On-line defect detection method and system for cable insulation layer

Abstract

The invention provides an online defect detection method and system for a cable insulation layer, and relates to the technical field of cables, and the method comprises the steps: obtaining an extrusion monitoring data set based on cable production; insulating layer defect online detection is carried out according to the extrusion monitoring data set; performing trend prediction according to the insulating layer defect first map; performing feedback adjustment on the insulation extrusion scheme according to the insulation layer defect second map to obtain an extrusion adjustment space; performing defect risk optimization on the extrusion adjustment space to obtain an extrusion candidate population; and based on the multi-dimensional defect risk prediction model, performing differential evolution optimization on the extrusion adjustment space according to the extrusion candidate population to obtain an extrusion adjustment optimization result, and executing cable production online optimization according to the extrusion adjustment optimization result. According to the invention, the technical problem of low defect detection accuracy of the cable insulation layer in the prior art can be solved, and the technical effect of improving the defect detection accuracy of the cable insulation layer is achieved.

Description

Technical Field

[0001] This application relates to the field of cable technology, and in particular to an online detection method and system for defects in cable insulation. Background Technology

[0002] With the continuous expansion of cable insulation production scale and the increase in production speed, the requirements for precise control of extrusion process parameters are also becoming increasingly stringent, including the coordinated optimization of multiple parameters such as screw speed, extrusion temperature, cooling rate, and traction speed. When the cable production line operates at high speed, any minute fluctuation in extrusion can lead to defects such as bubbles, depressions, and uneven fiber formation on the insulation surface. Simultaneously, microcracks or pores may form internally, further affecting insulation performance and electrical safety indicators.

[0003] Currently, existing cable insulation production technologies mainly rely on fixed extrusion schemes and empirical parameter adjustments, lacking real-time monitoring and intelligent control of the production process. Although some production lines are equipped with temperature and pressure sensors, these monitoring methods typically operate independently, failing to comprehensively analyze the correlation between surface defects, internal defects, and electrical performance, nor can they dynamically optimize and adjust for abnormal conditions. Existing defect detection methods largely depend on manual sampling or data collection from a single sensor, resulting in detection delays, low accuracy, and the inability to predict defect development trends.

[0004] In summary, the existing technology suffers from a lack of real-time monitoring, correlation analysis, and dynamic optimization adjustment mechanisms for multidimensional defects during the cable insulation production process. This results in the inability to detect and correct surface defects, internal cracks, and abnormal electrical performance in a timely manner, further affecting the overall quality, reliability, and production efficiency of the cable. Summary of the Invention

[0005] The purpose of this application is to provide an online detection method and system for defects in cable insulation layers, in order to solve the technical problem in the prior art that the lack of real-time monitoring, correlation analysis and dynamic optimization adjustment mechanism for multi-dimensional defects in cable insulation layers during the production process leads to the inability to detect and correct surface defects, internal cracks and abnormal electrical performance in a timely manner, which further affects the overall quality, reliability and production efficiency of the cable.

[0006] In view of the above problems, this application provides an online detection method and system for defects in cable insulation.

[0007] Firstly, this application provides an online defect detection method for cable insulation layers, implemented through an online defect detection system for cable insulation layers, comprising: performing cable production based on an insulation extrusion scheme and simultaneously acquiring an extrusion monitoring dataset; performing online defect detection of the insulation layer based on the extrusion monitoring dataset to construct a first defect map of the insulation layer; performing trend prediction based on the first defect map of the insulation layer to obtain a second defect map of the insulation layer; performing feedback adjustment of the insulation extrusion scheme based on the second defect map of the insulation layer to obtain an extrusion adjustment space; introducing a multidimensional defect risk prediction model to optimize the defect risk of the extrusion adjustment space to obtain an extrusion candidate population; performing differential evolution optimization of the extrusion adjustment space based on the multidimensional defect risk prediction model and the extrusion candidate population to obtain an extrusion adjustment optimization result, and performing online optimization of cable production based on the extrusion adjustment optimization result.

[0008] Preferably, the online defect detection method for cable insulation further includes: performing surface defect detection on the insulation layer based on the extrusion monitoring dataset to obtain surface defect detection results; performing internal defect detection on the insulation layer based on the extrusion monitoring dataset to obtain internal defect detection results; performing electrical performance defect detection on the insulation layer based on the extrusion monitoring dataset to obtain electrical performance defect detection results; and organizing the surface defect detection results, the internal defect detection results, and the electrical performance defect detection results to obtain a first defect map of the insulation layer.

[0009] Preferably, the online defect detection method for cable insulation further includes: capturing surface defect association features of the insulation layer based on the extrusion monitoring dataset to obtain a surface defect association feature set; performing supervised training based on the surface defect detection record set of the insulation layer to obtain a surface defect detection model; injecting perturbation into the surface defect detection record set of the insulation layer using an adversarial example generator to obtain a surface defect detection adversarial example set; performing robustness enhancement training on the surface defect detection model based on the surface defect detection adversarial example set to generate a surface defect detection channel; and inputting the surface defect association feature set into the surface defect detection channel to obtain the surface defect detection result.

[0010] Preferably, the online defect detection method for cable insulation further includes: activating the multidimensional defect risk prediction model, which includes a surface defect risk prediction model, an internal defect risk prediction model, and an electrical performance defect risk prediction model; optimizing the surface defect risk of the extrusion adjustment space according to the surface defect risk prediction model to obtain a first extrusion candidate seed; optimizing the internal defect risk of the extrusion adjustment space according to the internal defect risk prediction model to obtain a second extrusion candidate seed; optimizing the electrical performance defect risk of the extrusion adjustment space according to the electrical performance defect risk prediction model to obtain a third extrusion candidate seed; and adding the first extrusion candidate seed, the second extrusion candidate seed, and the third extrusion candidate seed to the extrusion candidate population.

