CN-121980934-A - Modeling analysis method for ablation characteristics of cable buffer layer

CN121980934ACN 121980934 ACN121980934 ACN 121980934ACN-121980934-A

Abstract

The invention discloses a modeling analysis method for ablation characteristics of a cable buffer layer, and particularly relates to the field of data analysis. The method comprises the steps of carrying out image detection and depth measurement on an ablation area on the surface of a buffer layer, obtaining temperature distribution and time sequence information under different working conditions, establishing a database, carrying out normalization, abnormal elimination, deletion completion and correlation analysis on the data, completing dimension reduction and feature selection to form a feature set, training a data driving model by combining an ablation mechanism and a dynamic equation, introducing physical constraint fusion to obtain a prediction model, quantifying the degree of ablation, grading risk, estimating the residual life and predicting short, medium and long term trends, and graphically displaying by using spatial distribution and evolution processes to provide basis for operation and maintenance decision.

Inventors

WANG HAORAN
YU ZHONG
BAI TINGWEI
XU JIAXIN
LI JIANFENG
YANG ZHENWEI
GAO MENG
WANG DA
LI YANLING
SUN QIANLIN

Assignees

辽宁电能发展股份有限公司
康威通信技术股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260121

Claims (10)

1. A modeling analysis method for cable buffer ablation features, comprising: Step S1, acquiring ablation data of a cable buffer layer, namely acquiring ablation original data of the cable buffer layer, acquiring an ablation image, recording the spatial distribution of an ablation area, measuring the ablation depth, acquiring temperature distribution data under different working states, recording ablation time sequence evolution data according to preset time intervals, and establishing a cable buffer layer ablation data acquisition database; Step S2, extracting ablation characteristic parameters, namely extracting geometrical form parameters of an ablation area from the database, performing texture analysis on an ablation image, analyzing color change characteristics of the ablation area, extracting temperature field characteristics from temperature distribution data, calculating evolution characteristics of the ablation parameters along with time, and forming a cable buffer layer ablation multidimensional characteristic parameter set; step S3, ablation characteristic data preprocessing, namely carrying out normalization processing on the multidimensional characteristic parameter set, removing abnormal values, carrying out missing value interpolation processing, analyzing characteristic correlation and identifying redundant characteristics, carrying out characteristic dimension reduction and selection, and forming a preprocessed cable buffer layer ablation characteristic data set; S4, constructing an ablation evolution model, namely performing ablation physical mechanism analysis to establish a theoretical basis model, establishing an ablation evolution kinetic equation, training a data driving model by utilizing the preprocessed cable buffer layer ablation characteristic data set, merging physical mechanism constraint into the data driving model, constructing a physical information neural network, and performing parameter optimization and cross validation on the model to obtain a cable buffer layer ablation evolution prediction model; Step5, quantitatively evaluating the ablation degree, namely establishing an ablation degree multistage classification standard, calculating an ablation comprehensive evaluation index, evaluating an ablation risk level, evaluating the residual service life of the buffer layer and generating an ablation degree quantitative evaluation report; step 6, ablation trend prediction analysis, namely short-term ablation prediction and long-term ablation prediction are carried out based on the cable buffer layer ablation evolution prediction model, scene simulation under different working conditions is carried out, prediction uncertainty is quantified, and a prediction result is compared with actual detection data for verification to form an ablation trend prediction analysis result; And S7, visually outputting analysis results, namely generating an ablation three-dimensional space distribution diagram and a two-dimensional expansion diagram, drawing an ablation parameter time sequence evolution curve and animation, generating a statistical chart, displaying comparison results of a predicted value, an actual value and different scenes, and constructing an interactive visual analysis interface to form an ablation analysis visual reporting system.
2. The method of claim 1, wherein in the step S1, the cable buffer layer ablation detection is carried out by adopting an industrial endoscope and a high-resolution CCD camera in combination, the resolution is set to be 0.01mm, the acquisition frequency is 10 frames per second, the endoscope probe scans along the axial direction of the cable at intervals of 5mm, 360-degree circumferential rotation acquisition is carried out at each scanning position, the ablation depth measurement is carried out by adopting a laser displacement sensor with the wavelength of 655nm, the measurement precision is +/-2 μm, the measurement range is 0-10mm, gridding scanning is carried out at the step interval of 0.1mm, the temperature field acquisition is carried out by adopting a thermal infrared imager, the temperature resolution is 0.1 ℃, the temperature measurement range is-20 ℃ to 400 ℃, the temperature is continuously acquired for 30 minutes under the rated load condition, the overload condition and the idle condition, the sampling interval is 10 seconds, the time sequence data record is comprehensively detected once every 7 days, the continuous monitoring period is 6 months, and the unique identification code is allocated to each ablation area, the depth and the volume evolution data are tracked.
