of the substance generated by the first process node includes the predicted value of the carbon dioxide purity of the first process node, reflecting the purity and weight of the carbon dioxide generated under the current conditions. If the predicted value of the carbon dioxide purity of the first process node is less than or equal to the preset first process node carbon dioxide purity threshold, the purification control parameter of the process node is marked as abnormal purity. The abnormal purity mark is used to prompt that the purification conditions of the current node may cause the purity to be unqualified, so as to take adjustment measures. If the predicted value of the carbon dioxide purity of the first process node is greater than the purity threshold of the node, that is, the predicted value meets the purity requirements, the predicted value of the weight of the substance generated by the first process node and the purification control parameter are used as input, and the result is passed to the next node for analysis and prediction. Through this node-by-node prediction, judgment and processing, the step-by-step monitoring of the purity of carbon dioxide is realized, and abnormal conditions can be identified and marked in real time, thereby ensuring that the purity of the final product meets the expected high quality requirements.

[0044] When the predicted carbon dioxide purity value is less than or equal to the carbon dioxide purity threshold, the carbon dioxide purification control parameter of the first process node, the carbon dioxide purification control parameter of the second process node, and the carbon dioxide purification control parameter of the Nth process node are marked as abnormal in purity.

[0045] When it is detected that the predicted value of the carbon dioxide purity of a certain process node is less than or equal to the preset carbon dioxide purity threshold, it indicates that the purification effect of the node does not meet the standard. At this time, all carbon dioxide purification control parameters from the first process node to the Nth process node are included in the purity abnormality identification range.

[0046] Furthermore, when the predicted value of carbon dioxide purity is less than or equal to the carbon dioxide purity threshold, the method further includes:

[0047] The user terminal updates the carbon dioxide purification control parameters of the first process node, the second process node, and the Nth process node, and performs purity detection in a loop until no purity abnormality mark is output.

[0048] The information of the abnormal identification is fed back to the user end, allowing the operator or system to update the carbon dioxide purification control parameters of the first process node to the Nth process node as needed. The update of these control parameters may include adjusting the temperature, pressure, flow rate, etc. to optimize the purification effect and improve the purity. The updated control parameters are re-input into the carbon dioxide purity prediction pipeline, and all process nodes are traversed again to recalculate and predict the carbon dioxide purity of each node; the purity prediction value is compared with the threshold again to check whether the abnormal situation has been eliminated; if the updated control parameters still cause the purity prediction value to fail to meet the standard, the control parameters will continue to be further adjusted through the user end; the purity test will be executed cyclically after each update, and the control parameters will be gradually optimized until the purity prediction values ​​of all process nodes are higher than the purity threshold and all abnormal identifications are eliminated; when the purity test no longer triggers any abnormal identification, the final carbon dioxide purity prediction result is output to confirm that the purification process meets the expected purity standard.

[0049] In summary, the embodiments of the present application have at least the following technical effects:

[0050] First, obtain the carbon dioxide purification process flow and the set of raw material component types. Then, disassemble the carbon dioxide purification process flow to obtain the first process node, the second process node, and the Nth process node. Further, according to the carbon dioxide purification process flow and the set of raw material component types, traverse the first process node, the second process node, and the Nth process node to train the carbon dioxide purity prediction model, and obtain the carbon dioxide purity prediction pipeline. Next, receive the raw material component weight ratio of the raw material component type set, as well as the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node. Then, through the carbon dioxide purity prediction pipeline, traverse the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node, and process them in combination with the raw material component weight ratio to obtain the carbon dioxide purity prediction value. Finally, when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold, the carbon dioxide purification control parameters of the first process node, the carbon dioxide purification control parameters of the second process node, and the carbon dioxide purification control parameters of the Nth process node are marked as abnormal in purity. It solves the technical problem that traditional processes cannot predict the purity of carbon dioxide, resulting in a lack of quantitative control methods, and achieves the technical effect of realizing quantitative control of carbon dioxide purity.

[0051] Embodiment 2, based on the same inventive concept as the intelligent detection method for purity of food-grade carbon dioxide in the above embodiment, Figure 2 As shown, the present application provides an intelligent detection device for the purity of food-grade carbon dioxide, wherein the device comprises:

[0052] A data acquisition module 11, the data acquisition module 11 is used to obtain a carbon dioxide purification process flow and a set of raw material component types; a disassembly module 12, the disassembly module 12 is used to disassemble the carbon dioxide purification process flow to obtain a first process node, a second process node, and finally an Nth process node; a training module 13, the training module 13 is used to traverse the first process node, the second process node, and finally the Nth process node according to the carbon dioxide purification process flow and the set of raw material component types to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline; a data receiving module 14, the data receiving module 14 is used to receive the raw material component weight ratio of the raw material component type set, as well as the first process node carbon dioxide purification control parameter, the second process node point carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter; a data processing module 15, the data processing module 15 is used to traverse the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter through the carbon dioxide purity prediction pipeline, and process in combination with the raw material component weight ratio to obtain a carbon dioxide purity prediction value; an abnormality identification module 16, the abnormality identification module 16 is used to mark the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter for purity abnormality when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold.