[0011] Preferably, the online defect detection method for cable insulation further includes: performing simulated production based on each extrusion adjustment scheme within the extrusion adjustment space to obtain extrusion simulation data for each scheme; performing risk analysis on the extrusion simulation data for each scheme based on the surface defect risk prediction model to construct a surface defect risk sequence; and performing iterative optimization of the surface defect risk in the extrusion adjustment space based on the surface defect risk sequence to obtain the first seed of the extrusion candidate.

[0012] Preferably, the online defect detection method for cable insulation further includes: based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the first extrusion candidate seed to obtain an extrusion evolution first space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the second extrusion candidate seed to obtain an extrusion evolution second space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the third extrusion candidate seed to obtain an extrusion evolution third space; configuring the weights of the multidimensional defect risk indicators based on the multidimensional defect risk prediction model to establish a global defect risk analysis model; and performing global defect risk minimization joint optimization on the first extrusion evolution space, the second extrusion evolution space, and the third extrusion evolution space according to the global defect risk analysis model to obtain the extrusion adjustment optimization result.

[0013] Preferably, the online defect detection method for cable insulation further includes: performing difference detection on the extrusion adjustment space based on the extrusion candidate first seed to obtain multiple extrusion difference vectors; performing mutation expansion on the multiple extrusion difference vectors to obtain multiple extrusion adjustment vector sets; performing cross mutation on the extrusion adjustment space based on the multiple extrusion adjustment vector sets to obtain a first extrusion adjustment mutation domain; and performing constraint optimization on the first extrusion adjustment mutation domain based on the multidimensional defect risk prediction model and multidimensional defect risk constraints to generate the first extrusion evolution space.

[0014] Preferably, the online defect detection method for cable insulation further includes: obtaining an extrusion monitoring data stream; cleaning the extrusion monitoring data stream to obtain the extrusion monitoring dataset.

[0015] Preferably, the online defect detection method for cable insulation further includes: obtaining an insulation defect alarm based on the second defect map of the insulation layer.

[0016] Secondly, this application also provides an online defect detection system for cable insulation layers, used to execute an online defect detection method for cable insulation layers as described in the first aspect, comprising: an extrusion monitoring dataset acquisition module, used to acquire an extrusion monitoring dataset simultaneously during cable production based on an insulation extrusion scheme; an insulation layer defect first map construction module, used to construct an insulation layer defect first map based on the extrusion monitoring dataset for online defect detection of the insulation layer; an insulation layer defect second map acquisition module, used to perform trend prediction based on the insulation layer defect first map to obtain an insulation layer defect second map; an extrusion adjustment space acquisition module, used to perform feedback adjustment of the insulation extrusion scheme based on the insulation layer defect second map to obtain an extrusion adjustment space; an extrusion candidate population acquisition module, used to introduce a multidimensional defect risk prediction model to optimize the defect risk of the extrusion adjustment space to obtain an extrusion candidate population; and an extrusion adjustment optimization result acquisition module, used to perform differential evolution optimization of the extrusion adjustment space based on the multidimensional defect risk prediction model and the extrusion candidate population to obtain an extrusion adjustment optimization result, and to perform online optimization of cable production based on the extrusion adjustment optimization result.

[0017] The technical solution provided in this application has at least the following technical effects or advantages: by achieving the technical goal of online defect detection and intelligent optimization adjustment in the cable insulation layer production process, it can detect and correct surface, internal and electrical performance defects in real time during the production process, thereby improving cable product quality, reducing defect rate, increasing production efficiency and ensuring reliability.

[0018] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an online defect detection method for cable insulation layers according to this application.

[0021] Figure 2 This is a schematic diagram of the structure of an online defect detection system for cable insulation layer according to this application.

[0022] Figure labeling: Module 1 for obtaining extrusion monitoring dataset, Module 2 for constructing the first spectrum of insulation layer defects, Module 3 for obtaining the second spectrum of insulation layer defects, Module 4 for obtaining extrusion adjustment space, Module 5 for obtaining extrusion candidate population, and Module 6 for obtaining extrusion adjustment optimization results. Detailed Implementation

[0023] This application provides an online defect detection method and system for cable insulation layers, solving the technical problem in existing technologies where the lack of real-time monitoring, correlation analysis, and dynamic optimization adjustment mechanisms for multi-dimensional defects during cable insulation layer production leads to the inability to promptly detect and correct surface defects, internal cracks, and abnormal electrical performance during production, further affecting the overall quality, reliability, and production efficiency of cables. The method achieves the technical goal of online defect detection and intelligent optimization adjustment during the cable insulation layer production process, enabling real-time detection and correction of surface, internal, and electrical performance defects during production, thereby improving cable product quality, reducing defect rates, increasing production efficiency, and ensuring reliability.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides an online detection method for defects in cable insulation layers, applied to an online detection system for defects in cable insulation layers, specifically including the following steps: S1: Cable production is carried out based on the insulation extrusion scheme, and extrusion monitoring datasets are acquired simultaneously.

[0026] Specifically, cable production is based on an insulation extrusion scheme, which involves manufacturing the cable insulation layer according to pre-set production processes and parameters. This insulation extrusion scheme includes information such as temperature profiles, pressure settings, screw speed, cooling rate, and material ratios. Simultaneously with the extrusion process, sensors and data acquisition systems collect multi-dimensional data in real time to obtain an extrusion monitoring dataset. For example, temperature sensors record the temperature of the heating zone at different locations, pressure sensors monitor the pressure inside the extruder cavity, and displacement sensors measure the material flow rate.

[0027] S2: Based on the extrusion monitoring dataset, perform online detection of insulation layer defects and construct the first spectrum of insulation layer defects.

