Cable state and life cycle evaluation method based on thermal and vibration information fusion

Abstract

Based on thermal and vibration information fusion, the cable state and life cycle evaluation method is constructed based on optical fiber sensing technology temperature and vibration data acquisition model, and the influence of non-electric parameters on the operation state of the cable is analyzed. Through the collection of temperature, vibration and other signals in the process of cable operation, combined with data mining and regression modeling technology, the health status and life cycle of the cable are evaluated. The method further uses multi-physical field simulation to predict the state of the cable under different working conditions, and improves the accuracy and response speed of fault prediction through model optimization, so as to realize the early warning of cable fault. This technology has important application value for improving the operation and maintenance efficiency of power system and reducing the operation and maintenance cost.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and more particularly to power equipment condition monitoring technology, specifically to a cable condition assessment method based on the fusion of temperature and vibration information. Background Technology

[0002] Cables are a crucial component of urban power distribution and transmission systems. Living in complex environments such as underground utility tunnels, underground galleries, or direct burial, they are susceptible to problems like insulation aging, localized overheating, and mechanical damage due to factors such as current-carrying heat, ambient temperature variations, and external disturbances. These potential hazards often exhibit characteristics of "weak early signs followed by sudden failures." Without continuous, online condition monitoring and assessment methods, power outages and increased maintenance costs are highly likely.

[0003] Existing cable condition monitoring methods mainly include regular power outage maintenance, manual inspection, partial discharge detection, and point-type temperature / vibration sensor monitoring. However, in scenarios with large-scale cable networks and complex laying environments, point-type sensors have limited coverage, high deployment and maintenance costs, and difficulty in achieving continuous monitoring along the line; offline maintenance and inspection suffer from long cycles and delayed detection, making it difficult to meet the need for early warning of potential hazards.

[0004] With the development of distributed optical fiber sensing technology, it has become possible to achieve long-distance, continuous monitoring of multiple parameters such as temperature, strain, and vibration using cable composite optical fiber or optical fiber laid along the cable. For example, publication number CN112697180A discloses a fusion-type distributed optical fiber sensing system and method for simultaneous measurement of temperature and vibration, which can simultaneously demodulate temperature and vibration signals in the same system, thereby obtaining dynamic and static information along the cable.

[0005] On the other hand, CN109870627A discloses a method for fault alarm and diagnosis of submarine cables based on distributed optical fiber temperature, strain, and vibration monitoring data. This method achieves fault alarm and location by performing moving average noise reduction on the monitoring data, difference normalization with reference data, and setting temperature / strain / vibration thresholds respectively. Furthermore, it combines statistical patterns of abnormal areas to determine the fault type. This type of solution emphasizes the use of multi-physical quantity data for fault alarm and diagnosis, which can reduce false alarms and missed alarms to a certain extent and provide maintenance information.

[0006] Therefore, developing cable condition monitoring technologies based on non-electrical signals such as temperature and vibration has become an important direction for improving the reliability and safety of power systems.

[0007] Existing technologies have several problems in assessing cable condition and lifespan through heat and vibration: 1. Multi-source information fusion mostly stays at the "alarm / diagnosis" level: Many solutions focus on setting thresholds for single or a few parameters such as temperature, strain or vibration to issue abnormal alarms. It is difficult to form a quantifiable and comparable health status index. Moreover, the thresholds are sensitive to changes in operating conditions and noise, which can easily lead to false alarms or missed alarms. 2. Lack of degradation modeling for the entire life cycle: Temperature-induced thermal aging and vibration-induced fatigue damage have cumulative and nonlinear coupling characteristics. Existing methods mostly lack a unified model that maps the combined effects of heat and vibration to the aging rate and remaining life, making it difficult to support closed-loop management of "condition assessment-life prediction-maintenance decision". 3. Insufficient data processing and model generalization capabilities: Distributed sensing and multi-sensor fusion will generate massive amounts of data. If there is a lack of effective feature extraction, noise reduction and online update mechanisms, the robustness and transferability of the model under different laying environments, load fluctuations and seasonal changes are difficult to guarantee. 4. Insufficient coupling between physical mechanisms and data-driven methods: Relying solely on empirical rules or a single mechanistic model makes it difficult to balance interpretability and prediction accuracy; it is necessary to combine mechanistic models, such as thermal aging laws, with data-driven methods such as machine learning to improve the ability to identify early-stage problems and the accuracy of lifespan prediction.