[0053] Furthermore, the training module 13 is used to perform the following method:

[0054] Taking the carbon dioxide purification process flow and the raw material component type set as constraints, collecting the purification log data of the first process node, wherein the purification log data of the first process node includes the raw material component weight record matrix, the carbon dioxide purification control record parameters and the first record data of the generated substance weight; taking the first record data of the generated substance weight as supervision, and taking the raw material component weight record matrix and the carbon dioxide purification control record parameters as input, training the carbon dioxide purity prediction model of the first process node; looping analysis until obtaining the carbon dioxide purity prediction model of the second process node until the carbon dioxide purity prediction model of the Nth process node; merging the output layer of the carbon dioxide purity prediction model of the first process node with an input node of the carbon dioxide purity prediction model of the second process node until merging the output layer of the carbon dioxide purity prediction model of the N-1th process node with an input node of the carbon dioxide purity prediction model of the Nth process node to generate the carbon dioxide purity prediction pipeline.

[0055] Furthermore, the training module 13 is used to perform the following method:

[0056] Taking the raw material component weight record matrix and the carbon dioxide purification control record parameters as dynamic constraints, and the carbon dioxide purification process and the raw material component type set as static constraints, collect the first process node purification incremental log data, wherein the first process node purification incremental log data includes a plurality of generated substance weight second record data; perform centralized value evaluation on the generated substance weight first record data and the plurality of generated substance weight second record data to obtain the generated substance weight identification value; take the generated substance weight identification value as supervision, and take the raw material component weight record matrix and the carbon dioxide purification control record parameters as input to train the first process node carbon dioxide purity prediction model.

[0057] Furthermore, the training module 13 is used to perform the following method:

[0058] Taking the carbon dioxide purification process flow and the raw material component type set as static constraints, the first-level purification incremental log data of the first process node is collected, wherein the first-level purification incremental log data of the first process node includes a first-level raw material component weight record matrix, a first-level carbon dioxide purification control record parameter and a first-level generated substance weight second record data; the matrix element deviation mean and matrix element deviation variance of the first-level raw material component weight record matrix and the raw material component weight record matrix are compared; the carbon dioxide purification control record parameter and the first-level carbon dioxide purification control record parameter are compared. The proportion of attributes whose control record parameter deviation is greater than or equal to the control parameter deviation threshold is set as the control record parameter deviation degree; when the matrix element deviation mean is less than or equal to the deviation mean threshold, and the matrix element deviation variance is less than or equal to the deviation variance threshold, and the control record parameter deviation degree is less than or equal to the deviation degree threshold, the first-level generated substance weight second record data of the first-level purification incremental log data of the first process node is added to the first process node purification incremental log data.

[0059] Furthermore, the training module 13 is used to perform the following method:

[0060] Arrange the first record data of the weight of the generated substance of any substance type and the second record data of the generated substance weight in order from small to large to obtain the weight of the generated substance with a serial number rounded up to one quarter and the weight of the generated substance with a serial number rounded down to three quarters; calculate the weight of the generated substance with the serial number rounded down to three quarters minus the weight of the generated substance with the serial number rounded up to one quarter, and obtain the range of the weight of the generated substance; calculate the weight of the generated substance with the serial number rounded down to three quarters plus 1.5 times the range of the generated substance weight to obtain the first boundary weight of the generated substance; calculate the weight of the generated substance with the serial number rounded up to one quarter minus 1.5 times the range of the generated substance weight to obtain the second boundary weight of the generated substance; count the average of the weight of the generated substances greater than or equal to the second boundary weight of the generated substance and less than or equal to the first boundary weight of the generated substance in the first record data of the generated substance weight and the second record data of the generated substance weight, and obtain the identification value of the generated substance weight.