[0028] Specifically, online detection of insulation defects is performed based on extrusion monitoring datasets. This involves utilizing real-time extrusion monitoring data collected during cable production, such as temperature, pressure, flow rate, and surface image data. Through automated analysis and model calculations, various defects appearing in the insulation layer during production can be identified in real time, including surface bubbles, scratches, internal cracks, or abnormal electrical performance. For example, out of 500 monitoring data points collected every minute, two surface bubbles and one internal crack may be detected, thus providing immediate feedback on production quality. The defect information obtained from online detection is integrated, visualized, and structured to construct a primary atlas of insulation defects. This atlas comprehensively displays the distribution, type, quantity, and severity of insulation defects. Each defect type corresponds to specific indicators, such as the number of surface defects, the length of internal cracks, and electrical performance deviations, which can intuitively reflect the overall situation and trends of defects during production.

[0029] S3: Based on the first spectrum of insulation layer defects, perform trend prediction to obtain the second spectrum of insulation layer defects.

[0030] Specifically, trend prediction is performed based on the first defect map of the insulation layer. This involves using historical data on defect types, quantities, distributions, and severity contained in the constructed first map, and employing time series analysis, machine learning, or predictive models to estimate potential defect trends in the insulation layer over a future period. Trend prediction can consider the rate of increase in defect quantity, the direction of defect expansion, and the impact of changes in production conditions. For example, if the number of surface defects increased from 10 to 30 in the past hour of production, the trend prediction model might predict that the number of defects will increase to 45 in the next 30 minutes. The results of the trend prediction are then visualized and structured to generate a new map, resulting in the second defect map of the insulation layer. This second map displays the types, distribution ranges, and severity of potential future defects, incorporating not only the current defect status but also predicted data. This provides a reference for production control and quality management. For instance, based on predictions, the number of internal cracks might increase from the current 8 to 12, and the number of cable samples with electrical performance anomalies might increase by 1.5 meters. This information is visually presented in the second map.

[0031] S4: The insulation extrusion scheme is adjusted based on the second pattern of insulation layer defects to obtain the extrusion adjustment space.

[0032] Specifically, the insulation extrusion scheme is adjusted based on the second defect map of the insulation layer. This involves using the defect prediction information contained in the second map, such as the potential number of surface bubbles, the length of internal cracks, or abnormal distribution of electrical properties, to adjust the current extrusion production scheme. This includes modifying parameters such as screw speed, temperature profile, cooling rate, or material ratio, thereby preventing or reducing defects in advance. For example, if it is predicted that the number of surface bubbles may increase by 50 in the next 30 minutes, the production system can reduce the extrusion temperature by 5 degrees Celsius or increase the cooling airflow to mitigate the defect risk. All possible adjustment schemes and their corresponding parameter ranges are combined into a multi-dimensional space to obtain the extrusion adjustment space, including feasible production schemes under different combinations of extrusion parameters. The potential defect risk corresponding to each scheme can also be reflected in this space. For example, the screw speed can vary between 100 rpm and 140 rpm, the extrusion temperature between 180 degrees Celsius and 200 degrees Celsius, and the cooling rate between 1 m / min and 3 m / min.

[0033] S5: Introduce a multidimensional defect risk prediction model to optimize the defect risk of the extrusion adjustment space and obtain the extrusion candidate population.

[0034] Specifically, a multidimensional defect risk prediction model is introduced to optimize defect risk in the extrusion adjustment space. This involves applying the multidimensional defect risk prediction model to the extrusion adjustment space to assess and optimize the defect risk for each possible combination of extrusion parameters. The multidimensional defect risk prediction model includes surface defect risk prediction, internal defect risk prediction, and electrical performance defect risk prediction models, each calculating potential risk values ​​for different defect types. Through defect risk optimization, several schemes that perform well across different defect dimensions are selected from the extrusion adjustment space, and these schemes are aggregated to form an extrusion candidate population, with each member corresponding to a specific combination of extrusion parameters.

[0035] S6: Based on the multidimensional defect risk prediction model, perform differential evolution optimization on the extrusion adjustment space according to the extrusion candidate population to obtain the extrusion adjustment optimization result, and perform online optimization of cable production according to the extrusion adjustment optimization result.

[0036] Specifically, based on a multidimensional defect risk prediction model, differential evolution optimization is performed on the extrusion adjustment space according to the extrusion candidate population. This involves using the multidimensional defect risk prediction model to assess the risk of each scheme in the extrusion candidate population, and then exploring and optimizing parameter combinations in the extrusion adjustment space using a differential evolution algorithm to gradually approach the scheme with the minimum defect risk. Multidimensional defect risks include surface defects, internal defects, and electrical performance defects. For example, when there are 10 schemes in the candidate population, the differential evolution process may go through 5 iterations, reducing the number of surface bubbles from 10 to 2, the number of internal cracks from 5 to 0, and the electrical performance anomaly risk value from 0.3 to 0.1. After differential evolution optimization, the extrusion parameter combination that performs best in terms of multidimensional defect risk is selected, obtaining the extrusion adjustment optimization results, including screw speed, extrusion temperature, and cooling rate, which can be directly applied to actual production.

[0037] Based on the results of extrusion adjustment optimization, online optimization of cable production is performed. The obtained optimal extrusion parameter combination is directly applied to the cable production process to realize real-time adjustment of production parameters, thereby reducing the defect rate of cable insulation layer. For example, in the continuous production of 100 meters of cable, the number of surface bubbles may be reduced from 8 to 1, the number of internal cracks may be reduced from 2 to 0, the electrical performance abnormality rate may be reduced by 50%, and the production quality may be significantly improved.

[0038] Furthermore, this application also includes: performing surface defect detection on the insulation layer based on the extrusion monitoring dataset to obtain surface defect detection results; performing internal defect detection on the insulation layer based on the extrusion monitoring dataset to obtain internal defect detection results; performing electrical performance defect detection on the insulation layer based on the extrusion monitoring dataset to obtain electrical performance defect detection results; and organizing the surface defect detection results, the internal defect detection results, and the electrical performance defect detection results to obtain a first defect map of the insulation layer.

