Crosslinked cable vulcanization process control system and method

Abstract

The invention provides a crosslinked cable vulcanization process control system and method, and relates to the field of intelligent control, and the method comprises the steps: firstly obtaining a prior data set, vulcanization target parameters, and parameters and time in a real-time monitoring vulcanization process, so as to obtain real-time vulcanization parameters; a data analysis and coding technology based on deep learning is used for coding a priori data set and real-time vulcanization parameters and carrying out decision anchor point dynamic response, and meanwhile, vulcanization target parameters are coded, so that vulcanization optimization control parameters are obtained according to prediction-target offset representation; complex nonlinear relations and high-dimensional features existing in the vulcanization process of the cross-linked cable are processed more effectively, so that prediction precision and a control effect are improved. And meanwhile, the internal structure of the system can be dynamically adjusted to adapt to new input variables or changed data distribution, so that the system keeps efficient operation in different types of production environments.

Description

Technical Field

[0001] This application relates to the field of intelligent control, and more particularly, in the embodiments of this application, it relates to a cross-linked cable vulcanization process control system and method. Background Art

[0002] Cross-linked cables are key components in power transmission and distribution systems and are widely used in power networks, industrial facilities, and construction projects. The vulcanization process is a crucial step in ensuring that the cable insulation and sheath materials have excellent mechanical and electrical properties. During the vulcanization process, precise control of parameters such as temperature, pressure, and time directly affects the quality of the final product.

[0003] The existing patent CN118519404A proposes an artificial intelligence-based cross-linked cable vulcanization control method. First, it collects data such as temperature, pressure, and time during the vulcanization process and cable performance indicators in real time through sensors. Then, it uses a density-based algorithm to process outliers and applies the Yeo-Johnson transformation for data normalization. Next, it uses the LightGBM algorithm to establish a prediction model and optimizes the hyperparameters through SMAC to improve accuracy. Finally, it deploys the optimized model to the production control system and uses edge computing technology to achieve real-time monitoring and automatic adjustment of the vulcanization process.

[0004] However, the patent solution mainly relies on a specific machine learning model (such as LightGBM) for prediction and optimization, and this single model may not be able to fully adapt to all types of production environments and changes. Moreover, there may be complex non-linear relationships or high-dimensional features among the key parameters in the actual vulcanization process of cross-linked cables. Although LightGBM is good at processing large-scale data sets, it may not be effective enough in dealing with such complex features. In addition, the optimization mentioned in the patent solution is mainly to optimize the hyperparameters of the LightGBM model through the SMAC algorithm, which is usually an offline process. In the actual production environment, production conditions may change frequently, and this offline optimization-based solution is difficult to meet the requirements of immediacy and dynamicity.

[0005] Therefore, an optimized cross-linked cable vulcanization process control scheme is desired. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a cross-linked cable vulcanization process control system and method. First, a prior dataset, vulcanization target parameters, and parameters and time during the vulcanization process are monitored in real time to obtain real-time vulcanization parameters. Deep learning-based data analysis and coding techniques are used to encode the prior dataset and real-time vulcanization parameters and perform decision anchor dynamic response, and at the same time, the vulcanization target parameters are encoded. Thus, vulcanization optimization control parameters are obtained according to the prediction-target offset representation to more effectively handle the complex non-linear relationships and high-dimensional features existing in the cross-linked cable vulcanization process, thereby improving the prediction accuracy and control effect. At the same time, its internal structure can be dynamically adjusted to adapt to new input variables or changing data distributions, enabling the system to operate efficiently in different types of production environments.

[0007] According to one aspect of the present application, a cross-linked cable vulcanization process control method is provided, which includes: obtaining a prior dataset, where each data sample in the prior dataset is {vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of the cable after vulcanization, elongation at break of the cable after vulcanization, electrical performance of the cable after vulcanization}; obtaining vulcanization target parameters input by the user, where the vulcanization target parameters include the target tensile strength of the cable after vulcanization, the target elongation at break of the cable after vulcanization, and the target electrical performance of the cable after vulcanization; monitoring the vulcanization temperature, vulcanization pressure, and vulcanization time during the vulcanization process in real time to obtain real-time vulcanization parameters; encoding the prior dataset and the real-time vulcanization parameters respectively to obtain a sequence of vulcanization prior parameter encoding vectors and vulcanization real-time parameter encoding vectors; performing decision anchor vulcanization effect dynamic response encoding on the vulcanization real-time parameter encoding vector and the sequence of vulcanization prior parameter encoding vectors to obtain a vulcanization effect dynamic response encoding vector; performing fully connected encoding on the vulcanization target parameters to obtain a vulcanization target parameter encoding vector; calculating a vulcanization effect prediction-target offset encoding vector between the vulcanization effect dynamic response encoding vector and the vulcanization target parameter encoding vector; and obtaining vulcanization optimization control parameters according to the vulcanization effect prediction-target offset encoding vector.

[0008] According to another aspect of the present application, a cross-linked cable vulcanization process control system is provided, which includes: a prior data set acquisition module for acquiring a prior data set, where each data sample in the prior data set is {vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of the cable after vulcanization, elongation at break of the cable after vulcanization, electrical performance of the cable after vulcanization}; a vulcanization target parameter acquisition module for acquiring vulcanization target parameters input by a user, where the vulcanization target parameters include the target tensile strength of the cable after vulcanization, the target elongation at break of the cable after vulcanization, and the target electrical performance of the cable after vulcanization; a real-time vulcanization parameter monitoring module for monitoring the vulcanization temperature, vulcanization pressure, and vulcanization time during the vulcanization process in real time to obtain real-time vulcanization parameters; a vulcanization parameter encoding module for encoding the prior data set and the real-time vulcanization parameters respectively to obtain a sequence of vulcanization prior parameter encoding vectors and vulcanization real-time parameter encoding vectors; a decision anchor vulcanization effect dynamic response encoding module for performing decision anchor vulcanization effect dynamic response encoding on the sequence of the vulcanization real-time parameter encoding vectors and the vulcanization prior parameter encoding vectors to obtain a vulcanization effect dynamic response encoding vector; a vulcanization target parameter fully connected module for performing fully connected encoding on the vulcanization target parameters to obtain a vulcanization target parameter encoding vector; a vulcanization effect prediction - target offset encoding module for calculating a vulcanization effect prediction - target offset encoding vector between the vulcanization effect dynamic response encoding vector and the vulcanization target parameter encoding vector; a vulcanization optimization control parameter module for obtaining vulcanization optimization control parameters according to the vulcanization effect prediction - target offset encoding vector.

[0009] Compared with the prior art, the cross-linked cable vulcanization process control system and method provided by the present application first acquire a prior data set, vulcanization target parameters, and monitor the parameters and time during the vulcanization process in real time to obtain real-time vulcanization parameters, use deep learning-based data analysis and encoding techniques to encode the prior data set and real-time vulcanization parameters and perform decision anchor dynamic response, and at the same time encode the vulcanization target parameters, so as to obtain vulcanization optimization control parameters according to the prediction - target offset representation, to more effectively handle the complex non-linear relationships and high-dimensional features existing in the cross-linked cable vulcanization process, thereby improving the prediction accuracy and control effect. At the same time, it can dynamically adjust its internal structure to adapt to new input variables or changing data distributions, enabling the system to operate efficiently in different types of production environments. Brief Description of the Drawings

[0010] The above and other objects, features, and advantages of the present application will become more apparent by describing the embodiments of the present application in more detail with reference to the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 It is a flowchart of a cross-linked cable vulcanization process control method according to an embodiment of the present application.