[0061] Furthermore, the data processing module 15 is used to execute the following method:

[0062] Through the carbon dioxide purity prediction pipeline, according to the first process node carbon dioxide purification control parameters, combined with the weight ratio of the raw material components, processing is performed to obtain a predicted value of the weight of the substance generated at the first process node, wherein the weight of the substance generated at the first process node includes the predicted value of the carbon dioxide purity at the first process node; when the predicted value of the carbon dioxide purity at the first process node is less than or equal to the carbon dioxide purity threshold of the first process node, the carbon dioxide purification control parameter of the first process node is marked as abnormal in purity; when the predicted value of the carbon dioxide purity at the first process node is greater than the carbon dioxide purity threshold of the first process node, processing is performed through the carbon dioxide purity prediction pipeline according to the predicted value of the weight of the substance generated at the first process node and the carbon dioxide purification control parameter of the first process node.

[0063] Furthermore, the abnormal identification module 16 is used to execute the following method:

[0064] The user terminal updates the carbon dioxide purification control parameters of the first process node, the second process node, and the Nth process node, and performs purity detection in a loop until no purity abnormality mark is output.

[0065] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0066] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0067] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An intelligent method for detecting the purity of food-grade carbon dioxide, characterized in that: The method comprises: Obtain a set of carbon dioxide purification process flow and raw material component types; Disassembling the carbon dioxide purification process flow to obtain a first process node, a second process node, and finally an Nth process node; According to the carbon dioxide purification process and the raw material component type set, traverse the first process node, the second process node until the Nth process node to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline; Receiving a weight ratio of the raw material components of the raw material component type set, and a first process node carbon dioxide purification control parameter, a second process node carbon dioxide purification control parameter, and up to an Nth process node carbon dioxide purification control parameter; Through the carbon dioxide purity prediction pipeline, the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter are traversed, and the raw material component weight ratio is combined for processing to obtain a carbon dioxide purity prediction value; When the predicted carbon dioxide purity value is less than or equal to the carbon dioxide purity threshold, the carbon dioxide purification control parameter of the first process node, the carbon dioxide purification control parameter of the second process node, and the carbon dioxide purification control parameter of the Nth process node are marked as abnormal in purity; According to the carbon dioxide purification process and the raw material component type set, the first process node, the second process node and the Nth process node are traversed to perform carbon dioxide purity prediction model training to obtain a carbon dioxide purity prediction pipeline, including: Taking the carbon dioxide purification process and the raw material component type set as constraints, collecting first process node purification log data, wherein the first process node purification log data includes a raw material component weight record matrix, carbon dioxide purification control record parameters and a first record data of generated substance weight; Using the first recorded data of the weight of the generated substance as supervision, and using the recorded matrix of the weight of the raw material components and the recorded parameters of the carbon dioxide purification control as input, training a first process node carbon dioxide purity prediction model; Circular analysis is performed until a carbon dioxide purity prediction model of a second process node is obtained, until a carbon dioxide purity prediction model of an Nth process node is obtained; The output layer of the carbon dioxide purity prediction model of the first process node is merged with an input node of the carbon dioxide purity prediction model of the second process node, until the output layer of the carbon dioxide purity prediction model of the N-1th process node is merged with an input node of the carbon dioxide purity prediction model of the Nth process node, to generate the carbon dioxide purity prediction pipeline.

2. The method for intelligent detection of purity of food-grade carbon dioxide according to claim 1, characterized in that: Taking the first recorded data of the weight of the generated substance as supervision, taking the raw material component weight record matrix and the carbon dioxide purification control record parameters as input, training the first process node carbon dioxide purity prediction model includes: Taking the raw material component weight record matrix and the carbon dioxide purification control record parameter as dynamic constraints, and taking the carbon dioxide purification process and the raw material component type set as static constraints, collecting the first process node purification incremental log data, wherein the first process node purification incremental log data includes a plurality of generated substance weight second record data; Performing a centralized value evaluation on the first recorded data of the weight portions of the generated substance and the plurality of second recorded data of the weight portions of the generated substance to obtain an identification value of the weight portions of the generated substance; The carbon dioxide purity prediction model of the first process node is trained by taking the weight portion identification value of the generated substance as supervision and taking the weight portion record matrix of the raw material components and the carbon dioxide purification control record parameters as input.

3. The method for intelligent detection of purity of food-grade carbon dioxide according to claim 2, characterized in that: The raw material component weight record matrix and the carbon dioxide purification control record parameters are used as dynamic constraints, and the carbon dioxide purification process and the raw material component type set are used as static constraints to collect the first process node purification incremental log data, wherein the first process node purification incremental log data includes a plurality of generated substance weight second record data, including: Taking the carbon dioxide purification process and the raw material component type set as static constraints, collecting the first-level purification incremental log data of the first process node, wherein the first-level purification incremental log data of the first process node includes a first-level raw material component weight record matrix, a first-level carbon dioxide purification control record parameter and a first-level generated substance weight second record data; Comparing the matrix element deviation mean and the matrix element deviation variance of the primary raw material component weight portion record matrix and the raw material component weight portion record matrix; Compare the carbon dioxide purification control record parameter and the first-level carbon dioxide purification control record parameter, and the proportion of attributes whose control record parameter deviation is greater than or equal to the control parameter deviation threshold, and set it as the control record parameter deviation degree; When the mean deviation of the matrix elements is less than or equal to the mean deviation threshold, and the variance of the matrix elements is less than or equal to the variance deviation threshold, and the deviation degree of the control record parameter is less than or equal to the deviation degree threshold, the second record data of the first-level generated material weight of the first-level purification incremental log data of the first process node is added to the first process node purification incremental log data.

