Crosslinked polyethylene online production monitoring method and system

Abstract

The invention relates to the technical field of plastic particles, in particular to a crosslinked polyethylene online production monitoring method and system. Firstly, reaction data, production data and image data are obtained; then, identifying the image data by using a crosslinked polyethylene feature identification model, and calculating according to the reaction data, the production data and model output data to obtain a first crosslinking degree monitoring value and a second defect feature monitoring value; the first crosslinking degree monitoring value and the second defect characteristic monitoring value are used for evaluating the crosslinking degree and the defect degree, so that the accuracy and the reliability of online production monitoring of the crosslinked polyethylene are improved; then, comprehensively evaluating the first crosslinking degree monitoring value and the second defect characteristic monitoring value to obtain a production quality monitoring value; and finally, comparing the production quality monitoring value with a production quality monitoring threshold value, if the production quality monitoring value exceeds the threshold value, judging that the product is unqualified, otherwise, generating environment temperature monitoring information according to a data result of the temperature prediction model.

Description

Technical Field

[0001] The invention relates to the technical field of plastic particles, and in particular to an online production monitoring method and system for cross-linked polyethylene. Background Art

[0002] Cross-linked polyethylene (XLPE) is a polymer material that forms a three-dimensional network structure by cross-linking polyethylene molecular chains through chemical or physical methods. Compared with ordinary polyethylene, cross-linked polyethylene has higher heat resistance, insulation, chemical resistance and mechanical strength. Cross-linked polyethylene is widely used in cable insulation materials, pipelines and building materials. However, online monitoring of the cross-linking process has always been a difficult problem in the production process. Effective monitoring of the cross-linking reaction process is crucial to ensure the performance and quality of the product.

[0003] However, the existing technology still has deficiencies in the online production monitoring of cross-linked polyethylene. On the one hand, the existing technology does not consider the defect degree and cross-linking time of cross-linked polyethylene during the monitoring process, and the consideration of factors is not comprehensive, which will lead to reduced accuracy and reliability of online production monitoring of cross-linked polyethylene; on the other hand, the existing technology does not consider the impact of ambient temperature on cross-linking and production, and does not provide effective ambient temperature monitoring, which will lead to reduced accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0004] Therefore, a method and system for online production monitoring of cross-linked polyethylene are proposed. Summary of the invention

[0005] The object of the present invention is to provide a method and system for online production monitoring of cross-linked polyethylene. First, reaction data, production data and image data are obtained; then, the image data is identified using a cross-linked polyethylene feature recognition model, and a first cross-linking degree monitoring value and a second defect feature monitoring value are calculated based on the reaction data, the production data and the model output data; the first cross-linking degree monitoring value and the second defect feature monitoring value are used to evaluate the cross-linking degree and the defect degree, so as to improve the accuracy and reliability of online production monitoring of cross-linked polyethylene; then, the first cross-linking degree monitoring value and the second defect feature monitoring value are comprehensively evaluated to obtain a production quality monitoring value; finally, the production quality monitoring value is compared with a production quality monitoring threshold value, and if it exceeds the threshold value, it is judged as unqualified, otherwise, environmental temperature monitoring information is generated according to the data results of the temperature prediction model.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for online production monitoring of cross-linked polyethylene, comprising:

[0008] Acquire reaction data of reaction sensors, production data of production sensors, and image data captured by microscope cameras;

[0009] The image data is processed using a cross-linked polyethylene feature recognition model to obtain plastic particle distribution data and defect feature data; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features;

[0010] The reaction data, the production data and the plastic particle distribution data are evaluated to calculate a first cross-linking degree monitoring value;

[0011] The color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature are evaluated to calculate a second defect feature monitoring value;

[0012] Performing a weighted summation on the first crosslinking degree monitoring value and the second defect characteristic monitoring value to obtain a production quality monitoring value;

[0013] The production quality monitoring value is compared with the production quality threshold. If it is higher than the production quality threshold, it is judged as unqualified; otherwise, the real-time monitoring data is input into the temperature prediction model to generate environmental temperature monitoring information based on the data results.

[0014] Furthermore, the specific implementation process of processing the image data using the cross-linked polyethylene feature recognition model to obtain the plastic particle distribution data and defect feature data includes:

[0015] Performing denoising processing on the image data to obtain denoised image data;

[0016] Using histogram equalization to adjust the contrast and brightness of the denoised image data to obtain enhanced image data;

[0017] Performing edge detection on the enhanced image data to obtain enhanced edge image data;

[0018] The enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition, thereby obtaining plastic particle distribution data and defect feature data.

[0019] Furthermore, the specific implementation process of evaluating the reaction data, the production data and the plastic particle distribution data and calculating the first crosslinking degree monitoring value includes:

[0020] Acquire reaction data of the sensor in the reaction device and production data of the sensor in the production device; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure;

[0021] Obtaining plastic particle distribution data outputted by the cross-linked polyethylene feature recognition model; wherein the plastic particle distribution data includes: the number of plastic particle distributions and the plastic particle distribution area;

[0022] Compare the cross-linking agent concentration, the reaction temperature and the reaction time with the cross-linking agent concentration standard value, the reaction temperature standard value and the reaction time standard value, respectively, to obtain a reaction monitoring value;

[0023] Comparing the production temperature and the production pressure with a standard production temperature value and a standard production pressure value, respectively, to obtain a production monitoring value;

[0024] Compare the plastic particle density monitoring value with the plastic particle density threshold to obtain a plastic particle distribution monitoring value; wherein the plastic particle density monitoring value is a ratio of the number of plastic particle distributions to the plastic particle distribution area;

[0025] The reaction monitoring value, the production monitoring value and the plastic particle distribution monitoring value are weightedly evaluated to obtain a first cross-linking degree monitoring value.