[0012] Figure 2 It is a schematic diagram of data flow of a cross-linked cable vulcanization process control method according to an embodiment of the present application.

[0013] Figure 3 It is a flowchart of performing decision anchor vulcanization effect dynamic response encoding on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector in a cross-linked cable vulcanization process control method according to an embodiment of the present application to obtain a vulcanization effect dynamic response encoding vector.

[0014] Figure 4 It is a flowchart of calculating decision anchor adaptive splicing weight factors for each vulcanization prior-real-time parameter local feature response anchor encoding matrix in the sequence of the vulcanization prior-real-time parameter local feature response anchor encoding matrix in a cross-linked cable vulcanization process control method according to an embodiment of the present application to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors.

[0015] Figure 5 It is a system block diagram of a cross-linked cable vulcanization process control system according to an embodiment of the present application. Detailed Embodiments

[0016] The following will describe various exemplary embodiments, features, and aspects of the present application in detail with reference to the accompanying drawings. The same reference numerals in the drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the drawings, the drawings do not have to be drawn to scale unless otherwise specified.

[0017] The special term "exemplary" here means "serving as an example, embodiment, or illustration". Any embodiment described as "exemplary" here does not have to be construed as superior to or better than other embodiments.

[0018] In addition, in order to better illustrate the present application, numerous specific details are given in the following detailed embodiments. Those skilled in the art should understand that the present application can be implemented without some specific details. In some instances, methods, means, elements, and circuits well-known to those skilled in the art are not described in detail so as to highlight the gist of the present application.

[0019] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, "a plurality of" means two or more, unless otherwise specifically defined.

[0020] In view of the above technical problems, the present application proposes a method for controlling the vulcanization process of cross-linked cables. Figure 1 It is a flowchart of the method for controlling the vulcanization process of cross-linked cables according to an embodiment of the present application. Figure 2 It is a schematic diagram of data flow of the method for controlling the vulcanization process of cross-linked cables according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the method for controlling the vulcanization process of cross-linked cables according to an embodiment of the present application includes: S110, obtaining a prior data set, where each data sample in the prior data set is {vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of the cable after vulcanization, elongation at break of the cable after vulcanization, electrical performance of the cable after vulcanization}; S120, obtaining vulcanization target parameters input by a user, where the vulcanization target parameters include the target tensile strength of the cable after vulcanization, the target elongation at break of the cable after vulcanization, and the target electrical performance of the cable after vulcanization}; S130, monitoring the vulcanization temperature, vulcanization pressure, and vulcanization time during the vulcanization process in real time to obtain real-time vulcanization parameters; S140, encoding the prior data set and the real-time vulcanization parameters respectively to obtain a sequence of vulcanization prior parameter encoding vectors and a vulcanization real-time parameter encoding vector; S150, performing decision anchor vulcanization effect dynamic response encoding on the vulcanization real-time parameter encoding vector and the sequence of vulcanization prior parameter encoding vectors to obtain a vulcanization effect dynamic response encoding vector; S160, performing fully connected encoding on the vulcanization target parameters to obtain a vulcanization target parameter encoding vector; S170, calculating a vulcanization effect prediction-target offset encoding vector between the vulcanization effect dynamic response encoding vector and the vulcanization target parameter encoding vector; S180, obtaining vulcanization optimization control parameters according to the vulcanization effect prediction-target offset encoding vector.

[0021] Accordingly, the technical concept of this application is to obtain a prior dataset, input the vulcanization target parameters by the user, and monitor the parameters and time during the vulcanization process in real time to obtain real-time vulcanization parameters. Then, use data analysis and coding techniques based on deep learning to encode the prior dataset and real-time vulcanization parameters. Next, perform decision anchor dynamic response on the encoded vulcanization real-time parameter encoding features and each vulcanization prior parameter encoding feature, and at the same time encode the vulcanization target parameters. Based on this, the prediction-target offset representation between the vulcanization effect dynamic response encoding feature and the vulcanization target parameter encoding feature after dynamic response is used to intelligently obtain the vulcanization optimization control parameters. This application can more effectively handle the complex non-linear relationships and high-dimensional features existing in the cross-linked cable vulcanization process, thereby improving the prediction accuracy and control effect. At the same time, it can dynamically adjust its internal structure to adapt to new input variables or changing data distributions. This flexibility enables the system to maintain efficient operation in different types of production environments.

[0022] In the above cross-linked cable vulcanization process control method, in step S110, a prior dataset is obtained. Each data sample in the prior dataset is {vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of the cable after vulcanization, elongation at break of the cable after vulcanization, electrical properties of the cable after vulcanization}. It should be understood that the vulcanization temperature is usually monitored in real time by high-precision temperature sensors such as thermocouples or infrared thermometers. These sensors can record the temperature changes during the vulcanization process with a time resolution of milliseconds, thus ensuring the accuracy and continuity of the temperature data. Similarly, the measurement of vulcanization pressure also requires the use of special pressure sensors, which can accurately capture the pressure fluctuations inside the mold during the vulcanization process. For the recording of the vulcanization time, it can be achieved by linking a timer with the control system to ensure that every second from the start to the end of vulcanization is accurately recorded. In this way, a complete database of vulcanization process parameters can be constructed to provide basic support for subsequent analysis. However, simply recording the physical conditions during the vulcanization process is not sufficient to fully reflect the quality characteristics of the product. Therefore, after vulcanization, a series of strict performance tests need to be carried out on the cable samples. Among them, in order to obtain the tensile strength and elongation at break, a universal material testing machine is usually used to apply a gradually increasing tensile force to the cable sample until the sample breaks. During this process, the testing machine will automatically record the maximum tensile force value (i.e., the tensile strength) and the maximum deformation amount that the sample can withstand before breaking (i.e., the elongation at break). In addition, parameters such as insulation resistance, breakdown voltage, and dielectric loss tangent value need to be measured by special electrical testing instruments. It is worth mentioning that in the selection of data samples, it is necessary to ensure their representativeness and diversity. This means not only covering different types of cable products, but also considering various factors that may affect the vulcanization effect, such as raw material formula, production process parameters, and environmental conditions. For example, for different rubber substrates (such as natural rubber, styrene-butadiene rubber, ethylene-propylene-diene monomer rubber, etc.), the optimal vulcanization temperature range may be different. Therefore, when constructing the prior dataset, data samples from different types of rubber substrates should be collected as much as possible so that the model can better understand and adapt to this difference. At the same time, attention should also be paid to avoiding bias problems caused by uneven sample distribution. If the number of data samples under certain specific conditions is too small, it may lead to large errors when the model predicts the vulcanization effect under these conditions. To this end, it is necessary to improve the uniformity of the sample distribution by increasing the number of experiments or introducing data augmentation techniques.