[0039] Specifically, surface defects of the insulation layer are detected based on the extrusion monitoring dataset to obtain surface defect detection results. That is, by using data such as temperature, pressure, and flow rate collected during the production process, combined with image acquisition and surface optical inspection, the appearance of the cable insulation layer is analyzed to identify whether there are surface defects such as bubbles, scratches, pits, or uneven textures.

[0040] Internal defects in the insulation layer are detected based on the extrusion monitoring dataset. The results of the internal defect detection are obtained by using methods such as ultrasonic, infrared imaging or electromagnetic detection, combined with extrusion data to analyze the internal structural integrity of the insulation layer, thereby identifying internal defects such as cracks, voids or impurities.

[0041] Based on the extrusion monitoring dataset, electrical performance defects of the insulation layer are detected to obtain the results. This involves using electrical strength testing, withstand voltage testing, or dielectric loss factor measurement, combined with electrical parameters from the extrusion process, to determine whether the insulation layer has a risk of breakdown or poor insulation. For example, when sampling 100 meters of cable during production, 2 meters of the sample may show electrical performance defects with high dielectric loss.

[0042] By compiling the surface defect detection results, internal defect detection results, and electrical performance defect detection results, a first defect map of the insulation layer is obtained. This map integrates and visualizes the defect results from the three dimensions, displaying the types, quantities, distribution, and severity of defects in a graphical form. It can intuitively reflect the overall defect situation of the insulation layer in a certain time period or batch. For example, in 10 batches of production data, if the number of surface defects increases from 20 to 50, the number of internal defects increases from 5 to 15, and the number of electrical performance defect samples increases from 2 meters to 6 meters, then the first defect map can clearly show the trend of defect growth with batches.

[0043] Furthermore, this application also includes: capturing surface defect association features of the insulation layer based on the extrusion monitoring dataset to obtain a surface defect association feature set; performing supervised training based on the surface defect detection record set of the insulation layer to obtain a surface defect detection model; injecting perturbation into the surface defect detection record set of the insulation layer using an adversarial example generator to obtain a surface defect detection adversarial example set; performing robustness enhancement training on the surface defect detection model based on the surface defect detection adversarial example set to generate a surface defect detection channel; and inputting the surface defect association feature set into the surface defect detection channel to obtain the surface defect detection result.

[0044] Specifically, the surface defect correlation feature is captured based on the extrusion monitoring dataset to obtain the surface defect correlation feature set. That is, by using data such as temperature, pressure and speed collected from the extrusion process, signal features that may be related to surface defects of the insulation layer are extracted, including texture fluctuations, temperature anomalies, pressure fluctuation amplitudes and surface gloss changes, which together constitute the surface defect correlation feature set.

[0045] Supervised training is performed on the insulation layer surface defect detection record set to obtain a surface defect detection model. This involves inputting the historical detection dataset with labeled defect categories and locations into a machine learning or deep learning algorithm. Through the training process, the learning model learns the correspondence between different defects and features. For example, a convolutional neural network is used to learn the correspondence between surface bubbles and temperature fluctuations. Finally, a surface defect detection model that can automatically identify surface defects is obtained.

[0046] By injecting perturbations into the insulation layer surface defect detection record set using an adversarial sample generator, a surface defect detection adversarial sample set is obtained. That is, by introducing adversarial sample generation technology, slight perturbations are added to the original detection data, such as adding random noise to image pixels or adding small deviations to numerical features, thereby generating new adversarial samples and making the dataset more diverse.

[0047] The surface defect detection model is robustly enhanced by training based on the adversarial sample set for surface defect detection, thereby generating a surface defect detection channel. This involves retraining the detection model using both original and adversarial samples, enabling the surface defect detection model to maintain stable detection capabilities even when facing noise, disturbances, or abnormal conditions, resulting in a more robust surface defect detection channel.

[0048] The surface defect associated feature set is input into the surface defect detection channel to obtain the surface defect detection result. That is, the captured surface defect related features are input into the enhanced surface defect detection channel. The specific identification result of the defect is obtained through the calculation of the surface defect detection channel, and the defect type and location are clearly stated in the output.

[0049] Furthermore, this application also includes: activating the multidimensional defect risk prediction model, which includes a surface defect risk prediction model, an internal defect risk prediction model, and an electrical performance defect risk prediction model; optimizing the surface defect risk of the extrusion adjustment space according to the surface defect risk prediction model to obtain a first extrusion candidate seed; optimizing the internal defect risk of the extrusion adjustment space according to the internal defect risk prediction model to obtain a second extrusion candidate seed; optimizing the electrical performance defect risk of the extrusion adjustment space according to the electrical performance defect risk prediction model to obtain a third extrusion candidate seed; and adding the first extrusion candidate seed, the second extrusion candidate seed, and the third extrusion candidate seed to the extrusion candidate population.

[0050] Specifically, activating the multidimensional defect risk prediction model means launching a computational model for assessing and predicting the potential defects in cable insulation. This model comprehensively considers the influencing factors of multiple defect types and can simultaneously analyze the potential risks of surface defects, internal defects, and electrical performance defects. The multidimensional defect risk prediction model includes a surface defect risk prediction model, an internal defect risk prediction model, and an electrical performance defect risk prediction model. Each sub-model performs risk assessment for different defect types. For example, the surface defect risk prediction model can predict the risk of bubbles based on temperature fluctuations and changes in surface texture.

[0051] The surface defect risk prediction model is used to optimize the surface defect risk in the extrusion adjustment space to obtain the first extrusion candidate seed. That is, in the extrusion adjustment space, all possible combinations of extrusion parameters are analyzed, the surface defect risk of each scheme is evaluated using the surface defect risk prediction model, and the scheme with the minimum surface defect risk is selected as the first extrusion candidate seed.