[0026] Furthermore, the specific implementation process of evaluating the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature and calculating the second defect feature monitoring value includes:

[0027] The image data is input into the cross-linked polyethylene feature recognition model for recognition processing;

[0028] Acquire defect features output by the cross-linked polyethylene feature recognition model; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features;

[0029] Respectively acquiring color information, bubble information, hole information, crack information and glue block information from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature;

[0030] The color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value.

[0031] Furthermore, the real-time monitoring data is input into the temperature prediction model, and the specific implementation process of generating the ambient temperature monitoring information according to the data results includes:

[0032] Obtain historical monitoring data from the historical database;

[0033] Acquire real-time monitoring data; wherein the real-time monitoring data includes: real-time crosslinking degree monitoring data and real-time defect characteristic monitoring data;

[0034] Inputting the historical monitoring data into a temperature prediction model for training to obtain temperature prediction parameters;

[0035] Input the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data into the temperature prediction model, update the model parameters, and output the real-time temperature parameters; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters;

[0036] The real-time temperature monitoring information and the real-time temperature parameters are comprehensively evaluated to obtain the ambient temperature monitoring information.

[0037] A cross-linked polyethylene online production monitoring system, comprising: an equipment control module, a data acquisition module, a data monitoring module and an output module;

[0038] Wherein, the device control module is used to control the start, pause and stop of the device;

[0039] The data acquisition module is used to obtain image data collected by the microscopic camera, reaction data and production data sent by the sensor, and historical monitoring data in the historical database;

[0040] The data monitoring module is used to process the image data, the reaction data and the production data to obtain monitoring data; wherein the data monitoring module includes: a cross-linking degree monitoring unit and a defect feature monitoring unit;

[0041] The output module is used to evaluate the monitoring data, output monitoring results and provide monitoring feedback; wherein the monitoring output module includes: a judgment unit and an ambient temperature monitoring unit.

[0042] Furthermore, the data monitoring module processes the image data using the cross-linked polyethylene feature recognition model to obtain the plastic particle distribution data and defect feature data. The specific implementation process includes:

[0043] Performing denoising processing on the image data to obtain denoised image data;

[0044] Using histogram equalization to adjust the contrast and brightness of the denoised image data to obtain enhanced image data;

[0045] Performing edge detection on the enhanced image data to obtain enhanced edge image data;

[0046] The enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition, thereby obtaining plastic particle distribution data and defect feature data.

[0047] Furthermore, the crosslinking degree monitoring unit evaluates the reaction data, the production data and the plastic particle distribution data to obtain the first crosslinking degree monitoring value, and the specific implementation process includes:

[0048] Acquire reaction data of the sensor in the reaction device and production data of the sensor in the production device; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure;

[0049] Obtaining plastic particle distribution data outputted by the cross-linked polyethylene feature recognition model; wherein the plastic particle distribution data includes: the number of plastic particle distributions and the plastic particle distribution area;

[0050] Comparing the cross-linking agent concentration, the reaction temperature and the reaction time with the cross-linking agent concentration standard value, the reaction temperature standard value and the reaction time standard value, respectively, to obtain a reaction monitoring value;

[0051] Comparing the production temperature and the production pressure with a standard production temperature value and a standard production pressure value, respectively, to obtain a production monitoring value;

[0052] Compare the plastic particle density monitoring value with the plastic particle density threshold to obtain a plastic particle distribution monitoring value; wherein the plastic particle density monitoring value is a ratio of the number of plastic particle distributions to the plastic particle distribution area;

[0053] The reaction monitoring value, the production monitoring value and the plastic particle distribution monitoring value are weightedly evaluated to obtain a first cross-linking degree monitoring value.

[0054] Furthermore, the defect feature monitoring unit evaluates the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature to obtain the second defect feature monitoring value, and the specific implementation process includes:

[0055] Inputting the image data into the cross-linked polyethylene feature recognition model for recognition processing;

[0056] Acquire defect features output by the cross-linked polyethylene feature recognition model; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features;

[0057] Respectively acquiring color information, bubble information, hole information, crack information and glue block information from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature;

[0058] The color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value.

[0059] Furthermore, the environmental temperature monitoring unit inputs the real-time monitoring data into the temperature prediction model, and the specific implementation process of generating environmental temperature monitoring information according to the data results includes:

[0060] Obtain historical monitoring data from the historical database;

[0061] Acquire real-time monitoring data; wherein the real-time monitoring data includes: real-time crosslinking degree monitoring data and real-time defect characteristic monitoring data;

[0062] Inputting the historical monitoring data into a temperature prediction model for training to obtain temperature prediction parameters;

[0063] Input the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data into the temperature prediction model, update the model parameters, and output the real-time temperature parameters; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters;

[0064] The real-time temperature monitoring information and the real-time temperature parameters are comprehensively evaluated to obtain the ambient temperature monitoring information.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. The present invention proposes a cross-linking degree monitoring function for monitoring the change of the cross-linking degree of cross-linked polyethylene; the function evaluates the reaction data, production data and plastic particle distribution data to obtain the reaction monitoring value, production monitoring value and plastic particle distribution monitoring value respectively; then, the function uses the weighted first cross-linking degree monitoring value to record the cross-linking degree of the cross-linked polyethylene; the function reflects the change of the cross-linking degree by monitoring the data from different influencing factors, so as to improve the accuracy and reliability of the online production monitoring of the cross-linked polyethylene.