[0023] In the above cross-linked cable vulcanization process control method, in step S120, the vulcanization target parameters input by the user are obtained. The vulcanization target parameters include the target tensile strength of the cable after vulcanization, the target elongation at break of the cable after vulcanization, and the target electrical performance of the cable after vulcanization. It should be understood that during the cross-linked cable vulcanization process, the vulcanization target parameters input by the user include the target tensile strength, the target elongation at break, and the target electrical performance of the cable after vulcanization, which jointly determine the mechanical and electrical performance of the cable. Considering that there are significant differences in the performance requirements of cables for different application scenarios. For example, in high-voltage transmission lines, the cable needs to have extremely high insulation performance and withstand voltage ability; while in building wiring, flexibility, anti-aging ability, and construction convenience are more emphasized. Therefore, allowing the user to directly participate in setting the target parameters can make the produced cable as close as possible to the actual usage requirements. Specifically, first, in order to clarify the specific expectations of the user, the system usually provides an interactive interface that allows the user to input specific values or ranges according to their own needs. However, relying solely on the original data provided by the user is not enough because these data are often subjective and uncertain. Therefore, a series of processing and verification are required to ensure their rationality and feasibility. When the user inputs the target parameters, the system will automatically check whether these parameters meet the relevant industry standards and technical specifications. For example, for mechanical performance indicators such as tensile strength and elongation at break, the system can evaluate whether the target values set by the user are reasonable by referring to the standards. If it is found that the target values deviate from the normal range, the system will prompt the user to readjust or provide further explanations. In addition, since the electrical performance involves multiple dimensions (such as insulation resistance, dielectric loss factor, breakdown voltage, etc.), the system also needs to comprehensively consider the mutual relationship between these parameters to avoid contradictory or unrealistic requirements. For example, a high breakdown voltage usually means a higher insulation thickness, which may have a negative impact on the flexibility of the cable. Therefore, when setting the target parameters, it is necessary to weigh the trade-offs between different performance indicators to find the optimal balance point. In the actual implementation process, considering the uncertainties and changing conditions in industrial production, the vulcanization target parameters input by the user also need to have a certain degree of flexibility. This means that not only the target values in the ideal state need to be set, but also the possible deviation ranges need to be considered. For example, the target tensile strength can be set as "30±2MPa", which means that as long as the actual value falls between 28MPa and 32MPa, it can be regarded as qualified. This flexibility not only improves the robustness of the system but also provides a greater space for subsequent optimization and adjustment.

[0024] In the above cross-linked cable vulcanization process control method, in step S130, the vulcanization temperature, vulcanization pressure, and vulcanization time during the vulcanization process are monitored in real time to obtain real-time vulcanization parameters. It should be understood that considering that these three key parameters of vulcanization temperature, vulcanization pressure, and vulcanization time directly affect the molecular structure change of the cable material and the mechanical and electrical properties of the final product. For example, too high vulcanization temperature may cause material aging or degradation, while too low temperature may not fully achieve the cross-linking reaction; insufficient vulcanization pressure may affect the compactness of the internal structure of the cable, thus reducing its voltage withstand capacity; too short vulcanization time may result in insufficient cross-linking degree, while too long time will waste energy and increase production costs. Therefore, by monitoring these parameters in real time, deviations can be detected and corrected in a timely manner, thus ensuring that the vulcanization process is always in an optimal state. Specifically, in order to achieve real-time monitoring of vulcanization temperature, vulcanization pressure, and vulcanization time, the system usually needs to rely on a series of advanced sensor technologies and data acquisition devices. First, for the monitoring of vulcanization temperature, high-precision temperature sensors such as thermocouples and infrared thermometers are widely used in modern industry. These sensors can capture the temperature changes in the vulcanization chamber with a time resolution of milliseconds and transmit the data to the control system in real time for analysis. For example, in actual production, if it is detected that the vulcanization temperature deviates from the set value, the system can quickly restore the target temperature range by adjusting the power of the heating device or changing the flow rate of the cooling medium. In addition, in order to improve the accuracy of temperature control, a PID (Proportional-Integral-Derivative) controller can also be introduced to further optimize the temperature regulation strategy through learning historical data and constructing a prediction model. For the monitoring of vulcanization pressure, it mainly relies on pressure sensors and pressure transmitter devices. These devices can accurately measure the pressure changes in the vulcanization chamber and convert them into electrical signals for the control system to process. In practical applications, the stability of vulcanization pressure is crucial for the quality of cable products. For example, when the pressure is too low, it may cause bubbles or voids on the surface of the cable, thus affecting its insulation performance; while too high pressure may cause die deformation or damage. Therefore, the system needs to dynamically adjust the working state of the compressed air supply or other pressurizing devices according to the target vulcanization pressure value input by the user and the real-time monitoring data to maintain the constant pressure in the vulcanization chamber. At the same time, in order to cope with emergencies, the system should also have an alarm function, which can immediately notify the operator to take measures when abnormal pressure is detected to avoid accidents. As for the monitoring of vulcanization time, this usually involves the coordinated work of a timer and a logic controller. The setting of vulcanization time needs to comprehensively consider factors such as the characteristics of the cable material, product specifications, and user requirements. For example, for thicker cables, a longer vulcanization time may be required to ensure that the cross-linking reaction is fully completed; while for thinner cables, the vulcanization time can be appropriately shortened to improve production efficiency. During the actual production process, the system will start timing according to the preset time parameters and automatically stop the vulcanization process after reaching the predetermined time.However, relying solely on fixed time settings is not sufficient to ensure product quality because there may be various uncertain factors in actual production, such as temperature fluctuations or pressure changes. Therefore, modern vulcanization control systems usually adopt a closed-loop feedback mechanism. By monitoring the changes in vulcanization temperature and vulcanization pressure in real time, the vulcanization time is dynamically adjusted to ensure that the performance of the final product meets the expected goals.

[0025] In the above cross-linked cable vulcanization process control method, in step S140, the prior data set and the real-time vulcanization parameters are respectively encoded to obtain a sequence of vulcanization prior parameter encoding vectors and vulcanization real-time parameter encoding vectors. It should be understood that considering the significant differences in the nature and uses between the prior data set and the real-time vulcanization parameters. Among them, the prior data set usually comes from historical production records, experimental studies or theoretical calculations, which reflects various situations that may be encountered in the vulcanization process and their corresponding performance manifestations. For example, through statistical analysis of vulcanization results under different temperature, pressure and time combinations, a set of empirical models regarding tensile strength, elongation at break and electrical properties can be obtained. These models not only help predict the product performance under new process conditions, but also can be used as a benchmark to evaluate whether the current production state deviates from the normal range. In contrast, the real-time vulcanization parameters are the data directly collected during the current production process, which represent the actual physical and chemical changes. Since real-time data is often affected by various uncertain factors (such as equipment aging, environmental fluctuations, etc.), its quality may not be as stable and reliable as the prior data. Therefore, by encoding the two separately, the deficiencies of each other can be made up to a certain extent, forming a more comprehensive and accurate characterization system. It should be noted that whether encoding the prior data set or the real-time vulcanization parameters, it is necessary to ensure that the encoding results have good interpretability and generalization ability. This means that the encoded vectors should not only be able to accurately reflect the main characteristics of the original data, but also should remain valid in new scenarios that have not been seen before. To this end, means such as cross-validation and regularization can be used to prevent the occurrence of overfitting phenomena.

[0026] In the embodiment of the present application, step S140 of respectively encoding the prior data set and the real-time vulcanization parameters to obtain a sequence of vulcanization prior parameter encoding vectors and a vulcanization real-time parameter encoding vector includes: S141, performing fully connected encoding based on a multi-layer perceptron model on each data sample in the prior data set to obtain the sequence of vulcanization prior parameter encoding vectors; S142, performing the fully connected encoding based on the multi-layer perceptron model on the real-time vulcanization parameters to obtain the vulcanization real-time parameter encoding vector.