[0052] The internal defect risk prediction model is used to optimize the internal defect risk of the extrusion adjustment space and obtain the second candidate seed for extrusion. That is, the internal defect risk is analyzed for the same combination of extrusion parameters using the internal defect risk prediction model, and the scheme with the minimum internal defect risk is selected as the second candidate seed.

[0053] Based on the electrical performance defect risk prediction model, the electrical performance defect risk of the extrusion adjustment space is optimized to obtain the third candidate seed for extrusion. That is, the electrical performance defect risk prediction model is used to predict the possible electrical performance defect risk for each combination of parameters in the extrusion adjustment space, and the combination with the smallest electrical performance defect risk is selected as the third candidate seed. For example, the scheme with the smallest predicted electrical performance defect risk among the same 100 combinations is selected as the third seed.

[0054] The first, second, and third extrusion candidate seeds are added to the extrusion candidate population, which means that the optimal solutions selected by each of the three sub-models are combined to form a candidate population containing the optimal extrusion solutions for different defect types, providing a basis for subsequent comprehensive optimization.

[0055] Furthermore, this application also includes: performing simulated production based on each extrusion adjustment scheme within the extrusion adjustment space to obtain extrusion simulation data for each scheme; performing risk analysis on the extrusion simulation data for each scheme based on the surface defect risk prediction model to construct a surface defect risk sequence; and performing iterative optimization of the surface defect risk in the extrusion adjustment space based on the surface defect risk sequence to obtain the first seed of the extrusion candidate.

[0056] Specifically, simulation production is carried out based on various extrusion adjustment schemes within the extrusion adjustment space to obtain extrusion simulation data for each scheme. This involves simulating the cable production process in a simulation environment using different preset parameter combinations within the extrusion adjustment space, such as screw speed, extrusion temperature, and cooling rate. This generates the result data that each scheme may produce in actual production. The simulation data includes surface texture, internal structure, and temperature and pressure curves. For example, in the simulation of 20 extrusion schemes, the number of surface bubbles for each scheme may range from 5 to 20.

[0057] Based on the surface defect risk prediction model, risk analysis is performed on the extrusion simulation data of each scheme, and a surface defect risk sequence is constructed. That is, the surface defect risk prediction model is used to analyze the data of each simulation scheme, calculate the corresponding surface defect risk value, including the surface defect risk coefficient corresponding to each extrusion adjustment scheme within the extrusion adjustment space, and form a risk sequence according to the scheme sequence to reflect the changing trend of surface defect risk under different schemes. For example, in 20 schemes, the risk value may increase from 0.1 to 0.8, forming a surface defect risk sequence from low to high.

[0058] Based on the surface defect risk sequence, the extrusion adjustment space is iteratively optimized to obtain the first extrusion candidate seed. That is, under the guidance of the risk sequence, the schemes in the extrusion adjustment space are iteratively optimized, and the parameter combination is gradually adjusted to reduce the surface defect risk. Finally, the scheme with the lowest risk, that is, the one with the smallest surface defect risk coefficient, is selected as the first extrusion candidate seed.

[0059] Furthermore, this application also includes: based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the first extrusion candidate seed to obtain a first extrusion evolution space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the second extrusion candidate seed to obtain a second extrusion evolution space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the third extrusion candidate seed to obtain a third extrusion evolution space; configuring weights for the multidimensional defect risk indicators based on the multidimensional defect risk prediction model to establish a global defect risk analysis model; and performing joint optimization to minimize global defect risk on the first extrusion evolution space, the second extrusion evolution space, and the third extrusion evolution space according to the global defect risk analysis model to obtain the extrusion adjustment optimization result.

[0060] Specifically, based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the first extrusion candidate seed to obtain the first extrusion evolution space. That is, with the first candidate seed as the benchmark, the multidimensional defect risk prediction model is used to perform differential calculation, mutation and crossover operations on each parameter combination in the extrusion adjustment space to generate a set of optimized extrusion schemes. For example, in the initial 100 parameter combinations, differential evolution may obtain 15 schemes with the smallest surface defects as the first space.

[0061] Based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the extrusion candidate second seed to obtain the extrusion evolution second space. That is, with the candidate second seed as the benchmark, the differential evolution algorithm is applied to explore the extrusion adjustment space, focusing on optimizing the internal defect risk, so as to obtain the set of extrusion schemes with the lowest internal defect risk. For example, after iterative calculation, 12 schemes with the fewest internal cracks may be selected from 100 schemes as the second space.

[0062] Based on the multidimensional defect risk prediction model, the extrusion adjustment space is differentially evolved according to the candidate third seed to obtain the extrusion evolution third space. That is, based on the candidate third seed, the extrusion adjustment space is optimized using the differential evolution algorithm, focusing on reducing the risk of electrical performance defects, thereby forming a set of optimal solutions for electrical performance defects. For example, among 100 parameter combinations, 10 solutions with the minimum dielectric loss may be selected as the third space.

[0063] Based on a multidimensional defect risk prediction model, a global defect risk analysis model is established by weighting multidimensional defect risk indicators. This involves assigning weights to risk indicators for surface defects, internal defects, and electrical performance defects. The multidimensional defect risk indicators include surface defect risk indicators, internal defect risk indicators, and electrical performance defect risk indicators. The global defect risk analysis model includes a global defect risk analysis formula: Global Defect Risk Coefficient = Surface Defect Risk Indicator Weight × Surface Defect Risk Coefficient + Internal Defect Risk Indicator Weight × Internal Defect Risk Coefficient + Electrical Performance Defect Risk Indicator Weight × Electrical Performance Defect Risk Coefficient. For example, with a surface defect risk indicator weight of 40%, an internal defect risk indicator weight of 35%, and an electrical performance defect risk indicator weight of 25%, a global analysis model reflecting the overall defect risk is established by integrating risks from all dimensions to guide multidimensional optimization decisions.