[0067] 2. The present invention proposes a defect feature monitoring function for performing feature evaluation on cross-linked polyethylene defects; this function performs weighted evaluation on the defect features and defect feature coefficients output by the cross-linked polyethylene feature recognition model, and uses the calculated second defect feature monitoring value to record the defect degree of the cross-linked polyethylene; the defect features include: color features, bubble features, hole features, crack features and rubber block features; this function evaluates and monitors different defect features, which can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0068] 3. The present invention proposes an environmental temperature monitoring function for monitoring the impact of environmental temperature on the production of cross-linked polyethylene; this function uses historical monitoring data to train a temperature prediction model to obtain temperature prediction parameters; uses the temperature prediction parameters and real-time monitoring data to update the model parameters to obtain real-time temperature parameters; this function evaluates the real-time temperature parameters and real-time temperature monitoring information to obtain environmental temperature monitoring information; the temperature parameters reflect the correlation between the monitoring data and the ambient temperature; this function can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic flow chart of a method for online production monitoring of cross-linked polyethylene of the present invention;

[0070] Figure 2 A schematic diagram of a cross-linked polyethylene production sample of the present invention;

[0071] Figure 3 It is a structural schematic diagram of the cross-linked polyethylene feature recognition model of the present invention;

[0072] Figure 4 It is a structural schematic diagram of a cross-linked polyethylene online production monitoring system of the present invention. DETAILED DESCRIPTION

[0073] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] Cross-linked polyethylene (XLPE) is a polymer material that forms a three-dimensional network structure by cross-linking polyethylene molecular chains through chemical or physical methods. Compared with ordinary polyethylene, cross-linked polyethylene has higher heat resistance, insulation, chemical resistance and mechanical strength. Cross-linked polyethylene is widely used in cable insulation materials, pipelines and building materials. However, online monitoring of the cross-linking process has always been a difficult problem in the production process. Effective monitoring of the cross-linking reaction process is crucial to ensure the performance and quality of the product.

[0075] However, the existing technology still has shortcomings in the online production monitoring of cross-linked polyethylene. On the one hand, the existing technology does not fully consider the influencing factors in the cross-linking process and production monitoring process, which reduces the accuracy and reliability of the online production monitoring of cross-linked polyethylene; on the other hand, the existing technology does not provide abnormal traceability monitoring information based on the correlation between historical abnormal data and historical abnormal traceability, which will reduce the accuracy and reliability of the online production monitoring of cross-linked polyethylene.

[0076] Embodiment 1

[0077] In the embodiment of the present application, the specific implementation process will be implemented by a cross-linked polyethylene online production monitoring method of the present invention, see Figure 1 The process of the method proposed in the present invention is described in the following; a method for online production monitoring of cross-linked polyethylene comprises:

[0078] S10. Obtaining reaction data, production data, image data and historical abnormal data;

[0079] S20. Inputting the image data into a cross-linked polyethylene feature recognition model to obtain plastic particle distribution data and defect feature data;

[0080] S30. Evaluate the reaction data, the production data and the plastic particle distribution data to obtain a first crosslinking degree monitoring value;

[0081] S40. Evaluate the defect feature data output by the cross-linked polyethylene feature recognition model to obtain a second defect feature monitoring value;

[0082] S50. Comprehensively evaluate the first crosslinking degree monitoring value and the second defect characteristic monitoring value to obtain a production quality monitoring value;

[0083] S60. Compare the production quality monitoring value with the production quality threshold. If it is higher than the production quality threshold, it is judged as unqualified; otherwise, the real-time monitoring data is input into the temperature prediction model to generate environmental temperature monitoring information based on the data results.

[0084] Furthermore, a specific implementation process of a cross-linked polyethylene online production monitoring method is as follows:

[0085] Acquire reaction data of reaction sensors, production data of production sensors, and image data captured by microscope cameras;

[0086] The image data is processed using a cross-linked polyethylene feature recognition model to obtain plastic particle distribution data and defect feature data; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features;

[0087] The reaction data, the production data and the plastic particle distribution data are evaluated to calculate a first cross-linking degree monitoring value;

[0088] The color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature are evaluated to calculate a second defect feature monitoring value;

[0089] Performing a weighted summation on the first crosslinking degree monitoring value and the second defect characteristic monitoring value to obtain a production quality monitoring value;

[0090] The production quality monitoring value is compared with the production quality threshold. If it is higher than the production quality threshold, it is judged as unqualified; otherwise, the real-time monitoring data is input into the temperature prediction model to generate environmental temperature monitoring information based on the data results.

[0091] In this embodiment, a cross-linked polyethylene online production monitoring method is proposed to monitor the quality of cross-linked polyethylene. Figure 2 As shown. First, reaction data, production data and image data are obtained; then, the image data is identified using a cross-linked polyethylene feature recognition model, and a first cross-linking degree monitoring value and a second defect feature monitoring value are calculated based on the reaction data, the production data and the model output data; the first cross-linking degree monitoring value and the second defect feature monitoring value are used to evaluate the cross-linking degree and the defect degree to improve the accuracy and reliability of online production monitoring of cross-linked polyethylene; then, the first cross-linking degree monitoring value and the second defect feature monitoring value are comprehensively evaluated to obtain a production quality monitoring value; finally, the production quality monitoring value is compared with a production quality monitoring threshold, and if the threshold is exceeded, it is judged as unqualified, otherwise, environmental temperature monitoring information is generated according to the data results of the temperature prediction model; this method can improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0092] Acquire reaction data of reaction sensors, production data of production sensors, and image data captured by microscope cameras;

[0093] Furthermore, the specific implementation process of processing the image data using the cross-linked polyethylene feature recognition model to obtain the plastic particle distribution data and defect feature data includes:

[0094] De-noising the image data collected by the microscope camera to obtain de-noised image data;

[0095] Using histogram equalization to adjust the contrast and brightness of the denoised image data to obtain enhanced image data;

[0096] Performing edge detection on the enhanced image data to obtain enhanced edge image data;

[0097] The enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition, thereby obtaining plastic particle distribution data and defect feature data.