[0027] Specifically, in the step S141, all data samples in the prior dataset are subjected to fully-connected encoding based on a multi-layer perceptron model to obtain a sequence of vulcanization prior parameter encoding vectors. It should be understood that considering that each data sample in the prior dataset contains multiple different types of parameters, such as vulcanization temperature, pressure, time, and various performance indicators of the cable after vulcanization. The dimensions, value ranges, and physical meanings of these parameters are different, and there are complex non-linear relationships between the parameters. For example, the combination of vulcanization temperature, pressure, and time will have a comprehensive impact on the tensile strength, elongation at break, and electrical properties of the cable, and this impact is not a simple linear superposition. Therefore, in order to capture the correlations between the parameters, the present application performs fully-connected encoding on all data samples in the prior dataset based on a multi-layer perceptron model to obtain a sequence of vulcanization prior parameter encoding vectors. Among them, the multi-layer perceptron model (MLP) has strong feature extraction capabilities. Through fully-connected encoding, these complex original data can be converted into low-dimensional, compact, and representative encoding vectors. For example, the MLP can automatically learn the potential non-linear relationship between vulcanization temperature and cable tensile strength and incorporate this relationship into the encoding vector. More specifically, the MLP consists of multiple neuron layers. Each layer of neurons is connected to the adjacent layer in a fully-connected manner and can use non-linear activation functions, which can effectively capture complex patterns and features in the data. By performing fully-connected encoding on the prior dataset, the MLP can learn these hidden patterns and provide more valuable information for subsequent decision-making.

[0028] Specifically, in the step S142, the real-time vulcanization parameters are subjected to the fully-connected encoding based on the multi-layer perceptron model to obtain the vulcanization real-time parameter encoding vector. It should be understood that the multi-layer perceptron model is a classic feed-forward neural network structure, consisting of an input layer, a hidden layer, and an output layer, where the hidden layer can contain one or more neuron layers. Each layer of neurons is connected to the adjacent layer in a fully-connected manner and uses non-linear activation functions (such as ReLU, Sigmoid, or Tanh) to introduce non-linear mapping capabilities. Considering that there are also non-linear correlations between the real-time vulcanization parameters, therefore, in the technical solution of the present application, the real-time vulcanization parameters are subjected to fully-connected encoding based on the multi-layer perceptron model to utilize the strong non-linear mapping capabilities of the multi-layer perceptron model. Through the fully-connected structure and activation functions, these complex features are automatically extracted from the real-time vulcanization parameters. For example, the combined effect of vulcanization temperature and vulcanization time on the cable performance may be non-linear, and the multi-layer perceptron model can capture this hidden feature information and reflect it in the encoding vector, thereby obtaining the vulcanization real-time parameter encoding vector.

[0029] In the above cross-linked cable vulcanization process control method, in step S150, a decision anchor vulcanization effect dynamic response encoding is performed on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a vulcanization effect dynamic response encoding vector. It should be understood that considering that the vulcanization real-time parameter encoding vector can be analogized to local features, which reflect the immediate and detailed information of the current vulcanization process. These data have high resolution and dynamics, but may be affected by noise interference or environmental fluctuations, so its stability is relatively weak; while the sequence of vulcanization prior parameter encoding vectors is similar to global features, covering the overall experience and rules of the historical vulcanization process. However, since they are generated based on historical data, they may not fully adapt to the subtle changes in the current specific scenario. Although these two types of features have significant differences in nature, they are highly complementary at different granularities, and there is a complex correlation relationship between the two. Direct splicing or simple fusion may lead to feature conflicts or information loss because there is a complex non-linear relationship between local features and global features, and simple combination methods are difficult to effectively capture this deep interaction pattern. Therefore, in the technical solution of this application, a decision anchor vulcanization effect dynamic response encoding mechanism is introduced to perform response analysis on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a vulcanization effect dynamic response encoding vector. In particular, this mechanism uses the hidden encoding of the vulcanization prior parameter encoding vector sequence as an anchor point to perform dynamic interaction modeling on the vulcanization real-time parameter encoding vector, which can effectively avoid the information distortion problem caused by simple fusion. The generated vulcanization effect dynamic response encoding vector can effectively capture the relationship between the global perspective (prior experience) and the detailed characteristics (real-time state) in the vulcanization process. This can comprehensively and accurately reflect the actual situation of the vulcanization process and provide richer and more accurate information for subsequent decision-making.

[0030] Figure 3 It is a flowchart of performing decision anchor vulcanization effect dynamic response encoding on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a vulcanization effect dynamic response encoding vector in the cross-linked cable vulcanization process control method according to an embodiment of the present application. As Figure 3As shown, in the embodiment of the present application, step S150 of performing decision anchor vulcanization effect dynamic response encoding on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a vulcanization effect dynamic response encoding vector includes: S151, performing deep implicit feature extraction on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a sequence of a vulcanization real-time parameter deep implicit encoding vector and a vulcanization prior parameter deep implicit encoding vector; S152, respectively inputting each vulcanization prior parameter deep implicit encoding vector in the sequence of the vulcanization real-time parameter deep implicit encoding vector and the vulcanization prior parameter deep implicit encoding vector into a semantic response decision anchoring component to obtain a sequence of vulcanization prior-real-time parameter local feature response anchoring encoding matrices; S153, calculating a decision anchor adaptive splicing weight factor for each vulcanization prior-real-time parameter local feature response anchoring encoding matrix in the sequence of the vulcanization prior-real-time parameter local feature response anchoring encoding matrices to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors; S154, fusing the sequence of the vulcanization prior-real-time parameter local feature response anchoring encoding matrices based on the sequence of the vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors to obtain the vulcanization effect dynamic response encoding vector.

[0031] In the embodiment of the present application, step S151 of performing deep implicit feature extraction on the sequences of the vulcanization real-time parameter encoding vector and the vulcanization prior parameter encoding vector to obtain a sequence of a vulcanization real-time parameter deep implicit encoding vector and a vulcanization prior parameter deep implicit encoding vector includes: S1511, performing deep implicit feature extraction based on fully connected encoding on the vulcanization real-time parameter encoding vector to obtain the vulcanization real-time parameter deep implicit encoding vector; S1512, performing deep implicit feature extraction based on fully connected encoding on each vulcanization prior parameter encoding vector in the sequence of the vulcanization prior parameter encoding vector to obtain a sequence of the vulcanization prior parameter deep implicit encoding vectors. Among them, step S1511 is represented by the vulcanization real-time parameter fully connected encoding formula as: , where is the vulcanization real-time parameter encoding vector, is matrix multiplication, and are respectively the vulcanization real-time parameter weight matrix and the vulcanization real-time parameter bias vector, is the activation function, It is the deep implicit encoding vector of vulcanization real-time parameters. It should be understood that through the mapping of the fully connected neural network, the vulcanization real-time parameter encoding vector is projected into a higher-dimensional latent space, where the features have a richer semantic representation ability. This process not only learns the global representation of the features through the network weights, but also captures the non-linear interaction relationships in the original features through the non-linear activation function, laying a foundation for the subsequent semantic alignment with local features. Specifically, the technical solution of this application first takes the original vulcanization real-time parameter encoding vector as input and passes it to the input layer of the fully connected neural network. Subsequently, each layer of neurons in the network combines the bias vector through matrix multiplication operations (i.e., linear transformations) and introduces non-linearity through the activation function, gradually extracting the complex patterns hidden in the data. Finally, the output after multiple layers of processing is the deep implicit encoding vector of vulcanization real-time parameters. This encoding vector not only retains the main characteristics of the original data but also enhances its adaptability to complex scenarios. It should be noted that the mapping process of the fully connected neural network is essentially a transformation from a low-dimensional space to a high-dimensional space. In this transformation process, the network weights play a key role, and they are automatically adjusted through the learning process in the training stage to capture the global representation of the input data. At the same time, the activation function (such as ReLU, Sigmoid, or Tanh) is responsible for introducing non-linearity, enabling the model to effectively handle the possible non-linear interaction relationships in the data.

