A cable insulation aging degree evaluation residual life early warning system

Abstract

This invention relates to the field of cable operation and maintenance, aiming to solve the problems of traditional cable insulation assessment neglecting multi-factor coupling and low accuracy of lifespan prediction, and to provide a precise early warning system. This invention includes modules for data acquisition, improved multi-dimensional coupled aging assessment, optimized lifespan decay prediction, early warning output, and data storage. It collects multi-dimensional electrical and environmental parameters through sensors, quantifies the aging level through coupled modeling and dynamic weight calculation, predicts the remaining lifespan based on an improved Arrhenius model and time-series residual correction, and outputs three levels of early warning through multiple channels. This invention improves the accuracy of aging assessment and lifespan prediction, supports cable operation and maintenance decisions, reduces fault risks, and is applicable to power system cable insulation condition monitoring scenarios.

Description

Technical Field

[0001] This invention relates to the field of cables, specifically to a cable insulation aging assessment and remaining life early warning system. Background Technology

[0002] As the core carrier of power transmission, the aging of the insulation layer of cables is a major cause of accidents such as short circuits and power outages. Existing insulation aging assessment and lifespan early warning systems have two major pain points, making it difficult to meet the needs of safe operation and maintenance of power systems: The multi-factor coupled aging assessment has large biases: Insulation aging is affected by multiple factors such as partial discharge, dielectric loss, and ambient temperature and humidity, and there are nonlinear coupling relationships between these factors (e.g., high temperature amplifies the destructive effect of partial discharge on insulation). Traditional systems often use single-parameter threshold judgment or simple weighted summation methods for assessment, without considering the coupling effect, resulting in an aging level misjudgment rate of over 30%, and easily leading to problems such as "mild aging being misjudged as moderate" or "severe aging being missed".

[0003] The accuracy of remaining lifetime prediction is severely affected: existing prediction algorithms are mostly based on fixed aging rate models, which cannot adapt to environmental parameter fluctuations (such as seasonal temperature and humidity changes) and aging abrupt changes (such as sudden enhancement of partial discharge), resulting in prediction errors generally exceeding 20%, and they cannot provide early warning of the risk of accelerated aging and degradation, making it difficult to support operation and maintenance decisions.

[0004] In addition, some systems suffer from lagging data collection and rigid early warning logic, which further reduces the operational support capabilities. However, these problems can be solved through conventional technical optimizations, and the core bottlenecks remain the two specific pain points mentioned above. Summary of the Invention

[0005] The purpose of this invention is to provide a cable insulation aging degree assessment and remaining life early warning system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a cable insulation aging degree assessment and remaining life early warning system, comprising a data acquisition module, an improved multi-dimensional coupled aging assessment module, an optimized life decay prediction module, an early warning output module, and a data storage module; the data acquisition module uses a high-frequency partial discharge sensor, a dielectric loss tester, and a temperature and humidity sensor to collect cable operating electrical parameters and environmental parameters in real time, obtaining multi-dimensional parameter data. The sampling frequency is 1 time / hour, and the data transmission delay is ≤10s, ensuring data real-time performance and integrity, and providing accurate input data for the two core innovative modules. The cable operating electrical parameters include partial discharge quantity. and dielectric loss tangent Environmental parameters include temperature and humidity ; The improved multi-dimensional coupled aging assessment module adopts an improved weighted adaptive coupling algorithm to model the coupling relationship of the multi-dimensional parameter data transmitted by the data acquisition module, perform weighted adaptive calculation and coupled aging index calculation, quantify the degree of insulation aging and output four aging levels, namely mild, moderate, severe and near failure. It solves the problem of large assessment deviation caused by single parameter dependence and neglect of the coupling effect of multiple factors in traditional assessment. The optimized lifespan decay prediction module employs an improved time-series residual correction algorithm. Based on the aging level output by the improved multi-dimensional coupled aging assessment module, it constructs an improved Arrhenius basic lifespan model. Through time-series residual extraction, mutation identification, and dynamic correction, it predicts the remaining lifespan and decay trend of the cable, fundamentally solving the problem of insufficient accuracy caused by environmental interference and aging mutations in traditional prediction. The early warning output module, based on the aging level of the improved multi-dimensional coupled aging assessment module and the lifespan result of the optimized lifespan decay prediction module, outputs three levels of early warning information according to preset thresholds (Level 1: Remaining lifespan ≥ 5 years, prompting routine maintenance; Level 2: 2 years ≤ Remaining lifespan < 5 years, prompting enhanced monitoring; Level 3: Remaining lifespan < 2 years, prompting immediate replacement). This information is pushed through multiple channels including the maintenance platform pop-up, SMS, and audible and visual alarms, simultaneously displaying core influencing factors to support maintenance decisions. The data storage module retains the multi-dimensional parameter data collected by the data acquisition module, the evaluation and prediction results of the improved multi-dimensional coupled aging assessment module and the optimized life decay prediction module, as well as the full-process operation data. The retention period is ≥5 years. This provides historical data support for the coupling coefficient and correction parameter iterative optimization of the improved multi-dimensional coupled aging assessment module and the optimized life decay prediction module, while ensuring that the system operation trajectory is traceable.