[0008] Therefore, the applicant proposes a cable condition and life cycle assessment method based on the fusion of thermal and vibration information. Summary of the Invention

[0009] The purpose of this invention is to address the technical shortcomings of existing cable condition and life cycle assessment technologies, such as false alarms and false negatives caused by single threshold alarms, lack of degradation modeling for the entire life cycle, weak data processing generalization ability, and insufficient coupling between physical mechanisms and data-driven methods, which lead to poor accuracy and effectiveness in cable condition and life cycle assessment.

[0010] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: The cable condition and life cycle assessment method based on the fusion of thermal and vibration information includes the following steps: Step 1: Perform multi-physical quantity data acquisition, processing, and manipulation; Step 2: Construct a cable health status assessment model; Step 3: Obtain the cable lifecycle assessment model; Step 4: Jointly optimize the cable health status assessment model and the cable life cycle assessment model; Step 5: Optimize real-time monitoring and self-correction of cable management, as well as intelligent decision-making throughout the cable lifecycle; Step 6: Conduct simulation calculations in different scenarios In step 1: The temperature field of the cable is obtained by real-time acquisition of multiple physical quantities of the cable through distributed fiber optic sensors and microelectromechanical systems (MEMS) sensors. Vibration signals and environmental data .

[0011] The cable status dataset is as follows: (1); in, For time, This indicates the cable location.

[0012] (2) Signal preprocessing: The acquired temperature and vibration signals are subjected to noise reduction, smoothing, and standardization. Temperature signal The vibration signal is smoothed using a Kalman filter. Frequency band decomposition is performed using wavelet packet transform.

[0013] The Kalman filter formula is: (2); The decomposition of the vibration signal using wavelet packets is as follows: (3); in ， These are the wavelet decomposition results at different scales. For node indexing; The number of components involved in the reconstruction.

[0014] (3) Feature extraction and fusion: Using principal component analysis to study temperature ,vibration and environmental data Dimensionality reduction is performed to obtain a comprehensive feature set of the cable's operating status: (4); in, These are the characteristic vectors of temperature, vibration, and environmental factors, respectively.

[0015] In step 2: (1) Health status prediction: Based on the extracted feature set A cable health status assessment model is established using Support Vector Machine (SVM), Random Forest (RF), or Deep Neural Network (DNN), and the model outputs a cable health status index. .

[0016] The health status prediction model is represented as follows: (5); in, It is a trained machine learning model. For model parameters, This is a cable health status index.

[0017] (2) Health Index Calculation: Cable health status index The Sigmoid function is used for calculation, and the formula is as follows: (6); in, The regression coefficients control for the contribution of different characteristics to health assessment.

[0018] (3) Cable condition classification: According to the health status index The cable condition is divided into multiple levels, each corresponding to a different fault risk: (7); If the health status index is within (0–0.2], it indicates a serious failure risk; if the health status index is within (0.2–0.5], it indicates a multi-quantile risk; if the health status index is within (0.5–0.8], it indicates a slight quantile risk; if the health status index is within (0.8–1], it indicates a normal operating condition.

[0019] The grading system helps to monitor the health status of cables in real time and provide early warnings.

[0020] (4) Thermal aging rate: Aging rate of cable insulation materials The modified Arrenius model indicates that: (8); in, As the pre-factor, Thermal activation energy, Boltzmann's constant, This represents the operating temperature of the cable. The formula illustrates the effect of temperature on the cable's aging rate.