[0032] Among them, the step S1512 is represented by the fully connected encoding of vulcanization prior parameters as: ; where is the sequence of the vulcanization prior parameter encoding vectors, and are respectively the 1st, 2nd, th, and th vulcanization prior parameter encoding vectors in the sequence of the vulcanization prior parameter encoding vectors, and are respectively the vulcanization prior parameter weight matrix and the vulcanization prior parameter bias vector, is the A sequence of vulcanization prior parameter deep implicit encoding vectors. It should be understood that the sequence of vulcanization prior parameter encoding vectors usually comes from historical production records or experimental studies. Although these data are stable and reliable, due to their high dimensionality and numerous variables, direct use may lead to an increase in computational complexity or even overfitting problems. Therefore, in the technical solution of this application, the sequence of the vulcanization prior parameter encoding vectors is independently processed, and a fully connected neural network is used to perform deep implicit feature extraction on each feature vector to reduce data redundancy and enhance interpretability and generalization ability. Among them, the combination of vulcanization temperature, vulcanization pressure, and vulcanization time may have a highly non-linear impact on the cable performance, and the fully connected neural network can automatically learn these potential relationships and incorporate them into the final encoding vector to obtain a sequence of vulcanization prior parameter deep implicit encoding vectors. Here, the input sequence of vulcanization prior parameter encoding vectors is different from the single representation of the vulcanization real-time parameter encoding vector. The sequence of vulcanization prior parameter encoding vectors emphasizes the global perspective, reflects the overall law of historical experience, and contains multiple features that need to be encoded one by one, while the vulcanization real-time parameter encoding vector focuses on local details and captures the immediate changes in the current production process. Therefore, since the vulcanization real-time parameter encoding vector is a single representation, feature extraction can be directly completed through a single complete mapping of the fully connected network; while the sequence of vulcanization prior parameter encoding vectors needs to be encoded for each sample to comprehensively reflect all the information in the sequence.

[0033] Specifically, in the step S152, each vulcanization prior parameter deep implicit encoding vector in the sequence of the vulcanization real-time parameter deep implicit encoding vector and the sequence of the vulcanization prior parameter deep implicit encoding vectors is input into the semantic response decision anchoring component to obtain a sequence of vulcanization prior-real-time parameter local feature response anchoring encoding matrices, which is expressed by the semantic response decision anchoring formula as: ; where is the transposed vector of is the length of is the A set of vulcanization prior-real-time parameter local feature response anchored encoding matrices. It should be understood that the core idea of this step is to generate a set of vulcanization prior-real-time parameter local feature response anchored encoding matrices with global reference characteristics between the sequences of vulcanization real-time parameter deep implicit encoding vectors and the vulcanization prior parameter deep implicit encoding vectors. The sequence of the vulcanization prior-real-time parameter local feature response anchored encoding matrices not only represents the interaction information between the first feature and the second feature, but also serves as a bridge to capture the global-local semantic alignment relationship. In the technical solution of this application, the sequence of the vulcanization prior-real-time parameter local feature response anchored encoding matrices is equivalent to the building block components that make up the final decision anchor, and it is suitable for being combined and spliced in a specific way to form a deep implicit representation of the complete decision anchor. Specifically, in the technical solution of this application, the sequences of vulcanization real-time parameter deep implicit encoding vectors and vulcanization prior parameter deep implicit encoding vectors respectively represent the immediate state of the current vulcanization process and the deep features of historical experience data. Although these two types of features come from different sources, there are complex correlation relationships between them. Through the action of the semantic response decision anchoring component, each element in the sequence of vulcanization real-time parameter deep implicit encoding vectors and the vulcanization prior parameter deep implicit encoding vectors is paired one by one, and based on the matching results of this similarity measure, a matrix form representation is constructed, that is, the corresponding vulcanization prior-real-time parameter local feature response anchored encoding matrix is generated, where each row or each column corresponds to the feature interaction information in a specific dimension. This matrix is essentially a quantitative representation of the interaction relationship between the two sets of features. It not only reflects the specific association between the vulcanization real-time parameter and the vulcanization prior parameter at a specific moment, but also can reveal the dynamic change law between the two.

[0034] Figure 4 A flowchart for calculating the decision anchor adaptive splicing weight factors of each vulcanization prior-real-time parameter local feature response anchored encoding matrix in the sequence of the vulcanization prior-real-time parameter local feature response anchored encoding matrices in the cross-linked cable vulcanization process control method according to an embodiment of the present application to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors. As Figure 4As shown, in the embodiment of the present application, in step S153, calculating the decision anchor adaptive splicing weight factor of each vulcanization prior-real-time parameter local feature response anchor encoding matrix in the sequence of vulcanization prior-real-time parameter local feature response anchor encoding matrices to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors includes: S1531, based on the feature distribution of each vulcanization prior-real-time parameter local feature response anchor encoding matrix in the sequence of vulcanization prior-real-time parameter local feature response anchor encoding matrices, determining the decision anchor adaptive splicing factor of each vulcanization prior-real-time parameter local feature response anchor encoding matrix to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing factors; S1532, performing a weighting process based on the Softmax function on the sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing factors to obtain the sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors.

[0035] In the embodiment of the present application, in step S1531, based on the feature distribution of each vulcanization prior-real-time parameter local feature response anchor encoding matrix in the sequence of vulcanization prior-real-time parameter local feature response anchor encoding matrices, determining the decision anchor adaptive splicing factor of each vulcanization prior-real-time parameter local feature response anchor encoding matrix to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing factors includes: obtaining the decision anchor adaptive splicing factor of the vulcanization prior-real-time parameter local feature response anchor encoding matrix based on the maximum value, mean value, variance, difference amplification coefficient, and drift coefficient of the vulcanization prior-real-time parameter local feature response anchor encoding matrix. Among them, step S1531 is represented by the vulcanization prior-real-time parameter decision anchor formula as: ; where represents the variance of represents the mean value of is for calculating the number of eigenvalues of represents the number of eigenvalues of represents the maximum value of ， represents the drift coefficient, is the A vulcanization prior-real-time parameter decision anchor adaptive splicing factor. It should be understood that the sequence of vulcanization prior-real-time parameter local feature response anchoring coding matrices is generated by pairing the vulcanization real-time parameter deep implicit coding vectors and the vulcanization prior parameter deep implicit coding vectors one by one, which not only reflects the interaction information between the two types of features but also can capture the global and local semantic alignment relationships. However, just generating the sequence of vulcanization prior-real-time parameter local feature response anchoring coding matrices is not sufficient to complete the task of the entire vulcanization effect dynamic response coding because their respective importance may vary depending on the specific scenario. Therefore, after generating the vulcanization prior-real-time parameter local feature response anchoring coding matrices, the module needs to further extract and model the feature contributions of each vulcanization prior-real-time parameter local feature response anchoring coding matrix, which is the role of the vulcanization prior-real-time parameter decision anchor adaptive splicing factor. By analyzing the feature distributions of the respective vulcanization prior-real-time parameter local feature response anchoring coding matrices, the vulcanization prior-real-time parameter decision anchor adaptive splicing factors of the respective vulcanization prior-real-time parameter local feature response anchoring coding matrices are determined. These vulcanization prior-real-time parameter decision anchor adaptive splicing factors measure the dominance of local features in a specific task or context and serve as the weight basis for subsequent feature fusion. Specifically, in the technical solution of the present application, by analyzing statistics such as the maximum value, mean value, and variance of each vulcanization prior-real-time parameter local feature response anchoring coding matrix and combining the difference amplification coefficient and the drift coefficient, the feature contribution of each matrix is quantified. Among them, the maximum value represents the strongest feature response intensity in the matrix, indicating that the feature interaction in a specific dimension is the most significant; the mean value reflects the average level of the overall feature response and reveals the overall activity degree of the matrix; while the variance is used to measure the fluctuation range of the feature response to help identify those regions with high uncertainty. In addition, the difference amplification coefficient further enhances the contrast of local features relative to global features, making some key features more prominent; the drift coefficient ensures that the feature distribution gradually changes from a weakly interpretable state to a strongly interpretable state, thereby enhancing the global dominance basis of local features. In this way, a decision anchor adaptive splicing factor can be assigned to each vulcanization prior-real-time parameter local feature response anchoring coding matrix. This factor is essentially a scalar value used to represent the relative importance of the matrix in the entire vulcanization process.