4. The method for intelligent detection of purity of food-grade carbon dioxide according to claim 2, characterized in that: The first recorded data of the weight of the generated substance and the plurality of second recorded data of the weight of the generated substance are evaluated for a centralized value to obtain an identification value of the weight of the generated substance, including: Arrange the first record data of the weight portion of the generated substance of any substance type and the second record data of the weight portions of the generated substance in order from small to large to obtain the weight portion of the generated substance with a sequence number of one quarter rounded up and the weight portion of the generated substance with a sequence number of three quarters rounded down; Calculate the weight of the substance generated by rounding down the three-quarters serial number minus the weight of the substance generated by rounding up the one-quarter serial number to obtain the range of the weight of the generated substance; Calculate the weight of the substance generated by rounding down the three-quarters sequence number plus 1.5 times the range of the weight of the substance generated to obtain the weight of the substance generated by the first boundary; Calculate the weight of the substance generated by rounding up the serial number of one quarter minus 1.5 times the range of the weight of the substance generated to obtain the weight of the substance generated at the second boundary; The first record data of the generated substance weight portions and the average of the generated substance weight portions greater than or equal to the second boundary generated substance weight portions and less than or equal to the first boundary generated substance weight portions among the plurality of second record data of the generated substance weight portions are counted to obtain the generated substance weight portion identification value.

5. The method for intelligent detection of purity of food-grade carbon dioxide according to claim 1, characterized in that: Also includes: Through the carbon dioxide purity prediction pipeline, according to the first process node carbon dioxide purification control parameter, combined with the raw material component weight ratio, a prediction value of the weight of the substance generated at the first process node is obtained, wherein the weight of the substance generated at the first process node includes the prediction value of the carbon dioxide purity at the first process node; When the predicted value of carbon dioxide purity at the first process node is less than or equal to the carbon dioxide purity threshold at the first process node, marking the carbon dioxide purification control parameter at the first process node as abnormal in purity; When the predicted value of carbon dioxide purity of the first process node is greater than the carbon dioxide purity threshold of the first process node, the carbon dioxide purity prediction pipeline is used to generate a predicted value of material weight and a carbon dioxide purification control parameter of the first process node for processing.

6. The method for intelligent detection of purity of food-grade carbon dioxide according to claim 5, characterized in that: When the predicted value of carbon dioxide purity is less than or equal to the carbon dioxide purity threshold, it also includes: updating the carbon dioxide purification control parameters of the first process node, the second process node carbon dioxide purification control parameters until the Nth process node carbon dioxide purification control parameters through the user terminal and cyclically performing purity detection until no purity abnormality mark is output.

7. An intelligent detection device for the purity of food-grade carbon dioxide, characterized in that: A method for intelligently detecting the purity of food-grade carbon dioxide according to any one of claims 1 to 6, the device comprising: A data acquisition module, the data acquisition module is used to obtain a carbon dioxide purification process and a set of raw material component types; A disassembly module, wherein the disassembly module is used to disassemble the carbon dioxide purification process flow to obtain a first process node, a second process node, and up to an Nth process node; A training module, wherein the training module is used to traverse the first process node, the second process node, and finally the Nth process node to perform carbon dioxide purity prediction model training according to the carbon dioxide purification process and the raw material component type set, so as to obtain a carbon dioxide purity prediction pipeline; A data receiving module, the data receiving module is used to receive the raw material component weight ratio of the raw material component type set, and the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter; A data processing module, the data processing module is used to traverse the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter until the Nth process node carbon dioxide purification control parameter through the carbon dioxide purity prediction pipeline, and perform processing in combination with the raw material component weight ratio to obtain a carbon dioxide purity prediction value; An abnormality identification module is used to identify the purity abnormality of the first process node carbon dioxide purification control parameter, the second process node carbon dioxide purification control parameter, and the Nth process node carbon dioxide purification control parameter when the carbon dioxide purity prediction value is less than or equal to the carbon dioxide purity threshold.

Images