[0021] (5) Vibration aging rate: Vibration has a significant impact on cable fatigue damage, and the vibration aging rate... It can be calculated using the following formula: (9); in, The power spectral density of the vibration signal. Let be the frequency sensitivity function of the vibration. The cutoff frequency is the vibration frequency.

[0022] (6) Overall aging rate: The aging rate of a cable, which combines the effects of temperature and vibration, is expressed as: (10); This formula reflects the overall degradation rate of cables under different operating conditions.

[0023] (7) Remaining useful life prediction: By comprehensive aging rate The following integral formula is used to predict the remaining useful life (RUL) of the cable: (11); in, Indicates the remaining service life. For the current moment, For the target time, The overall aging rate varies over time.

[0024] In step 3: (1) Multi-task learning MTL optimization: A multi-task learning MTL model is used to jointly optimize the cable health status assessment model and the life cycle assessment model. The loss function used is: (12); in, For health status assessment loss function, For lifecycle prediction loss function, The coefficient is used to balance the weights of the two tasks.

[0025] (2) Dynamic updates and learning: Use incremental learning algorithms to refine model parameters. Dynamic updates are performed to allow the model to adapt to changes in the cable's operating condition. The update formula is: (13); in, For learning rate, The gradient of the loss function. These are the old parameter values ​​for the model. These are the updated parameter values ​​for the model.

[0026] In step 4: (1) Real-time monitoring and self-correction function: Based on sensor output errors and environmental changes, the system provides self-diagnosis and self-correction functions to ensure the accuracy of cable condition assessment. Through online learning methods, the model can gradually update and adjust its parameters as new health status data is input. When dynamically updating the cable condition model, real-time feedback and self-calibration further optimize the results.

[0027] (2) Intelligent decision-making throughout the entire cable lifecycle: By combining real-time monitoring data with historical data, intelligent decision support is provided for cable maintenance, repair, and replacement, thereby optimizing the operation and maintenance of the power system.

[0028] In step 5: Different scenarios include: 1) Scenario I: Considering only the anomaly detection and health status assessment of temperature information, the health status index and classification results are output through Kalman filtering of temperature signals, PCA feature dimensionality reduction, and a type of SVM classification health assessment model, serving as a baseline comparison for the multi-source fusion scheme.

[0029] 2) Scenario II: Considering the combined effects of temperature and vibration information, a comprehensive assessment of cable health status and life cycle is conducted. This involves temperature Kalman filtering, vibration wavelet packet decomposition, PCA feature fusion, and an RF health assessment model. The thermal aging rate is calculated based on the modified Arrenius model, and the vibration aging rate is calculated by combining the vibration power spectral density. The resulting comprehensive aging rate is used to predict the remaining service life. The cable health status index, graded early warning results, and remaining service life prediction values ​​are output simultaneously. The accuracy, false alarm / missed alarm rate, early warning lead time, and life prediction error are comprehensively compared with Scenario I to verify the comprehensive advantages of multi-source information fusion in improving status assessment and achieving full life cycle management.

[0030] This process, through data collection, processing, feature extraction, model building, and optimization, comprehensively improves the accuracy and efficiency of cable condition assessment and lifecycle prediction, thereby optimizing the operation and maintenance management of the power system.

[0031] Compared with the prior art, the present invention has the following technical effects: 1) This invention uses distributed fiber optic sensors and MEMS sensors to achieve real-time acquisition of cable temperature field, vibration and environmental quantities, making the state characterization more comprehensive and closer to the real working conditions. Kalman filtering and wavelet packet decomposition are used to perform noise reduction processing on temperature / vibration signals respectively, and then PCA dimensionality reduction and fusion are used to form a high-quality comprehensive feature set, thereby improving the stability and accuracy of health assessment. 2) Furthermore, this invention utilizes models such as SVM / RF / DNN to output health status indices and classify them, facilitating online monitoring and early warning. At the same time, it establishes temperature thermal aging and vibration fatigue aging models and integrates degradation rates to achieve prediction of remaining service life (RUL), further supporting full life cycle management. 3) This invention optimizes "health assessment + life prediction" through multi-task learning and dynamically updates model parameters by incremental learning to adapt to changes in operating conditions. With the help of self-diagnosis, self-correction and intelligent decision support, it can identify risks earlier, improve the accuracy of fault prediction and response speed, and ultimately improve operation and maintenance efficiency, reduce operation and maintenance costs and enhance the safety and reliability of the power system. Attached Figure Description