3. The method according to claim 1, wherein the ablation physical mechanism analysis in the step S4 comprises the steps of performing an ablation process on a cable buffer layer material which is polypropylene or polyethylene polymer material, wherein the ablation process comprises three stages of thermal oxidative degradation, pyrolysis carbonization and mechanical exfoliation, and establishing an ablation influence factor system which comprises a temperature factor, a current factor, an environment factor and a material factor.
4. The method of claim 1, wherein the physical constraint fusion in the step S4 comprises introducing an ablation dynamics equation as a constraint into a neural network training, constructing a physical information neural network, wherein the output of the physical information neural network is an ablation depth predicted value and an ablation area predicted value, adding physical characteristics of temperature, current and oxygen concentration into the input of the neural network, and performing pre-training and physical constraint fine-tuning by adopting a transfer learning strategy.
5. The method according to claim 1, wherein the model parameter optimization and verification in the step S4 includes performing super-parameter search by using bayesian optimization and establishing a proxy model of super-parameter and model performance by using a gaussian process, selecting super-parameter by using expected improvement as an acquisition function, evaluating model generalization performance by using k-fold cross verification, and selecting a model configuration with optimal performance as a cable buffer ablation evolution prediction model by using root mean square error, average absolute error and decision coefficient as performance indexes.
6. The method according to claim 1, wherein the ablation levels in the step S5 are divided into five levels, namely a normal level, a slight ablation level, a moderate ablation level, a severe ablation level and a severe ablation level, wherein the normal level corresponds to an ablation depth d <0.1mm and an ablation area ratio A% <1%, the slight ablation level corresponds to an ablation depth 0.1mm < d <0.5mm and an ablation area ratio 1% < A% <5%, the moderate ablation level corresponds to an ablation depth 0.5mm < d <1.0mm and an ablation area ratio 5% < A% <15%, the severe ablation level corresponds to an ablation depth 1.0mm < d <1.5mm and an ablation area ratio 15% < A% <30%, the severe ablation level corresponds to an ablation depth d > 1.5mm and an ablation area ratio A% > 30%, and when the ablation depth increase rate is greater than 0.1 mm/month or the ablation area increase rate is greater than 5%/month, the ablation level is up-regulated and the classification result of the level is output.
7. The method according to claim 1, wherein the ablation risk assessment in the step S5 comprises the steps of establishing an ablation risk score model, wherein the ablation risk score considers the current state, the development trend and the influence result of ablation, the severity of the failure result is determined according to the importance level, the load type and the replacement difficulty of the cable, five grades of scores of 1-5 are adopted, the ablation risk score RS is in a value range of [0,5], and risk grades are classified according to a risk matrix, wherein RS <1 is low risk, RS <1 is less than or equal to 2 is low risk, RS <2 is less than or equal to 3 is medium risk, RS <3 is less than or equal to 4 is medium and high risk, and RS is more than or equal to 4 is high risk.
8. The method according to claim 1, wherein the residual life estimation in the step S5 comprises the steps of setting a failure threshold value to be 80% of the ablation depth reaching the thickness of the buffer layer or the ablation area ratio reaching 40%, predicting an ablation parameter for 24 months in the future by using a cable buffer layer ablation evolution prediction model, predicting a step length to be 1 month, determining the moment when the failure threshold value is reached for the first time to obtain the residual life, adding random disturbance to the model input parameter and performing 1000 Monte Carlo simulation to obtain a probability distribution of the residual life, marking the residual life as emergency replacement for less than 6 months, marking the residual life as planned replacement for 6 to 12 months, and marking the residual life as continuous monitoring for more than 12 months.
9. The method according to claim 1, wherein the step S6 comprises the steps of predicting ablation development for 1 to 3 months in the future, predicting ablation evolution trend for 6 to 24 months in the future by short-term ablation prediction, simulating different working conditions, wherein the normal working conditions set the cable operation current to 80 to 100% of rated current, the overload working conditions set the cable operation current to 110 to 120% of rated current in summer in high-temperature period and the duration to 3 months, and comparing the prediction result with actual detection data to verify the prediction accuracy.
10. The method of claim 1, wherein the interactive visual interface in the step S7 is constructed by using a Web technology stack, the front end is visually displayed by using HTML5, CSS3 and JavaScript and using d3.Js, ECharts and three.js, the back end is provided with a data interface by using Python Flask framework, the interface is laid out by using a dashboard and provides a space distribution view, a time-sequence evolution view, a statistical analysis view, a prediction comparison view and a report deriving function, supports graph scaling, translation, rotation and data point hovering prompt, supports deriving PNG, SVG or PDF format graphs and deriving Excel or CSV format analysis data, supports generating PDF or HTML format analysis report, and realizes multi-cable comparison and early warning reminding.