[0098] The cross-linked polyethylene feature recognition model in this embodiment is a deep learning model based on a convolutional neural network, and its structure is as follows: Figure 3 As shown, it mainly includes: an input layer, a feature extraction layer, a feature marking layer, a feature recognition layer and an output layer; the input layer converts the image data collected by the microscope camera from the image space to the feature space to facilitate subsequent feature processing; the feature extraction layer uses 3×3, 5×5 and 7×7 convolution operations to extract multi-scale features and fuses the multi-scale features through channel merging operations; the feature marking layer is used to obtain the anchor frame of the cross-linked polyethylene feature; the feature recognition layer recognizes the target feature according to the anchor frame; the output layer is used to output the target feature; wherein the target feature includes the distribution feature of plastic particles and the defect feature.

[0099] The preprocessing process of image data in this embodiment includes: Gaussian denoising, histogram equalization and edge detection; the Gaussian denoising is used to reduce the interference noise of image data to improve the accuracy of the feature recognition model; the histogram equalization is used to enhance the contrast and brightness of the image to improve the detection and recognition capabilities of edge and sharp image information; the edge detection is used to improve the feature recognition model's ability to detect the structure of plastic particles and defect details such as cracks and bubbles; this process can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0100] Furthermore, the specific implementation process of evaluating the reaction data, the production data and the plastic particle distribution data and calculating the first crosslinking degree monitoring value includes:

[0101] Acquire reaction data of the sensor in the reaction device and production data of the sensor in the production device; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure;

[0102] Further, obtaining the plastic particle distribution data outputted by the cross-linked polyethylene feature recognition model; wherein the plastic particle distribution data includes: the number of plastic particle distributions and the plastic particle distribution area;

[0103] Further, the cross-linking agent concentration, the reaction temperature and the reaction time are respectively compared with the cross-linking agent concentration standard value, the reaction temperature standard value and the reaction time standard value to obtain a reaction monitoring value; wherein the calculation formula of the reaction monitoring value is:

[0104]

[0105] α1+α2+α3=1.0;

[0106] Among them, FYJC i It is represented as the reaction monitoring value of the ith data. The larger the value, the closer the actual reaction is to the standard reaction process. α1 is represented as the weight factor of the cross-linking agent concentration. nd i The cross-linking agent concentration represented by the i-th data; nd th represents the standard value of the cross-linking agent concentration; α2 represents the weight factor of the reaction temperature; wd i Represented as the reaction temperature of the i-th data; wd th α3 represents the weight factor of the reaction time; sj represents the standard value of the reaction temperature; i Represented as the reaction time of the i-th data; sj th It is indicated as the standard value of the reaction time.

[0107] In this embodiment, the weights of the reaction monitoring values ​​are set to 0.35, 0.35 and 0.3 respectively.

[0108] In this embodiment, the standard value of the cross-linking agent concentration, the standard value of the reaction temperature and the standard value of the reaction time are set by relevant personnel in this field based on historical experience. Therefore, the setting of the standard values ​​is not unique and can be adjusted according to actual conditions.

[0109] Furthermore, the production temperature and the production pressure are compared with the production temperature standard value and the production pressure standard value respectively to obtain the production monitoring value; wherein the calculation formula of the production monitoring value is:

[0110]

[0111] Among them, SCJC i It is represented as the production monitoring value of the i-th data. The larger the value, the closer the actual production is to the standard production process; β is represented as the weight factor of the production temperature; pt i Represented as the production temperature of the i-th data; pt th Expressed as the standard value of the production temperature; pp i The production pressure represented by the i-th data; pp th Indicated as standard value of the production pressure.

[0112] In this embodiment, the weight of the production monitoring value is set to 0.5.

[0113] Further, the plastic particle density monitoring value is compared with the plastic particle density threshold value to obtain a plastic particle distribution monitoring value; wherein the plastic particle density monitoring value is the ratio of the number of plastic particle distributions to the plastic particle distribution area;

[0114] Furthermore, the reaction monitoring value, the production monitoring value and the plastic particle distribution monitoring value are weightedly evaluated to obtain a first crosslinking degree monitoring value; wherein the calculation formula of the first crosslinking degree monitoring value is:

[0115]

[0116] γ1+γ2+γ3=1.0;

[0117] Wherein, DCJC represents the first cross-linking degree monitoring value, the larger the value, the closer the actual cross-linking degree is to the standard cross-linking degree; N represents the number of monitoring data; γ1 represents the weight factor of the reaction monitoring value; FYJC i is represented by the reaction monitoring value of the i-th data; γ2 is represented by the weight factor of the production monitoring value; SCJC i γ3 represents the weight factor of the plastic particle density monitoring value; sl represents the production monitoring value of the i-th data; i mj is the number of plastic particles distributed as the i-th data; i The plastic particle distribution area represented by the i-th data; cc th Expressed as the plastic particle density threshold.