[0032] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a comparison chart of the working characteristic curves in the embodiments of the present invention; Figure 3 This is a statistical comparison chart of the distribution of early warning lead time in embodiments of the present invention; Figure 4 This is a comparison chart of remaining lifetime predictions in embodiments of the present invention. Detailed Implementation

[0033] like Figure 1 As shown, the cable condition and life cycle assessment method based on the fusion of thermal and vibration information includes the following steps: Step 1: Acquisition, processing, and manipulation of multi-physical quantity data; (1) Signal acquisition: The temperature field of the cable is obtained by real-time acquisition of multiple physical quantities of the cable through distributed fiber optic sensors and microelectromechanical systems (MEMS) sensors. Vibration signals and environmental data .

[0034] The cable status dataset is as follows: (1); in, For time, This indicates the cable location.

[0035] (2) Signal preprocessing: The acquired temperature and vibration signals are subjected to noise reduction, smoothing, and standardization. Temperature signal The vibration signal is smoothed using a Kalman filter. Frequency band decomposition is performed using wavelet packet transform.

[0036] The Kalman filter formula is: (2); The decomposition of the vibration signal using wavelet packets is as follows: (3); in ， These are the wavelet decomposition results at different scales. For node indexing; The number of components involved in the reconstruction.

[0037] (3) Feature extraction and fusion: Using principal component analysis to study temperature ,vibration and environmental data Dimensionality reduction is performed to obtain a comprehensive feature set of the cable's operating status: (4); in, These are the characteristic vectors of temperature, vibration, and environmental factors, respectively.

[0038] Step 2: Cable health status and life cycle assessment model; (1) Health status prediction: Based on the extracted feature set A cable health status assessment model is established using Support Vector Machine (SVM), Random Forest (RF), or Deep Neural Network (DNN), and the model outputs a cable health status index. .

[0039] The health status prediction model is represented as follows: (5); in, It is a trained machine learning model. For model parameters, This is a cable health status index.

[0040] (2) Health Index Calculation: Cable health status index The Sigmoid function is used for calculation, and the formula is as follows: (6); in, The regression coefficients control for the contribution of different characteristics to health assessment.

[0041] (3) Cable condition classification: According to the health status index The cable condition is divided into multiple levels, each corresponding to a different fault risk: (7); If the health status index is within (0–0.2], it indicates a serious failure risk; if the health status index is within (0.2–0.5], it indicates a multi-quantile risk; if the health status index is within (0.5–0.8], it indicates a slight quantile risk; if the health status index is within (0.8–1], it indicates a normal operating condition.

[0042] The grading system helps to monitor the health status of cables in real time and provide early warnings.

[0043] (4) Thermal aging rate: Aging rate of cable insulation materials The modified Arrenius model indicates that: (8); in, As the pre-factor, Thermal activation energy, Boltzmann's constant, This represents the operating temperature of the cable. The formula illustrates the effect of temperature on the cable's aging rate.

[0044] (5) Vibration aging rate: Vibration has a significant impact on cable fatigue damage, and the vibration aging rate... It can be calculated using the following formula: (9); in, The power spectral density of the vibration signal. Let be the frequency sensitivity function of the vibration. The cutoff frequency is the vibration frequency.

[0045] (6) Overall aging rate: The aging rate of a cable, which combines the effects of temperature and vibration, is expressed as: (10); This formula reflects the overall degradation rate of cables under different operating conditions.

[0046] (7) Remaining useful life (RUL) prediction: By comprehensive aging rate The following integral formula is used to predict the remaining useful life (RUL) of the cable: (11); in, Indicates the remaining service life. For the current moment, For the target time, The overall aging rate varies over time.