[0036] Among them, the drift coefficient is obtained by performing local-global smooth transition processing on the vulcanization prior-real-time parameter local feature response anchoring coding matrix and is expressed by the drift coefficient calculation formula as: ; where is the intermediate state transition, is the eigenvalues, representing the drift coefficient.

[0037] It should be understood that for the drift coefficient in the vulcanization prior-real-time parameter decision anchor adaptive splicing factor , for the state transition of the input feature set distribution of the vulcanization prior-real-time parameter local feature response anchor encoding matrix from weak overall interpretability of the mean to strong local interpretability of the maximum, through the drift coefficient from weak to strong interpretable generalization to enhance the global dominance basis of local features.

[0038] Among them, is used as the intermediate state transition representation from weak interpretability to strong interpretability. For each eigenvalue of the vulcanization prior-real-time parameter local feature response anchor encoding matrix, it is used as the importance score of the input vulcanization prior-real-time parameter local feature response anchor encoding matrix for the global smooth state transition, to perform importance score weight global control on the intermediate state transition relative to the global state transition, so as to achieve the interpretable generalization inference of the weight basis of the vulcanization prior-real-time parameter decision anchor adaptive splicing factor.

[0039] Specifically, in the step S1532, weight processing based on the Softmax function is performed on the sequence of the vulcanization prior-real-time parameter decision anchor adaptive splicing factor to obtain the sequence of the vulcanization prior-real-time parameter decision anchor adaptive splicing weight factor, which is expressed by the weight processing formula as: ; among them, is the normalization function, is the A sulfurization prior-real-time parameter decision anchor adaptive splicing weight factor. It should be understood that the sequence of the sulfurization prior-real-time parameter decision anchor adaptive splicing factor is weighted based on the Softmax function, that is, the sequence of the sulfurization prior-real-time parameter decision anchor adaptive splicing factor is normalized into a weight sequence with the property of probability distribution by using the Softmax function. The Softmax transformation can not only convert the adaptive sulfurization prior-real-time parameter decision anchor adaptive splicing factor into an interpretable weight distribution, but also further amplify the significant differences of local features through the characteristics of the exponential function to enhance the distribution discrimination ability of features. To ensure adaptability in various task scenarios. Specifically, the sequence of the sulfurization prior-real-time parameter decision anchor adaptive splicing factor may have different dimensions or ranges, and directly using them may lead to the problem of unbalanced weight allocation. Therefore, it is necessary to perform normalization processing to convert it into a weight sequence with the property of probability distribution. In the technical solution of this application, the softmax function realizes the normalization operation by applying the exponential operation to each sulfurization prior-real-time parameter decision anchor adaptive splicing factor and dividing it by the sum of the exponents of all factors. This processing method not only ensures that the sum of the finally generated weight factors is equal to 1, but also retains the relative relationship between the original factors. This mechanism enables the system to more flexibly adjust the contribution ratio of each matrix, so as to better balance the relationship between the global trend and local details.

[0040] Specifically, in the step S154, based on the sequence of the sulfurization prior-real-time parameter decision anchor adaptive splicing weight factor, the sequence of the sulfurization prior-real-time parameter local feature response anchor encoding matrix is fused to obtain the sulfurization effect dynamic response encoding vector, which is expressed by the fusion formula as: ; where is the sulfurization effect dynamic response encoding matrix, is the shape reshaping operation, is the dynamic response encoding vector of the vulcanization effect. It should be understood that considering that the vulcanization process is essentially a multivariable, nonlinear and highly dynamic physical and chemical process, its results are affected by many factors, including process parameters such as vulcanization temperature, vulcanization pressure, vulcanization time, and the properties of the material itself. In order to accurately describe the behavior of this complex system, historical experience (i.e., prior data) and current state (i.e., real-time data) must be considered simultaneously. However, simply superimposing or averaging the two may result in information loss or confusion, because they each reflect characteristics at different time and space scales. Therefore, the introduction of the vulcanization prior-real-time parameter decision anchor adaptive splicing weight factor as a bridge can dynamically adjust the contribution ratio of the two according to the specific scenario, so as to better balance the relationship between global trends and local details. In the fusion process, each vulcanization prior-real-time parameter local feature response anchor encoding matrix is ​​assigned a corresponding weight factor. This weight distribution mechanism ensures that important matrices obtain higher weights, while secondary matrices are appropriately reduced in weight, thereby avoiding excessive dilution or distortion of information. Finally, all matrices are weighted and summed according to the proportion of their weight factors, and converted into the form of a dynamic response encoding vector of the vulcanization effect through a shape reshaping operation.

[0041] In the above-mentioned cross-linked cable vulcanization process control method, the step S160, performing full connection encoding on the vulcanization target parameters to obtain the vulcanization target parameter encoding vector, includes: performing the full connection encoding based on the multi-layer perceptron model on the vulcanization target parameters to obtain the vulcanization target parameter encoding vector. It should be understood that considering the complex relationship between these parameters in the vulcanization target parameters and between them and the real-time parameters and prior parameters in the vulcanization process. Based on this, the vulcanization target parameters are fully connected encoded based on the multi-layer perceptron model to obtain the vulcanization target parameter encoding vector. That is, the fully connected structure of the multi-layer perceptron model can deeply explore the potential features and patterns in these parameters, and abstract the vulcanization target parameters from the original specific numerical form to a more representative feature vector. For example, through the nonlinear transformation of the multi-layer perceptron model, the intrinsic connection between different target performance indicators can be found and reflected in the encoding vector, providing more valuable information for subsequent decision-making.

[0042] In the above cross-linked cable vulcanization process control method, in step S170, calculate the vulcanization effect prediction-target offset coding vector between the vulcanization effect dynamic response coding vector and the vulcanization target parameter coding vector. It should be understood that considering that the vulcanization effect dynamic response coding vector reflects the current vulcanization effect predicted based on real-time vulcanization parameters and prior parameters, while the vulcanization target parameter coding vector represents the desired vulcanization target. Therefore, the gap between the prediction result and the target can be clearly presented in a quantitative manner. In this application, by calculating the vulcanization effect prediction-target offset coding vector between the vulcanization effect dynamic response coding vector and the vulcanization target parameter coding vector, the system can accurately know the deviation degree of the current vulcanization process from the ideal state, providing a basis for more accurate adjustment.