[0118] In this embodiment, the weights of the first cross-linking degree monitoring values ​​are set to 0.4, 0.3 and 0.3 respectively.

[0119] The weight factors allocated in the calculation formulas of the reaction monitoring value, the production monitoring value and the first cross-linking degree monitoring value in this embodiment can be adjusted according to actual conditions.

[0120] In this embodiment, a cross-linking degree monitoring function is proposed to monitor the changes in the degree of cross-linking of cross-linked polyethylene; this function evaluates the reaction data, production data and plastic particle distribution data to obtain reaction monitoring values, production monitoring values ​​and plastic particle distribution monitoring values ​​respectively; then, this function uses the weighted first cross-linking degree monitoring value to record the cross-linking degree of the cross-linked polyethylene; this function reflects the changes in the degree of cross-linking by monitoring data from different influencing factors to improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0121] Furthermore, the specific implementation process of evaluating the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature and calculating the second defect feature monitoring value includes:

[0122] The image data is input into the cross-linked polyethylene feature recognition model for recognition processing;

[0123] Further, the defect features output by the cross-linked polyethylene feature recognition model are obtained; wherein the defect feature data include: color features, bubble features, hole features, crack features and rubber block features;

[0124] Further, color information, bubble information, hole information, crack information and glue block information are obtained from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature respectively;

[0125] Further, the color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value; wherein the calculation formula of the second defect characteristic monitoring value is:

[0126]

[0127] Wherein, TZJC represents the second defect characteristic monitoring value; M represents the number of characteristic samples; η1 represents the color characteristic coefficient; ys j It is represented as the color uniformity of the jth sample; η2 is represented as the bubble characteristic coefficient; qp j Expressed as the maximum number of bubbles in the jth sample; is the diameter of the j1th bubble of the jth sample; η3 is the hole characteristic coefficient; kd j It is expressed as the maximum number of holes in the jth sample; It is expressed as the area of ​​the j2th hole of the jth sample; is the depth of the j2nd hole of the jth sample; η4 is the crack characteristic coefficient; lw j It is expressed as the maximum number of cracks in the jth sample; It is expressed as the length of the j3rd crack of the jth sample; It is represented by the width of the j3rd crack of the jth sample; η5 is represented by the characteristic coefficient of the rubber block; jk j It is expressed as the maximum number of glue blocks of the jth sample; It is expressed as the length of the j4th glue block of the jth sample; Expressed as the width of the j4th glue block of the jth sample.

[0128] In this embodiment, the shape of the rubber block is set to a rectangle to facilitate feature calculation; the color characteristic coefficient, bubble characteristic coefficient, hole characteristic coefficient, crack characteristic coefficient and rubber block characteristic coefficient are set according to the different degrees of influence on the quality monitoring of cross-linked polyethylene, and are set to 0.3, 0.1, 0.2, 0.1 and 0.3 respectively; the setting of different characteristic coefficients can be adjusted according to actual conditions.

[0129] In this embodiment, a defect feature monitoring function is proposed for performing feature evaluation on cross-linked polyethylene defects; this function performs weighted evaluation on the defect features and defect feature coefficients output by the cross-linked polyethylene feature recognition model, and uses the calculated second defect feature monitoring value to record the defect degree of the cross-linked polyethylene; the defect features include: color features, bubble features, hole features, crack features and rubber block features; this function evaluates and monitors different defect features, which can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0130] Further, the first cross-linking degree monitoring value and the second defect characteristic monitoring value are comprehensively evaluated to obtain a production quality monitoring value;

[0131] Furthermore, the production quality monitoring value is compared with the production quality threshold, and if it is higher than the production quality threshold, it is judged as unqualified; otherwise, the real-time monitoring data is input into the temperature prediction model, and the environmental temperature monitoring information is generated according to the data results;

[0132] The specific implementation process of the ambient temperature monitoring information includes:

[0133] Obtain historical monitoring data from the historical database;

[0134] Further, real-time monitoring data is obtained; wherein the real-time monitoring data includes: real-time cross-linking degree monitoring data and real-time defect characteristic monitoring data;

[0135] Further, the historical monitoring data is input into a temperature prediction model for training to obtain temperature prediction parameters;

[0136] Further, the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data are input into the temperature prediction model, the model parameters are updated, and the real-time temperature parameters are output; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters;

[0137] The temperature prediction model in this embodiment uses a hybrid network combining CNN and Transformer; the hybrid network adopts a dual-branch structure, and the input data is input into the two sub-networks of CNN and Transformer for feature processing, and finally the output features of different networks are fused to obtain the prediction result; CNN can capture spatial context features and use local convolution operations to learn the contextual relationship between monitoring data; Transformer can extract long-distance dependency features and use the self-attention mechanism to learn the global relationship between temperature and monitoring data.

[0138] Furthermore, the real-time temperature monitoring information and the real-time temperature parameter are comprehensively evaluated to obtain the ambient temperature monitoring information; wherein the calculation formula of the ambient temperature monitoring information is:

[0139]

[0140] Wherein, ETJC represents the environmental temperature monitoring information; P represents the number of data samples; ω (i,fy) INF is the real-time reaction temperature parameter of the i-th sample; (i,fy) Represented as the real-time reaction temperature information of the i-th sample; ω (i,sc) INF is the real-time production temperature parameter of the i-th sample; (i,sc) Represented as the real-time production temperature information of the i-th sample.