[0047] Step 3: Jointly optimize the cable health status assessment model and the cable life cycle assessment model; (1) Multi-task learning MTL optimization: A multi-task learning MTL model is used to jointly optimize the cable health status assessment model and the life cycle assessment model. The loss function used is: (12); in, For health status assessment loss function, For lifecycle prediction loss function, The coefficient is used to balance the weights of the two tasks.

[0048] (2) Dynamic updates and learning: Use incremental learning algorithms to refine model parameters. Dynamic updates are performed to allow the model to adapt to changes in the cable's operating condition. The update formula is: (13); in, For learning rate, The gradient of the loss function. These are the old parameter values ​​for the model. These are the updated parameter values ​​for the model.

[0049] Step 4: Optimize real-time monitoring and self-correction of cable management, as well as intelligent decision-making throughout the cable lifecycle; (1) Real-time monitoring and self-correction function: Based on sensor output errors and environmental changes, the system provides self-diagnosis and self-correction functions to ensure the accuracy of cable condition assessment. Through online learning methods, the model can gradually update and adjust its parameters as new health status data is input. When dynamically updating the cable condition model, real-time feedback and self-calibration further optimize the results.

[0050] (2) Intelligent decision-making throughout the entire cable lifecycle: By combining real-time monitoring data with historical data, intelligent decision support is provided for cable maintenance, repair, and replacement, thereby optimizing the operation and maintenance of the power system.

[0051] Example: (a) Example Structure The case study is based on multi-source monitoring data during cable operation and is organized according to the following method: "data input—scenario comparison—unified calculation process—index evaluation—result comparison". Figure 1 As shown.

[0052] (II) Scene Description 1) Scenario I: Considering only the anomaly detection and health status assessment of temperature information, the health status index and classification results are output through Kalman filtering of temperature signals, PCA feature dimensionality reduction, and a type of SVM classification health assessment model, serving as a baseline comparison for the multi-source fusion scheme.

[0053] 2) Scenario II: Considering the combined effects of temperature and vibration information, a comprehensive assessment of cable health status and life cycle is conducted. This is achieved through temperature Kalman filtering, vibration wavelet packet decomposition, PCA feature fusion, and an RF health assessment model. The thermal aging rate is calculated based on the modified Arrenius model, and the vibration aging rate is calculated by combining the vibration power spectral density. The resulting comprehensive aging rate is used to predict the remaining service life. The cable health status index, graded early warning results, and remaining service life prediction values ​​are output simultaneously. The accuracy, false alarm / missed alarm rate, early warning lead time, and life prediction error are comprehensively compared with Scenario I to verify the comprehensive advantages of multi-source information fusion in improving status assessment and achieving full life cycle management.

[0054] (III) Scene Calculation The specific steps for conducting simulation calculations in different scenarios are as follows: S1) Signal acquisition and data construction; Deploying a distributed network of fiber optic sensors and microelectromechanical systems (MEMS) sensors, the system synchronously acquires data in real time during cable operation to obtain the temperature field. Vibration signals and environmental data With timestamps With cable location To index, construct a multidimensional state dataset. This provides a unified data input for solving subsequent scenarios.

[0055] S2) Signal preprocessing and noise reduction; Temperature signal Kalman filtering is used for smoothing to suppress noise and abrupt changes; for vibration signals Wavelet packet decomposition is used for frequency band division and reconstruction to extract effective vibration components and reduce interference; environmental data... Missing values ​​were imputed and standardized to create a preprocessed dataset of consistent quality.

[0056] S3) Feature extraction and fusion; Feature construction was performed on the preprocessed temperature, vibration, and environmental data, and principal component analysis (PCA) was used for dimensionality reduction and fusion to obtain a low-dimensional comprehensive feature vector. .Will It serves as a unified input feature for health assessment and life prediction models, enabling the same-scale expression of multi-source information.