[0043] In the above cross-linked cable vulcanization process control method, in step S180, the vulcanization optimization control parameters are obtained based on the vulcanization effect prediction-target offset coding vector, including: inputting the vulcanization effect prediction-target offset coding vector into a decoder-based reinforcement learning engine to obtain the vulcanization optimization control parameters. That is, decoding processing is performed using the vulcanization effect prediction-target offset coding vector obtained by calculating the offset between the vulcanization effect dynamic response coding vector and the vulcanization target parameter coding vector, so as to intelligently obtain the vulcanization optimization control parameters. It should be understood that the decoder-based reinforcement learning engine has a powerful non-linear mapping ability and can handle such complex relationships. It can extract key information from the offset coding vector and convert it into specific adjustment suggestions for control parameters such as temperature, pressure, and time during the vulcanization process, so as to generate optimization control parameters more comprehensively and accurately, thereby improving the vulcanization quality of cross-linked cables and ensuring that product performance meets requirements. In this process, the offset calculation result between the vulcanization effect dynamic response coding vector and the vulcanization target parameter coding vector is used as the core input. With its powerful non-linear mapping ability, the reinforcement learning engine can deeply explore the complex relationship between the two, thus realizing the precise optimization of the vulcanization process. Specifically, the control parameters mainly include key variables such as vulcanization temperature, vulcanization pressure, and vulcanization time. These parameters directly determine the molecular structure change and the final performance of cross-linked cables. For example, temperature control affects the cross-linking reaction rate, pressure adjustment ensures the density of the internal structure of the cable, and time setting determines whether the cross-linking degree is sufficient. The reinforcement learning engine extracts key information about the difference between the current vulcanization state and the target performance through decoding processing of the offset coding vector and converts it into specific adjustment suggestions, such as increasing or decreasing the temperature setting value, increasing or decreasing the vulcanization time, and adjusting the pressure range. This intelligent parameter generation method not only improves the controllability and efficiency of the vulcanization process but also effectively responds to uncertainties in production, ensuring that the mechanical properties (such as tensile strength and elongation at break) and electrical properties (such as insulation resistance and breakdown voltage) of the final product meet the expected standards, thereby comprehensively improving the vulcanization quality of cross-linked cables and meeting the requirements of diverse application scenarios.

[0044] In summary, the cross-linked cable vulcanization process control method based on the embodiments of the present application is elucidated. First, a prior dataset, vulcanization target parameters, and real-time monitoring of the parameters and time during the vulcanization process are obtained to obtain real-time vulcanization parameters. Deep learning-based data analysis and coding techniques are used to encode the prior dataset and real-time vulcanization parameters and perform decision anchor dynamic response. At the same time, the vulcanization target parameters are encoded, and thus, vulcanization optimization control parameters are obtained according to the prediction-target offset representation to more effectively handle the complex non-linear relationships and high-dimensional features existing in the cross-linked cable vulcanization process, thereby improving the prediction accuracy and control effect. At the same time, its internal structure can be dynamically adjusted to adapt to new input variables or changing data distributions, enabling the system to operate efficiently in different types of production environments.

[0045] Figure 5 FIG. is a system block diagram of a cross-linked cable vulcanization process control system according to an embodiment of the present application. As Figure 5 shown, the cross-linked cable vulcanization process control system 100 according to an embodiment of the present application includes: a prior dataset acquisition module 110 for acquiring a prior dataset, and each data sample in the prior dataset is {vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of the cable after vulcanization, elongation at break of the cable after vulcanization, electrical performance of the cable after vulcanization}; a vulcanization target parameter acquisition module 120 for acquiring vulcanization target parameters input by a user, where the vulcanization target parameters include the target tensile strength of the cable after vulcanization, the target elongation at break of the cable after vulcanization, and the target electrical performance of the cable after vulcanization; a real-time vulcanization parameter monitoring module 130 for real-time monitoring of the vulcanization temperature, vulcanization pressure, and vulcanization time during the vulcanization process to obtain real-time vulcanization parameters; a vulcanization parameter encoding module 140 for encoding the prior dataset and the real-time vulcanization parameters respectively to obtain a sequence of vulcanization prior parameter encoding vectors and vulcanization real-time parameter encoding vectors; a decision anchor vulcanization effect dynamic response encoding module 150 for performing decision anchor vulcanization effect dynamic response encoding on the sequence of the vulcanization real-time parameter encoding vectors and the vulcanization prior parameter encoding vectors to obtain a vulcanization effect dynamic response encoding vector; a vulcanization target parameter fully connected module 160 for performing fully connected encoding on the vulcanization target parameters to obtain a vulcanization target parameter encoding vector; a vulcanization effect prediction-target offset encoding module 170 for calculating a vulcanization effect prediction-target offset encoding vector between the vulcanization effect dynamic response encoding vector and the vulcanization target parameter encoding vector; a vulcanization optimization control parameter module 180 for obtaining vulcanization optimization control parameters according to the vulcanization effect prediction-target offset encoding vector.

[0046] Here, those skilled in the art can understand that the specific operations of each step in the above cross-linked cable vulcanization process control system have been described above with reference to Figures 1 to 4The description of the cross-linked cable vulcanization process control method has been introduced in detail, and thus, its repeated description will be omitted.

[0047] As described above, the cross-linked cable vulcanization process control system 100 according to the embodiments of the present application can be implemented in various terminal devices. In one example, the cross-linked cable vulcanization process control system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the cross-linked cable vulcanization process control system 100 can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the cross-linked cable vulcanization process control system 100 can also be one of the many hardware modules of the terminal device.

[0048] Alternatively, in another example, the cross-linked cable vulcanization process control system 100 and the terminal device can also be separate devices, and the cross-linked cable vulcanization process control system 100 can be connected to the terminal device through a wired and / or wireless network, and transmit and interact information in accordance with a predefined data format.

[0049] In summary, the cross-linked cable vulcanization process control system based on the embodiments of the present application is elucidated. It first obtains a prior data set, vulcanization target parameters, and monitors the parameters and time during the vulcanization process in real time to obtain real-time vulcanization parameters. It uses deep learning-based data analysis and coding techniques to encode the prior data set and real-time vulcanization parameters and perform decision anchor dynamic response, while encoding the vulcanization target parameters, so as to obtain vulcanization optimization control parameters according to the prediction-target offset representation, to more effectively handle the complex non-linear relationships and high-dimensional features existing in the cross-linked cable vulcanization process, thereby improving the prediction accuracy and control effect. At the same time, it can dynamically adjust its internal structure to adapt to new input variables or changing data distributions, enabling the system to operate efficiently in different types of production environments.

Claims

1. A method for controlling the vulcanization process of a cross-linked cable, characterized in that: include: Acquire a priori data set, wherein each data sample in the priori data set is vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of cable after vulcanization, elongation at break of cable after vulcanization, and electrical performance of cable after vulcanization; acquire vulcanization target parameters input by a user, wherein the vulcanization target parameters include target tensile strength of cable after vulcanization, target elongation at break of cable after vulcanization, and target electrical performance of cable after vulcanization; monitor the vulcanization temperature, vulcanization pressure and vulcanization time during the vulcanization process in real time to obtain real-time vulcanization parameters; encode the priori data set and the real-time vulcanization parameters respectively to obtain the real-time vulcanization parameters; A sequence of vulcanization prior parameter coding vectors and a vulcanization real-time parameter coding vector are obtained; decision anchor point vulcanization effect dynamic response coding is performed on the sequence of the vulcanization real-time parameter coding vector and the vulcanization prior parameter coding vector to obtain a vulcanization effect dynamic response coding vector; full connection coding is performed on the vulcanization target parameter to obtain a vulcanization target parameter coding vector; a vulcanization effect prediction-target offset coding vector is calculated between the vulcanization effect dynamic response coding vector and the vulcanization target parameter coding vector; and vulcanization optimization control parameters are obtained according to the vulcanization effect prediction-target offset coding vector.