[0141] In this embodiment, an ambient temperature monitoring function is proposed to monitor the impact of ambient temperature on the production of cross-linked polyethylene; this function uses historical monitoring data to train a temperature prediction model to obtain temperature prediction parameters; the model parameters are updated using the temperature prediction parameters and real-time monitoring data to obtain real-time temperature parameters; this function evaluates the real-time temperature parameters and real-time temperature monitoring information to obtain ambient temperature monitoring information; the temperature parameters reflect the correlation between the monitoring data and the ambient temperature; this function can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0142] Embodiment 2

[0143] As an embodiment of the present invention, refer to Figure 4 ,A cross-linked polyethylene online production monitoring system comprises: an equipment control module, a data acquisition module, a data monitoring module and an output module;

[0144] Wherein, the device control module is used to control the start, pause and stop of the device;

[0145] The data acquisition module is used to obtain image data collected by the microscope camera, reaction data and production data sent by the sensor, and historical monitoring data in the historical database; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure;

[0146] The data monitoring module is used to process the image data, reaction data and production data to obtain monitoring data; wherein the data monitoring module uses the cross-linked polyethylene feature recognition model to process the image data to obtain the plastic particle distribution data and defect feature data. The specific implementation process includes:

[0147] Performing denoising processing on the image data to obtain denoised image data;

[0148] Further, the contrast and brightness of the denoised image data are adjusted using histogram equalization to obtain enhanced image data;

[0149] Further, edge detection is performed on the enhanced image data to obtain enhanced edge image data;

[0150] Furthermore, the enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition to obtain plastic particle distribution data and defect feature data.

[0151] The cross-linked polyethylene feature recognition model in this embodiment is a deep learning model based on a convolutional neural network, and mainly includes: an input layer, a feature extraction layer, a feature marking layer, a feature recognition layer and an output layer; the input layer converts the image data collected by the microscope camera from the image space to the feature space to facilitate subsequent feature processing; the feature extraction layer uses convolution operations of different sizes to extract multi-scale features and fuses the multi-scale features through channel merging operations; the feature marking layer is used to obtain the anchor frame of the cross-linked polyethylene feature; the feature recognition layer recognizes the target feature according to the anchor frame; the output layer is used to output the target feature; wherein the target feature includes plastic particle distribution features and defect features.

[0152] The data monitoring module includes: a cross-linking degree monitoring unit and a defect characteristic monitoring unit;

[0153] The specific implementation process of the cross-linking degree monitoring unit in this embodiment includes: first, obtaining reaction data, production data and plastic particle distribution data; wherein the plastic particle distribution data is obtained by identifying image data through a cross-linked polyethylene feature recognition model; then, evaluating the reaction data, the production data and the plastic particle distribution data to obtain a first cross-linking degree monitoring value.

[0154] The calculation formula of the first cross-linking degree monitoring value is:

[0155]

[0156] γ1+γ2+γ3=1.0;

[0157] Wherein, DCJC represents the first cross-linking degree monitoring value, the larger the value, the closer the actual cross-linking degree is to the standard cross-linking degree; N represents the number of monitoring data; γ1 represents the weight factor of the reaction monitoring value; FYJC i It is represented as the reaction monitoring value of the i-th data; γ2 is represented as the weight factor of the production monitoring value; SCJC i It is represented as the production monitoring value of the ith data; γ3 is represented as the weight factor of the plastic particle density monitoring value; sl i Represented as the number of plastic particles distributed in the i-th data; mj i Represented as the plastic particle distribution area of ​​the i-th data; cc th Expressed as plastic particle density threshold.

[0158] Wherein, the calculation formula of the reaction monitoring value is:

[0159]

[0160] α1+α2+α3=1.0;

[0161] Among them, FYJC i It is represented as the reaction monitoring value of the ith data. The larger the value, the closer the actual reaction is to the standard reaction process. α1 is represented as the weight factor of the cross-linking agent concentration. nd i It is represented as the cross-linker concentration of the ith data; nd th is the standard value of the cross-linking agent concentration; α2 is the weighting factor of the reaction temperature; wd i Represents the reaction temperature of the i-th data; wd th It is the standard value of reaction temperature; α3 is the weight factor of reaction time; sj i It is expressed as the reaction time of the i-th data; sj th Expressed as the standard value of reaction time.

[0162] The calculation formula of the production monitoring value is:

[0163]

[0164] Among them, SCJC i It is represented as the production monitoring value of the i-th data. The larger the value, the closer the actual production is to the standard production process; β is represented as the weight factor of the production temperature; pt iRepresented as the production temperature of the i-th data; pt th Expressed as the standard value of production temperature; pp i Expressed as the production pressure of the i-th data; pp th Expressed as standard value of production pressure.

[0165] Furthermore, the defect feature monitoring unit evaluates the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature to obtain the second defect feature monitoring value, and the specific implementation process includes:

[0166] Inputting the image data into the cross-linked polyethylene feature recognition model for recognition processing;

[0167] Further, the defect features output by the cross-linked polyethylene feature recognition model are obtained; wherein the defect feature data include: color features, bubble features, hole features, crack features and rubber block features;

[0168] Further, color information, bubble information, hole information, crack information and glue block information are obtained from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature respectively;

[0169] Further, the color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value; wherein the calculation formula of the second defect characteristic monitoring value is:

[0170]

[0171] η1+η2+η3+η4+η5=1.0;

[0172] Wherein, TZJC represents the second defect characteristic monitoring value; M represents the number of characteristic samples; η1 represents the color characteristic coefficient; ys j It is represented as the color uniformity of the jth sample; η2 is represented as the bubble characteristic coefficient; qp j Expressed as the maximum number of bubbles in the jth sample; is the diameter of the j1th bubble of the jth sample; η3 is the hole characteristic coefficient; kd j It is expressed as the maximum number of holes in the jth sample; It is expressed as the area of ​​the j2th hole of the jth sample; is the depth of the j2nd hole of the jth sample; η4 is the crack characteristic coefficient; lw j It is expressed as the maximum number of cracks in the jth sample; It is expressed as the length of the j3rd crack of the jth sample; It is represented by the width of the j3rd crack of the jth sample; η5 is represented by the characteristic coefficient of the rubber block; jk j It is expressed as the maximum number of glue blocks of the jth sample; It is expressed as the length of the j4th glue block of the jth sample; Expressed as the width of the j4th glue block of the jth sample.