[0057] S4) Cable health status assessment; (1) Based on feature vectors We construct and train Support Vector Machine (SVM), Random Forest (RF), and Deep Neural Network (DNN) models to output a cable health status index. ; (2) Using the Sigmoid function to... Perform normalization mapping to obtain comparable health indices; (3) Classify the status according to the health index: severe fault (0–0.2), multiple status (0.2–0.5), slight status (0.5–0.8), normal operation (0.8–1), to realize status identification and real-time early warning triggering.

[0058] S5) Cable life cycle assessment; (1) Based on temperature data, the thermal aging rate was calculated using the modified Arrenius model. ; (2) Calculate the vibration aging rate based on characterization parameters such as vibration power spectral density. ; (3) The overall aging rate is obtained by taking into account the combined effects of temperature and vibration. Based on this, a cumulative degradation calculation is performed to predict the remaining useful life.

[0059] (iv) Results Analysis; In Scenario I, anomaly detection and health status assessment are based solely on temperature information, serving as a baseline control with only one information source. In contrast, Scenario II further integrates vibration information, resulting in a more comprehensive state representation. As shown in Table 1, the accuracy of Scenario II improved from 0.959 to 0.983, the precision from 0.562 to 0.830, and the F1 score from 0.705 to 0.838. Furthermore, the 95% confidence interval of F1 score improved from [0.644, 0.765] to [0.782, 0.885], indicating that the overall recognition performance and comprehensive stability are significantly enhanced after the temperature and vibration fusion.

[0060] Furthermore, Table 1 also shows that the recall rate for Scenario II decreased from 0.942 to 0.846, indicating that while the fusion solution achieves "fewer false positives," its anomaly coverage capability is somewhat reduced. On the one hand, the addition of vibration information enhances the model's ability to distinguish non-fault disturbances, making the judgment more cautious, thus improving precision and F1 score. On the other hand, when anomalies are in their early stages or have weak representations, a conservative judgment boundary may lead to some anomalies not being identified in time, resulting in a certain risk of missed detections. Therefore, the fusion solution is more inclined towards "reducing false positives," and the threshold or loss weight can be optimized in the future based on business risk preferences to balance "fewer false positives" and "fewer missed detections."

[0061] Table 2 presents the AUC and false positive / false negative rates for alarm and ranking performance: In Scenario II, the AUC improved from 0.993 to 0.995 (95% CI improved from [0.990, 0.996] to [0.993, 0.997]), the false positive rate decreased from 0.040 to 0.009, but the false negative rate increased from 0.058 to 0.154. This result further confirms that the fusion solution is stronger in suppressing false positives, but the cost of false negatives increases under the current strategy. Therefore, in engineering applications, alarm thresholds should be recalibrated or cost-sensitive constraints should be adopted, taking into account the operational goals of "more sensitive to false negatives / more sensitive to false positives," to obtain an alarm strategy that better meets actual needs.

[0062] Meanwhile, statistics on early warning lead times show that the mean for scenario I is 23.933 and the median is 26.000, while the mean for scenario II is 22.467 and the median is 22.000, showing a slight overall decline. However, at the lower quantiles, P10 improved from 1.600 to 7.200, and P90 remained largely consistent (35.600 vs 35.800). This indicates that the fusion scheme provides better early warning stability for the "worst samples." On one hand, the fused information provides additional anomaly evidence for some difficult samples, enabling earlier alarm triggering in extreme cases; on the other hand, the more cautious overall judgment leads to delayed triggering for some samples, thus lowering the mean / median. Therefore, reasonable thresholds and alarm strategies are expected to further improve the overall early warning lead time while maintaining the improvement in P10.

[0063] Finally, regarding lifetime prediction, the MAE in Scenario II decreased from 2.757 steps to 1.500 steps, the MAPE decreased from 11.045% to 5.932%, and the absolute error decreased from 6.000 to 4.000. This indicates that the fusion scheme not only improved the classification performance of health assessment but also achieved better results in lifetime prediction accuracy, thus forming a closed-loop gain across the entire life cycle of "state assessment + lifetime prediction".