[0064] Based on the global defect risk analysis model, the extrusion evolution first space, extrusion evolution second space and extrusion evolution third space are jointly optimized to minimize the global defect risk, and the extrusion adjustment optimization result is obtained. That is, the global defect risk analysis model is used to jointly evaluate and optimize the schemes in the three evolution spaces, and select the comprehensive optimal scheme with the minimum surface, internal and electrical performance defect risks. For example, the final three global optimal schemes are selected from 15 schemes in the first space, 12 schemes in the second space and 10 schemes in the third space.

[0065] Furthermore, this application also includes: performing difference detection on the extrusion adjustment space based on the first extrusion candidate seed to obtain multiple extrusion difference vectors; performing mutation expansion on the multiple extrusion difference vectors to obtain multiple extrusion adjustment vector sets; performing cross mutation on the extrusion adjustment space based on the multiple extrusion adjustment vector sets to obtain a first extrusion adjustment mutation domain; and performing constraint optimization on the first extrusion adjustment mutation domain based on the multidimensional defect risk prediction model and multidimensional defect risk constraints to generate the first extrusion evolution space.

[0066] Specifically, the extrusion adjustment space is analyzed based on the first extrusion candidate seed to obtain multiple extrusion difference vectors. That is, the extrusion parameter combination corresponding to the first candidate seed is used as a benchmark to compare the parameter differences of other schemes in the extrusion adjustment space and represent the differences in vector form. Each vector records the change in parameters such as screw speed, extrusion temperature, and cooling rate. For example, a certain vector may represent an increase of 5 revolutions per minute in screw speed, a decrease of 3 degrees Celsius in extrusion temperature, and an increase of 0.5 meters per minute in cooling rate.

[0067] By mutating and expanding multiple extrusion differential vectors, multiple sets of extrusion adjustment vectors are obtained. That is, by making appropriate random perturbations or adjustments to the differential vectors, more possible combinations of extrusion parameters are generated to increase the diversity of schemes and the search space. For example, after expanding the original differential vectors by 5 times, 50 different extrusion adjustment vectors can be obtained, each vector representing a possible parameter fine-tuning scheme.

[0068] The extrusion adjustment space is cross-mutated based on multiple sets of extrusion adjustment vectors to obtain the first extrusion adjustment variation domain. This involves combining or mixing different vectors in the adjustment vector set. For example, the screw speed is used with the value of one vector and the extrusion temperature is used with the value of another vector, thereby generating a new parameter combination region and forming a variation domain containing multiple new extrusion schemes in order to explore better extrusion schemes. For example, cross-mutation can generate 20 new combinations, covering parameter ranges that have not been explored before.

[0069] Based on a multidimensional defect risk prediction model, constraint optimization is performed on the first extrusion adjustment variation domain according to multidimensional defect risk constraints to generate the first extrusion evolution space. Specifically, the multidimensional defect risk prediction model evaluates all schemes in the variation domain and selects the parameter combination with the lowest risk based on constraints related to surface defects, internal defects, and electrical performance defects, forming the first extrusion evolution space. For example, if there are 50 schemes in the variation domain, constraint optimization may ultimately select 15 schemes with the lowest risk as the first evolution space. The multidimensional defect risk constraints include surface defect risk thresholds, internal defect risk thresholds, and electrical performance defect risk thresholds. By comparing these three aspects, if the risk is less than all three thresholds, it is considered to satisfy the multidimensional defect risk constraints and is then added to the first extrusion evolution space.

[0070] Furthermore, this application also includes: obtaining an extrusion monitoring data stream; cleaning the extrusion monitoring data stream to obtain the extrusion monitoring dataset.

[0071] Specifically, during the extrusion production of cable insulation, sensors and detection equipment deployed on the production line are used to collect continuously changing data such as temperature, pressure, extrusion speed, current, voltage, and material flow rate in real time, thereby obtaining an extrusion monitoring data stream.

[0072] The collected extrusion monitoring data stream is organized and processed, including removing invalid data, correcting abnormal data, filling in missing data, and standardizing the timestamp format to make the data more accurate and standardized, resulting in an extrusion monitoring dataset.

[0073] Furthermore, this application also includes: obtaining an insulation layer defect alarm based on the second insulation layer defect map.

[0074] Specifically, based on the future defect prediction information contained in the second insulation layer defect map, such as the potential increase in the number of surface bubbles, the potential length of internal cracks, or the potential range of abnormal electrical performance, an insulation layer defect alarm is generated to provide a basis for timely action. When the number, distribution, or severity of defects exceeds a preset threshold, an alarm message is automatically generated to remind production personnel or the control system to intervene or make adjustments. For example, when the predicted number of surface bubbles exceeds 50, or the predicted length of abnormal electrical performance exceeds 2 meters, a defect alarm is triggered, and adjustments to temperature or cooling rate may be suggested to reduce the risk.

[0075] In summary, the online defect detection method for cable insulation provided in this application has the following technical effects: by achieving the technical goal of online defect detection and intelligent optimization adjustment of the cable insulation production process, it can detect and correct surface, internal and electrical performance defects in real time during the production process, thereby improving cable product quality, reducing defect rate, increasing production efficiency and ensuring reliability.

[0076] Example 2: Based on the same inventive concept as the online defect detection method for cable insulation layers in the foregoing examples, this application also provides an online defect detection system for cable insulation layers. Please refer to the appendix. Figure 2 The system includes: an extrusion monitoring dataset acquisition module 1, used for cable production based on the insulation extrusion scheme and simultaneously acquiring the extrusion monitoring dataset; an insulation layer defect first map construction module 2, used for online detection of insulation layer defects based on the extrusion monitoring dataset and constructing an insulation layer defect first map; an insulation layer defect second map acquisition module 3, used for trend prediction based on the insulation layer defect first map and obtaining an insulation layer defect second map; an extrusion adjustment space acquisition module 4, used for feedback adjustment of the insulation extrusion scheme based on the insulation layer defect second map and obtaining an extrusion adjustment space; an extrusion candidate population acquisition module 5, used for introducing a multidimensional defect risk prediction model to optimize the defect risk of the extrusion adjustment space and obtaining an extrusion candidate population; and an extrusion adjustment optimization result acquisition module 6, used for differential evolution optimization of the extrusion adjustment space based on the multidimensional defect risk prediction model and the extrusion candidate population to obtain an extrusion adjustment optimization result, and performing online optimization of cable production based on the extrusion adjustment optimization result.