[0173] The output module is used to evaluate the monitoring data, output monitoring results and provide monitoring feedback; wherein the monitoring output module includes: a judgment unit and an ambient temperature monitoring unit.

[0174] The specific implementation process of the ambient temperature monitoring unit inputting the real-time monitoring data into the temperature prediction model and generating ambient temperature monitoring information according to the data results includes:

[0175] Obtain historical monitoring data from the historical database;

[0176] Further, real-time monitoring data is obtained; wherein the real-time monitoring data includes: real-time cross-linking degree monitoring data and real-time defect characteristic monitoring data;

[0177] Further, the historical monitoring data is input into a temperature prediction model for training to obtain temperature prediction parameters;

[0178] Further, the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data are input into the temperature prediction model, the model parameters are updated, and the real-time temperature parameters are output; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters;

[0179] Furthermore, the real-time temperature monitoring information and the real-time temperature parameter are comprehensively evaluated to obtain the ambient temperature monitoring information. The calculation formula of the ambient temperature monitoring information is:

[0180]

[0181] Wherein, JTJC represents the ambient temperature monitoring information; P represents the number of data samples; ω (i,fy) INF is the real-time reaction temperature parameter of the i-th sample; (i,fy) Represented as the real-time reaction temperature information of the i-th sample; ω (i,sc) INF is the real-time production temperature parameter of the i-th sample; (i,sc) Represented as the real-time production temperature information of the i-th sample.

[0182] In this embodiment, a cross-linked polyethylene online production monitoring system is proposed for monitoring the online production quality of cross-linked polyethylene; the system analyzes and monitors the internal structure, cross-linking state and defect conditions of the cross-linked polyethylene, and uses a cross-linked polyethylene feature recognition model to identify and process image data to obtain distribution characteristics and defect characteristics; the system monitors the quality of cross-linked polyethylene from two perspectives: the degree of cross-linking and the degree of defects, and obtains a production quality monitoring value by comprehensively evaluating the calculated first degree of cross-linking monitoring value and the second defect feature monitoring value; then, the production quality monitoring value is compared with the production quality threshold, and the monitoring result is output; the system can effectively improve the accuracy and reliability of online production monitoring of cross-linked polyethylene.

[0183] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for online production monitoring of cross-linked polyethylene, characterized in that: include: Acquire reaction data of reaction sensors, production data of production sensors, and image data captured by microscope cameras; The image data is processed using a cross-linked polyethylene feature recognition model to obtain plastic particle distribution data and defect feature data; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features; The reaction data, the production data and the plastic particle distribution data are evaluated to calculate a first cross-linking degree monitoring value; The color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature are evaluated to calculate a second defect feature monitoring value; Performing a weighted summation on the first crosslinking degree monitoring value and the second defect characteristic monitoring value to obtain a production quality monitoring value; The production quality monitoring value is compared with the production quality threshold. If it is higher than the production quality threshold, it is judged as unqualified; otherwise, the real-time monitoring data is input into the temperature prediction model to generate environmental temperature monitoring information based on the data results.

2. A method for online production monitoring of cross-linked polyethylene according to claim 1, characterized in that: The specific implementation process of processing the image data using the cross-linked polyethylene feature recognition model to obtain plastic particle distribution data and defect feature data includes: De-noising the image data collected by the microscope camera to obtain de-noised image data; Using histogram equalization to adjust the contrast and brightness of the denoised image data to obtain enhanced image data; Performing edge detection on the enhanced image data to obtain enhanced edge image data; The enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition, thereby obtaining plastic particle distribution data and defect feature data.

3. A method for online production monitoring of cross-linked polyethylene according to claim 1, characterized in that: The specific implementation process of evaluating the reaction data, the production data and the plastic particle distribution data and calculating the first crosslinking degree monitoring value includes: Acquire reaction data of the sensor in the reaction device and production data of the sensor in the production device; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure; Obtaining plastic particle distribution data outputted by the cross-linked polyethylene feature recognition model; wherein the plastic particle distribution data includes: the number of plastic particle distributions and the plastic particle distribution area; Comparing the cross-linking agent concentration, the reaction temperature and the reaction time with the cross-linking agent concentration standard value, the reaction temperature standard value and the reaction time standard value, respectively, to obtain a reaction monitoring value; Comparing the production temperature and the production pressure with a standard production temperature value and a standard production pressure value, respectively, to obtain a production monitoring value; Compare the plastic particle density monitoring value with the plastic particle density threshold to obtain a plastic particle distribution monitoring value; wherein the plastic particle density monitoring value is a ratio of the number of plastic particle distributions to the plastic particle distribution area; The reaction monitoring value, the production monitoring value and the plastic particle distribution monitoring value are weightedly evaluated to obtain a first cross-linking degree monitoring value.