[0064] Table 1: Classification Indicators;

[0065] Table 2: Ranking and Alarm Indicators;

[0066] Table 3. RUL (Remaining Life) Prediction Error;

[0067] Table 4: Early Warning Lead Time;

[0068] In summary, this invention first collects time-series temperature data, vibration signals, and environmental data of the target cable to form a multi-source state dataset. Then, the temperature signal is smoothed using Kalman filtering, the vibration signal is decomposed into frequency bands and denoised using wavelet packet decomposition, and the environmental data is processed for missing data and standardized to obtain preprocessed multi-source signals. Based on this, temperature features, vibration features, and environmental features are extracted and fused. Principal component analysis is used to reduce the dimensionality of the fused features to obtain a comprehensive feature vector. Then, using this comprehensive feature vector as input, a deep neural network health assessment model is constructed to output a health status index, which is then processed using a Sigmoid function. The mapping normalizes and calibrates the health status index, and classifies the health status according to preset thresholds. Furthermore, it calculates the thermal aging rate based on temperature data and the vibration aging rate based on vibration data, and merges them to obtain the comprehensive aging rate. The remaining life is predicted by integrating the comprehensive aging rate. At the same time, multi-task learning is used to jointly optimize and train the health assessment task and the life prediction task. During online operation, incremental learning is performed based on newly collected data to dynamically update the model parameters, thereby realizing the real-time assessment output of the cable's health status and remaining life, and providing monitoring, alarm and decision support for the whole life cycle operation and maintenance.

Claims

1. A cable condition and life cycle assessment method based on the fusion of thermal and vibration information, characterized in that, Includes the following steps: Step 1: Perform multi-physical quantity data acquisition, processing, and manipulation; Step 2: Construct a cable health status assessment model; Step 3: Obtain the cable lifecycle assessment model; Step 4: Jointly optimize the cable health status assessment model and the cable life cycle assessment model; Step 5: Optimize real-time monitoring and self-correction of cable management, as well as intelligent decision-making throughout the cable lifecycle; Step 6: Conduct simulation calculations in different scenarios.

2. The method according to claim 1, characterized in that, In step 1, the following steps are specifically adopted: 1-1) The temperature field of the cable is obtained by real-time acquisition of multiple physical quantities of the cable through distributed fiber optic sensors and microelectromechanical systems (MEMS) sensors. Vibration signals and environmental data ; 1-2) Perform signal preprocessing, including the temperature signal. The vibration signal is smoothed using a Kalman filter. Frequency band decomposition is performed using wavelet packet transform; 1-3) Feature extraction and fusion; including using principal component analysis (PCA) for temperature... ,vibration and environmental data Dimensionality reduction is performed to obtain a comprehensive feature set of the cable's operating status.

3. The method according to claim 2, characterized in that, In step 1-1), the cable status dataset obtained is as follows: （1）； in, For time, Location of the cable; In steps 1-2), the Kalman filter formula used is: （2）； The decomposition of the vibration signal using wavelet packets is as follows: （3）； in ， The wavelet decomposition results are shown at different scales. For node indexing; The number of components participating in the reconstruction; In steps 1-3), the resulting comprehensive feature set is: （4）； in, These are the characteristic vectors of temperature, vibration, and environmental factors, respectively.

4. The method according to any one of claims 1 to 3, characterized in that, In step 2, constructing the cable health status assessment model specifically includes: 2-1) Obtain a health status prediction model: Based on the extracted feature set A cable health status assessment model is established, and the model outputs the cable health status index. ; The health status prediction model is represented as follows: （5）； in, It is a trained machine learning model. For model parameters, This refers to the cable health status index. 2-2) Calculate the health index: Cable health status index The Sigmoid function is used for calculation, and the formula is as follows: （6）； in, The regression coefficients control for the contribution of different characteristics to the health assessment. 2-3) Cable condition classification: According to the health status index The cable condition is divided into multiple levels, each corresponding to a different fault risk: （7） If the health status index is within (0–0.2], it indicates a serious failure risk; if the health status index is within (0.2–0.5], it indicates a multi-quantile risk; if the health status index is within (0.5–0.8], it indicates a slight quantile risk; if the health status index is within (0.8–1], it indicates a normal operating condition. The grading system helps to monitor the health status of cables in real time and provide early warnings.