[0077] Furthermore, the online defect detection system for cable insulation layers is also used for: detecting surface defects in the insulation layer based on the extrusion monitoring dataset to obtain surface defect detection results; detecting internal defects in the insulation layer based on the extrusion monitoring dataset to obtain internal defect detection results; detecting electrical performance defects in the insulation layer based on the extrusion monitoring dataset to obtain electrical performance defect detection results; and organizing the surface defect detection results, the internal defect detection results, and the electrical performance defect detection results to obtain a first defect map of the insulation layer.

[0078] Furthermore, the online defect detection system for cable insulation layers is also used for: capturing surface defect association features of the insulation layer based on the extrusion monitoring dataset to obtain a surface defect association feature set; performing supervised training based on the surface defect detection record set of the insulation layer to obtain a surface defect detection model; injecting perturbation into the surface defect detection record set of the insulation layer using an adversarial example generator to obtain a surface defect detection adversarial example set; performing robustness enhancement training on the surface defect detection model based on the surface defect detection adversarial example set to generate a surface defect detection channel; and inputting the surface defect association feature set into the surface defect detection channel to obtain the surface defect detection result.

[0079] Furthermore, the online defect detection system for cable insulation layers is also used for: activating the multidimensional defect risk prediction model, which includes a surface defect risk prediction model, an internal defect risk prediction model, and an electrical performance defect risk prediction model; optimizing the surface defect risk of the extrusion adjustment space according to the surface defect risk prediction model to obtain a first extrusion candidate seed; optimizing the internal defect risk of the extrusion adjustment space according to the internal defect risk prediction model to obtain a second extrusion candidate seed; optimizing the electrical performance defect risk of the extrusion adjustment space according to the electrical performance defect risk prediction model to obtain a third extrusion candidate seed; and adding the first extrusion candidate seed, the second extrusion candidate seed, and the third extrusion candidate seed to the extrusion candidate population.

[0080] Furthermore, the online defect detection system for cable insulation is also used for: performing simulated production based on each extrusion adjustment scheme within the extrusion adjustment space to obtain extrusion simulation data for each scheme; performing risk analysis on the extrusion simulation data for each scheme based on the surface defect risk prediction model to construct a surface defect risk sequence; and performing iterative optimization of the surface defect risk in the extrusion adjustment space based on the surface defect risk sequence to obtain the first seed of the extrusion candidate.

[0081] Furthermore, the online defect detection system for cable insulation layers is also used for: based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the first extrusion candidate seed to obtain an extrusion evolution first space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the second extrusion candidate seed to obtain an extrusion evolution second space; based on the multidimensional defect risk prediction model, performing differential evolution on the extrusion adjustment space according to the third extrusion candidate seed to obtain an extrusion evolution third space; configuring weights for the multidimensional defect risk indicators based on the multidimensional defect risk prediction model to establish a global defect risk analysis model; and performing joint optimization to minimize global defect risk on the first extrusion evolution space, the second extrusion evolution space, and the third extrusion evolution space according to the global defect risk analysis model to obtain the extrusion adjustment optimization result.

[0082] Furthermore, the online defect detection system for cable insulation layers is also used for: performing difference detection on the extrusion adjustment space based on the extrusion candidate first seed to obtain multiple extrusion difference vectors; performing mutation expansion on the multiple extrusion difference vectors to obtain multiple extrusion adjustment vector sets; performing cross mutation on the extrusion adjustment space based on the multiple extrusion adjustment vector sets to obtain a first extrusion adjustment mutation domain; and performing constraint optimization on the first extrusion adjustment mutation domain based on the multidimensional defect risk prediction model and multidimensional defect risk constraints to generate the first extrusion evolution space.

[0083] Furthermore, the online defect detection system for cable insulation is also used to: obtain an extrusion monitoring data stream; clean the extrusion monitoring data stream to obtain the extrusion monitoring dataset.

[0084] Furthermore, the online defect detection system for cable insulation is also used to: obtain an insulation defect alarm based on the second defect map of the insulation layer.

[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The online detection method and specific examples of cable insulation layer defects in the aforementioned embodiment one are also applicable to the online detection system of cable insulation layer defects in this embodiment. Through the foregoing detailed description of the online detection method of cable insulation layer defects, those skilled in the art can clearly understand the online detection system of cable insulation layer defects in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for online detection of defects in cable insulation, characterized in that, The method includes: Cable production is based on an insulation extrusion scheme, and extrusion monitoring datasets are acquired simultaneously. Based on the extrusion monitoring dataset, online detection of insulation layer defects is performed to construct a first atlas of insulation layer defects. Based on the first spectrum of insulation layer defects, a trend prediction is performed to obtain a second spectrum of insulation layer defects. The insulation extrusion scheme is adjusted based on the second defect map of the insulation layer to obtain the extrusion adjustment space. A multidimensional defect risk prediction model is introduced to optimize the defect risk in the extrusion adjustment space and obtain the extrusion candidate population. Based on the multidimensional defect risk prediction model, differential evolution optimization is performed on the extrusion adjustment space according to the extrusion candidate population to obtain the extrusion adjustment optimization result, and online optimization of cable production is performed according to the extrusion adjustment optimization result.