4. A method for online production monitoring of cross-linked polyethylene according to claim 1, characterized in that: The specific implementation process of evaluating the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature to calculate and obtain the second defect feature monitoring value includes: Inputting the image data into the cross-linked polyethylene feature recognition model for recognition processing; Acquire defect features output by the cross-linked polyethylene feature recognition model; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features; Respectively acquiring color information, bubble information, hole information, crack information and glue block information from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature; The color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value.

5. The method for online production monitoring of cross-linked polyethylene according to claim 1, characterized in that: The specific implementation process of inputting real-time monitoring data into the temperature prediction model and generating ambient temperature monitoring information based on the data results includes: Obtain historical monitoring data from the historical database; Acquire real-time monitoring data; wherein the real-time monitoring data includes: real-time crosslinking degree monitoring data and real-time defect characteristic monitoring data; Inputting the historical monitoring data into a temperature prediction model for training to obtain temperature prediction parameters; Input the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data into the temperature prediction model, update the model parameters, and output the real-time temperature parameters; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters; The real-time temperature monitoring information and the real-time temperature parameters are comprehensively evaluated to obtain the ambient temperature monitoring information.

6. A cross-linked polyethylene online production monitoring system, characterized in that: include: Equipment control module, data acquisition module, data monitoring module and output module; Wherein, the device control module is used to control the start, pause and stop of the device; The data acquisition module is used to obtain image data collected by the microscopic camera, reaction data and production data sent by the sensor, and historical monitoring data in the historical database; The data monitoring module is used to process the image data, the reaction data and the production data to obtain monitoring data; wherein the data monitoring module includes: a cross-linking degree monitoring unit and a defect feature monitoring unit; The output module is used to evaluate the monitoring data, output monitoring results and provide monitoring feedback; wherein the monitoring output module includes: a judgment unit and an ambient temperature monitoring unit.

7. A cross-linked polyethylene online production monitoring system according to claim 6, characterized in that: The data monitoring module processes the image data using the cross-linked polyethylene feature recognition model to obtain the plastic particle distribution data and defect feature data. The specific implementation process includes: Performing denoising processing on the image data to obtain denoised image data; Using histogram equalization to adjust the contrast and brightness of the denoised image data to obtain enhanced image data; Performing edge detection on the enhanced image data to obtain enhanced edge image data; The enhanced edge image data and the enhanced image data are input into a cross-linked polyethylene feature recognition model for recognition, thereby obtaining plastic particle distribution data and defect feature data.

8. A cross-linked polyethylene online production monitoring system according to claim 6, characterized in that: The specific implementation process of the cross-linking degree monitoring unit evaluating the reaction data, the production data and the plastic particle distribution data to obtain the first cross-linking degree monitoring value includes: Acquire reaction data of the sensor in the reaction device and production data of the sensor in the production device; wherein the reaction data includes: cross-linking agent concentration, reaction temperature and reaction time; the production data includes: production temperature and production pressure; Obtaining plastic particle distribution data outputted by the cross-linked polyethylene feature recognition model; wherein the plastic particle distribution data includes: the number of plastic particle distributions and the plastic particle distribution area; Comparing the cross-linking agent concentration, the reaction temperature and the reaction time with the cross-linking agent concentration standard value, the reaction temperature standard value and the reaction time standard value, respectively, to obtain a reaction monitoring value; Comparing the production temperature and the production pressure with a standard production temperature value and a standard production pressure value, respectively, to obtain a production monitoring value; Compare the plastic particle density monitoring value with the plastic particle density threshold to obtain a plastic particle distribution monitoring value; wherein the plastic particle density monitoring value is a ratio of the number of plastic particle distributions to the plastic particle distribution area; The reaction monitoring value, the production monitoring value and the plastic particle distribution monitoring value are weightedly evaluated to obtain a first cross-linking degree monitoring value.

9. A cross-linked polyethylene online production monitoring system according to claim 6, characterized in that: The defect feature monitoring unit evaluates the color feature, the bubble feature, the hole feature, the crack feature and the rubber block feature to obtain the second defect feature monitoring value, and the specific implementation process includes: Inputting the image data into the cross-linked polyethylene feature recognition model for recognition processing; Acquire defect features output by the cross-linked polyethylene feature recognition model; wherein the defect feature data includes: color features, bubble features, hole features, crack features and rubber block features; Respectively acquiring color information, bubble information, hole information, crack information and glue block information from the color feature, the bubble feature, the hole feature, the crack feature and the glue block feature; The color information, the bubble information, the hole information, the crack information and the rubber block information are weighted with the color characteristic coefficient, the bubble characteristic coefficient, the hole characteristic coefficient, the crack characteristic coefficient and the rubber block characteristic coefficient respectively to calculate a second defect characteristic monitoring value.

10. A cross-linked polyethylene online production monitoring system according to claim 6, characterized in that: The specific implementation process of the ambient temperature monitoring unit inputting real-time monitoring data into the temperature prediction model and generating ambient temperature monitoring information according to the data results includes: Obtain historical monitoring data from the historical database; Acquire real-time monitoring data; wherein the real-time monitoring data includes: real-time crosslinking degree monitoring data and real-time defect characteristic monitoring data; Inputting the historical monitoring data into a temperature prediction model for training to obtain temperature prediction parameters; Input the temperature prediction parameters, the real-time crosslinking degree monitoring data and the real-time defect characteristic monitoring data into the temperature prediction model, update the model parameters, and output the real-time temperature parameters; wherein the real-time temperature parameters include: real-time reaction temperature parameters and real-time production temperature parameters; The real-time temperature monitoring information and the real-time temperature parameters are comprehensively evaluated to obtain the ambient temperature monitoring information.

Images