5. The method according to any one of claims 1 to 4, characterized in that, In step 3, obtaining the cable lifecycle assessment model specifically includes: 3-1) Thermal aging rate parameters: Aging rate of cable insulation materials The modified Arrenius model indicates that: （8）； in, As the pre-factor, Thermal activation energy, Boltzmann's constant, The operating temperature of the cable; this formula represents the effect of temperature on the aging rate of the cable. 3-2) Vibration aging rate parameters: Vibration has a significant impact on cable fatigue damage, and the vibration aging rate... It can be calculated using the following formula: （9）； in, The power spectral density of the vibration signal. Let be the frequency sensitivity function of the vibration. The cutoff frequency is the vibration frequency. 3-3) Overall aging rate parameters: The aging rate of a cable, which combines the effects of temperature and vibration, is expressed as: （10）； This formula reflects the overall degradation rate of cables under different operating conditions.

6. The method according to claim 5, characterized in that, This also includes steps 3-4) predicting the remaining useful life (RUL): By comprehensive aging rate The following integral formula is used to predict the remaining service life (RUL) of the cable: （11）； in, Indicates the remaining service life. For the current moment, For the target time, The overall aging rate varies over time.

7. The method according to claim 1, 2, 3, 4, or 6, characterized in that, In step 4, the joint optimization specifically includes: 4-1) The multi-task learning model (MTL) is used to jointly optimize the cable health status assessment model and the cable life cycle assessment model; 4-2) Use incremental learning algorithms to dynamically update the model parameters, enabling the model to adapt to changes in the cable's operating state; the update formula is: （13）； in, For learning rate, The gradient of the loss function. These are the old parameter values ​​for the model. These are the updated parameter values ​​for the model.

8. The method according to claim 7, characterized in that, The loss function used in step 4-1) is: （12）； in, For health status assessment loss function, For lifecycle prediction loss function, The coefficient is used to balance the weights of the two tasks.

9. The method according to claim 1, characterized in that, Step 5 specifically includes: 5-1) Real-time monitoring and self-correction function: Based on sensor output errors and environmental changes, it provides self-diagnosis and self-correction functions to ensure the accuracy of cable condition assessment; through online learning methods, the model can gradually update and adjust its parameters as new health status data is input. When dynamically updating the cable condition model, real-time feedback and self-calibration will further optimize the results. 5-2) Intelligent decision-making throughout the entire cable lifecycle: By combining real-time monitoring data with historical data, intelligent decision support is provided for cable maintenance, repair, and replacement, thereby optimizing the operation and maintenance of the power system.

10. The method according to claim 1, characterized in that, In step 6, simulation calculations are performed under different scenarios; 1) Scenario I: Considering only the anomaly detection and health status assessment of temperature information, the health status index and classification results are output through Kalman filtering of temperature signals, PCA feature dimensionality reduction, and a class of SVM classification health assessment models, serving as a baseline comparison for the multi-source fusion scheme. 2) Scenario II: Considering the combined effects of temperature and vibration information, a comprehensive assessment of cable health status and life cycle is conducted. This involves temperature Kalman filtering, vibration wavelet packet decomposition, PCA feature fusion, and an RF health assessment model. The thermal aging rate is calculated based on the modified Arrenius model, and the vibration aging rate is calculated by combining the vibration power spectral density. The resulting comprehensive aging rate is used to predict the remaining service life. The cable health status index, graded early warning results, and remaining service life prediction values ​​are output simultaneously. The accuracy, false alarm / missed alarm rate, early warning lead time, and life prediction error are comprehensively compared with Scenario I to verify the comprehensive advantages of multi-source information fusion in improving status assessment and achieving full life cycle management.

Images