2. The cross-linked cable vulcanization process control method according to claim 1, characterized in that: The priori data set and the real-time vulcanization parameters are respectively encoded to obtain a sequence of vulcanization priori parameter encoding vectors and a vulcanization real-time parameter encoding vector, including: performing full connection encoding based on a multi-layer perceptron model on each data sample in the priori data set to obtain a sequence of the vulcanization priori parameter encoding vectors; performing the full connection encoding based on the multi-layer perceptron model on the real-time vulcanization parameters to obtain the vulcanization real-time parameter encoding vector.

3. The cross-linked cable vulcanization process control method according to claim 2, characterized in that: The decision anchor point vulcanization effect dynamic response coding is performed on the sequence of the vulcanization real-time parameter coding vector and the vulcanization prior parameter coding vector to obtain the vulcanization effect dynamic response coding vector, including: performing deep implicit feature extraction on the sequence of the vulcanization real-time parameter coding vector and the vulcanization prior parameter coding vector to obtain a sequence of the vulcanization real-time parameter deep implicit coding vector and the vulcanization prior parameter deep implicit coding vector; inputting each vulcanization prior parameter deep implicit coding vector in the sequence of the vulcanization real-time parameter deep implicit coding vector and the vulcanization prior parameter deep implicit coding vector into the semantic response vector respectively. A decision anchoring component is used to obtain a sequence of vulcanization prior-real-time parameter local feature response anchoring coding matrices; the decision anchor adaptive splicing weight factors of each vulcanization prior-real-time parameter local feature response anchoring coding matrix in the sequence of vulcanization prior-real-time parameter local feature response anchoring coding matrices are calculated to obtain a sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors; based on the sequence of vulcanization prior-real-time parameter decision anchor adaptive splicing weight factors, the sequence of vulcanization prior-real-time parameter local feature response anchoring coding matrices is fused to obtain the vulcanization effect dynamic response coding vector.

4. The cross-linked cable vulcanization process control method according to claim 3, characterized in that: Performing deep implicit feature extraction on the sequence of the vulcanization real-time parameter coding vector and the vulcanization prior parameter coding vector to obtain a sequence of the vulcanization real-time parameter deep implicit coding vector and the vulcanization prior parameter deep implicit coding vector, including: performing deep implicit feature extraction based on fully connected coding on the vulcanization real-time parameter coding vector to obtain the vulcanization real-time parameter deep implicit coding vector; performing deep implicit feature extraction based on fully connected coding on each vulcanization prior parameter coding vector in the sequence of the vulcanization prior parameter coding vector to obtain a sequence of the vulcanization prior parameter deep implicit coding vector.

5. The cross-linked cable vulcanization process control method according to claim 4, characterized in that: The decision anchor adaptive splicing weight factors of each sulfurization prior-real-time parameter local feature response anchor coding matrix in the sequence of the sulfurization prior-real-time parameter local feature response anchor coding matrix are calculated to obtain a sequence of sulfurization prior-real-time parameter decision anchor adaptive splicing weight factors, including: based on the feature distribution of each sulfurization prior-real-time parameter local feature response anchor coding matrix in the sequence of the sulfurization prior-real-time parameter local feature response anchor coding matrix, determining the decision anchor adaptive splicing factors of each sulfurization prior-real-time parameter local feature response anchor coding matrix to obtain a sequence of sulfurization prior-real-time parameter decision anchor adaptive splicing factors; performing weighting processing based on the Softmax function on the sequence of sulfurization prior-real-time parameter decision anchor adaptive splicing factors to obtain a sequence of sulfurization prior-real-time parameter decision anchor adaptive splicing weight factors.

6. The cross-linked cable vulcanization process control method according to claim 5, characterized in that: Based on the characteristic distribution of each sulfurization prior-real-time parameter local feature response anchor coding matrix in the sequence of the sulfurization prior-real-time parameter local feature response anchor coding matrix, the decision anchor adaptive splicing factor of each sulfurization prior-real-time parameter local feature response anchor coding matrix is ​​determined to obtain a sequence of sulfurization prior-real-time parameter decision anchor adaptive splicing factors, including: based on the maximum value, mean, variance and difference amplification coefficient and drift coefficient of the sulfurization prior-real-time parameter local feature response anchor coding matrix, the decision anchor adaptive splicing factor of the sulfurization prior-real-time parameter local feature response anchor coding matrix is ​​obtained; wherein the drift coefficient is obtained by performing local-to-global smooth transition processing on the sulfurization prior-real-time parameter local feature response anchor coding matrix.

7. The cross-linked cable vulcanization process control method according to claim 6, characterized in that: Performing full connection encoding on the vulcanization target parameter to obtain a vulcanization target parameter encoding vector includes: performing full connection encoding based on the multi-layer perceptron model on the vulcanization target parameter to obtain the vulcanization target parameter encoding vector.

8. The cross-linked cable vulcanization process control method according to claim 7, characterized in that: According to the vulcanization effect prediction-target offset coding vector, a vulcanization optimization control parameter is obtained, including: inputting the vulcanization effect prediction-target offset coding vector into a decoder-based reinforcement learning engine to obtain the vulcanization optimization control parameter.

9. A cross-linked cable vulcanization process control system, characterized in that: include: A priori data set acquisition module is used to acquire a priori data set, wherein each data sample in the priori data set is vulcanization temperature, vulcanization pressure, vulcanization time, tensile strength of cable after vulcanization, elongation at break of cable after vulcanization, and electrical performance of cable after vulcanization; a vulcanization target parameter acquisition module is used to acquire vulcanization target parameters input by a user, wherein the vulcanization target parameters include target tensile strength of cable after vulcanization, target elongation at break of cable after vulcanization, and target electrical performance of cable after vulcanization; a real-time vulcanization parameter monitoring module is used to monitor the vulcanization temperature, vulcanization pressure and vulcanization time in the vulcanization process in real time to obtain real-time vulcanization parameters; A vulcanization parameter encoding module is used to encode the priori data set and the real-time vulcanization parameters respectively to obtain a sequence of vulcanization priori parameter encoding vectors and a vulcanization real-time parameter encoding vector; a decision anchor vulcanization effect dynamic response encoding module is used to perform decision anchor vulcanization effect dynamic response encoding on the sequence of the vulcanization real-time parameter encoding vector and the vulcanization priori parameter encoding vector to obtain a vulcanization effect dynamic response encoding vector; a vulcanization target parameter fully connected module is used to perform fully connected encoding on the vulcanization target parameter to obtain a vulcanization target parameter encoding vector; a vulcanization effect prediction-target offset encoding module is used to calculate the vulcanization effect prediction-target offset encoding vector between the vulcanization effect dynamic response encoding vector and the vulcanization target parameter encoding vector; a vulcanization optimization control parameter module is used to obtain the vulcanization optimization control parameter according to the vulcanization effect prediction-target offset encoding vector.

10. The cross-linked cable vulcanization process control system according to claim 9, characterized in that: The vulcanization parameter encoding module is used to: perform full connection encoding based on a multi-layer perceptron model on each data sample in the priori data set to obtain a sequence of the vulcanization priori parameter encoding vectors; perform full connection encoding based on a multi-layer perceptron model on the real-time vulcanization parameter to obtain the vulcanization real-time parameter encoding vector.

Images