Description

Modeling analysis method for ablation characteristics of cable buffer layer Technical Field The invention relates to the technical field of data analysis, in particular to a modeling analysis method aiming at ablation characteristics of a cable buffer layer. Background In the prior art, in order to analyze the ablation condition of a cable buffer layer, a relatively similar method is generally developed according to the process of detection acquisition, characteristic acquisition and result output, wherein the surface of the buffer layer is firstly detected, an ablation related image is acquired, a suspected ablation area is positioned, then the depth measurement is carried out on a target area, the operation environment information such as temperature distribution and the like is acquired under different operation working conditions, and then a data set is formed by filing the detection results for a plurality of times according to time for manual interpretation or primary evaluation according to an experience threshold value. However, the method has the defects in practical application that firstly, a single detection result or a small amount of geometric quantity is mainly used, the ablation form, texture/color change, temperature field characteristics and time sequence evolution are difficult to simultaneously describe, so that the characteristic expression is insufficient, secondly, the modeling analysis mainly adopts empirical rules or pure data driving fitting, constraint and systematic data preprocessing aiming at an ablation mechanism are lacked, characteristic screening is not carried out, model stability and interpretation are not enough, thirdly, evaluation output always stays at an ablation/light and heavy or simple report level, comprehensive quantification, risk prompt, service life estimation and visual presentation facing to operation and maintenance are lacked, and a decision closed loop is difficult to form. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a modeling analysis method for ablation characteristics of a cable buffer layer, which solves the problems in the background art through the following scheme. In order to achieve the purpose, the invention provides the following technical scheme that the modeling analysis method for the ablation characteristics of the cable buffer layer comprises the following steps: Step S1, acquiring ablation data of a cable buffer layer, namely acquiring ablation original data of the cable buffer layer, acquiring an ablation image, recording the spatial distribution of an ablation area, measuring the ablation depth, acquiring temperature distribution data under different working states, recording ablation time sequence evolution data according to preset time intervals, and establishing a cable buffer layer ablation data acquisition database; Step S2, extracting ablation characteristic parameters, namely extracting geometrical form parameters of an ablation area from the database, performing texture analysis on an ablation image, analyzing color change characteristics of the ablation area, extracting temperature field characteristics from temperature distribution data, calculating evolution characteristics of the ablation parameters along with time, and forming a cable buffer layer ablation multidimensional characteristic parameter set; step S3, ablation characteristic data preprocessing, namely carrying out normalization processing on the multidimensional characteristic parameter set, removing abnormal values, carrying out missing value interpolation processing, analyzing characteristic correlation and identifying redundant characteristics, carrying out characteristic dimension reduction and selection, and forming a preprocessed cable buffer layer ablation characteristic data set; S4, constructing an ablation evolution model, namely performing ablation physical mechanism analysis to establish a theoretical basis model, establishing an ablation evolution kinetic equation, training a data driving model by utilizing the preprocessed cable buffer layer ablation characteristic data set, merging physical mechanism constraint into the data driving model, constructing a physical information neural network, and performing parameter optimization and cross validation on the model to obtain a cable buffer layer ablation evolution prediction model; Step5, quantitatively evaluating the ablation degree, namely establishing an ablation degree multistage classification standard, calculating an ablation comprehensive evaluation index, evaluating an ablation risk level, evaluating the residual service life of the buffer layer and generating an ablation degree quantitative evaluation report; step 6, ablation trend prediction analysis, namely short-term ablation prediction and long-term ablation prediction are carried out based on the cable buffer layer ablation evolution prediction model, scene simulation under different