2. The online defect detection method for cable insulation as described in claim 1, characterized in that, Based on the extrusion monitoring dataset, online detection of insulation layer defects is performed, and a first atlas of insulation layer defects is constructed, including: The surface defects of the insulation layer are detected based on the extrusion monitoring dataset to obtain the surface defect detection results; Based on the extrusion monitoring dataset, internal defects in the insulation layer are detected to obtain the internal defect detection results; Based on the extrusion monitoring dataset, electrical performance defects of the insulation layer are detected to obtain electrical performance defect detection results. By compiling the surface defect detection results, the internal defect detection results, and the electrical performance defect detection results, a first spectrum of insulation layer defects is obtained.

3. The online defect detection method for cable insulation as described in claim 2, characterized in that, Based on the extrusion monitoring dataset, surface defects in the insulation layer are detected to obtain surface defect detection results, including: Based on the extrusion monitoring dataset, surface defect correlation features of the insulation layer are captured to obtain a surface defect correlation feature set; Supervised training is performed based on the insulation layer surface defect detection record set to obtain a surface defect detection model; The surface defect detection adversarial sample set is obtained by perturbating the insulation layer surface defect detection record set using an adversarial sample generator. The surface defect detection model is robustly enhanced by training based on the adversarial example set for surface defect detection, thereby generating a surface defect detection channel; The surface defect associated feature set is input into the surface defect detection channel to obtain the surface defect detection result.

4. The online defect detection method for cable insulation as described in claim 1, characterized in that, A multidimensional defect risk prediction model is introduced to optimize the defect risk in the extrusion adjustment space, obtaining an extrusion candidate population, including: Activate the multidimensional defect risk prediction model, which includes a surface defect risk prediction model, an internal defect risk prediction model, and an electrical performance defect risk prediction model. Based on the surface defect risk prediction model, the surface defect risk of the extrusion adjustment space is optimized to obtain the first seed of extrusion candidates. Based on the internal defect risk prediction model, the internal defect risk of the extrusion adjustment space is optimized to obtain the second seed of extrusion candidates. Based on the electrical performance defect risk prediction model, the electrical performance defect risk of the extrusion adjustment space is optimized to obtain the third seed candidate for extrusion. The first extrusion candidate seed, the second extrusion candidate seed, and the third extrusion candidate seed are added to the extrusion candidate population.

5. The online defect detection method for cable insulation as described in claim 4, characterized in that, The surface defect risk is optimized in the extrusion adjustment space according to the surface defect risk prediction model to obtain the first seed of extrusion candidates, including: Simulation production is carried out based on each extrusion adjustment scheme within the extrusion adjustment space to obtain extrusion simulation data for each scheme. Based on the surface defect risk prediction model, risk analysis is performed on the extrusion simulation data of each scheme to construct a surface defect risk sequence. Based on the surface defect risk sequence, the extrusion adjustment space is iteratively optimized to obtain the first seed of the extrusion candidate.

6. The online defect detection method for cable insulation as described in claim 1, characterized in that, Based on the multidimensional defect risk prediction model, differential evolution optimization is performed on the extrusion regulation space according to the extrusion candidate population to obtain the extrusion regulation optimization results, including: Based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the first extrusion candidate seed to obtain the first extrusion evolution space. Based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the extrusion candidate second seed to obtain the extrusion evolution second space; Based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the third extrusion candidate seed to obtain the third extrusion evolution space. Based on the multidimensional defect risk prediction model, the weights of the multidimensional defect risk indicators are configured to establish a global defect risk analysis model. Based on the global defect risk analysis model, the global defect risk is jointly minimized in the first space, the second space, and the third space of extrusion evolution to obtain the extrusion adjustment optimization result.

7. The online defect detection method for cable insulation as described in claim 6, characterized in that, Based on the multidimensional defect risk prediction model, differential evolution is performed on the extrusion adjustment space according to the first extrusion candidate seed to obtain the first extrusion evolution space, including: Based on the first extrusion candidate seed, the extrusion adjustment space is subjected to difference detection to obtain multiple extrusion difference vectors; Based on the multiple extrusion differential vectors, a set of multiple extrusion adjustment vectors is obtained by mutation expansion. The extrusion adjustment space is cross-mutated according to the multiple sets of extrusion adjustment vectors to obtain the first extrusion adjustment variation domain; Based on the multidimensional defect risk prediction model, the first extrusion adjustment variation domain is constrained and optimized according to the multidimensional defect risk constraints to generate the first space of extrusion evolution.

8. The online defect detection method for cable insulation as described in claim 1, characterized in that, Cable production is based on an insulation extrusion scheme, and extrusion monitoring datasets are acquired simultaneously, including: Obtain extrusion monitoring data stream; The extrusion monitoring data stream is cleaned to obtain the extrusion monitoring dataset.

9. The online defect detection method for cable insulation as described in claim 1, characterized in that, An insulation layer defect alarm is obtained based on the second spectrum of insulation layer defects.

10. An online defect detection system for cable insulation, characterized in that, The steps for implementing the online defect detection method for a cable insulation layer according to any one of claims 1 to 9 include: The extrusion monitoring dataset acquisition module is used to acquire extrusion monitoring datasets simultaneously during cable production based on insulation extrusion schemes. The insulation layer defect first map construction module is used to perform online detection of insulation layer defects based on the extrusion monitoring dataset and construct the insulation layer defect first map. The second spectrum of insulation layer defects acquisition module is used to perform trend prediction based on the first spectrum of insulation layer defects to obtain the second spectrum of insulation layer defects. The extrusion adjustment space acquisition module is used to adjust the insulation extrusion scheme based on the second pattern of insulation layer defects to obtain the extrusion adjustment space; The extrusion candidate population acquisition module is used to introduce a multi-dimensional defect risk prediction model to optimize the defect risk of the extrusion adjustment space and obtain the extrusion candidate population. The extrusion adjustment optimization result acquisition module is used to perform differential evolution optimization on the extrusion adjustment space based on the multidimensional defect risk prediction model and the extrusion candidate population to obtain the extrusion adjustment optimization result, and to perform online optimization of cable production based on the extrusion adjustment optimization result.

Images