[0007] Preferably, the specific implementation steps of the improved multi-dimensional coupled aging assessment module are as follows: Step 1: Acquisition and Standardization of Evaluation Factors: Obtain four core evaluation factors from the data acquisition module, namely, partial discharge quantity. Dielectric loss tangent Ambient temperature Ambient humidity All standardized to Intervals, eliminating the influence of dimensions; where Reflects the degree of internal defects in the insulation. Reflects the degradation of the insulating dielectric properties. , As an environmental coupling factor; Step 2, Coupling Relationship Modeling and Adaptive Weight Calculation: Construct the "Electrical Factor-Environmental Factor" coupling matrix, and calculate the coupling strength between factors using an improved grey relational analysis algorithm, letting... and Coupling coefficient , and Coupling coefficient , and The coupling coefficient is Coupling coefficient range The larger the value, the stronger the coupling effect; the weights are dynamically allocated based on the coupling strength: the basic weight of the electrical factor is 0.6, the basic weight of the environmental factor is 0.4, and then adjusted according to the coupling coefficient: , , , This ensures that the weights dynamically adapt to the coupling relationship; Step 3, Calculation of Coupled Aging Index: The aging index is calculated using an improved coupled weighted summation formula. : At the same time, a coupling correction term is introduced. Final aging index The range of values ​​is ; Step 4, Aging Level Determination: According to The aging process is divided into four levels: mild aging (0-0.3), moderate aging (0.3-0.6), severe aging (0.6-0.9), and near-failure (0.9-1.2). The system outputs the precise aging level and core influencing factors (e.g., "moderate aging is mainly affected by the coupling effect of high temperature and partial discharge").

[0008] Preferably, the specific implementation steps of step 2, coupling relationship modeling and weight adaptive calculation, are as follows: Step 2.1: Construct the electrical factor-environmental factor coupling matrix: Based on the obtained four types of core factors, according to electrical factors (partial discharge quantity)... Dielectric loss tangent - Environmental factors (temperature) ,humidity The interaction logic of ) is used to construct a coupling matrix and clarify the coupling relationships that need to be quantified as " and " and (These two sets of factors account for over 80% of the influence of insulation aging, while the influence of other cross-relationships is negligible.) The coupling coefficients in the coupling matrix relate electrical factors to environmental factors, which are the core aging influence mechanism. The first set of couplings... Partial discharge quantity temperature High temperature amplification The second set of couplings is affected by the breakdown of the insulation. Dielectric loss tangent humidity High humidity exacerbates The rise in temperature accelerates insulation aging. Step 2.2: Calculate the coupling coefficient using the improved grey relational analysis algorithm: An improved grey relational analysis algorithm (introducing a factor time-series correlation correction term to address the problem of insufficient adaptation to dynamic data in traditional algorithms) is used to calculate the coupling coefficient of the two sets of coupling relationships. (Coupling coefficient between temperature and partial discharge quantity) (Coupling coefficient between humidity and dielectric loss tangent), specific operation: Data preprocessing: Extract time series data for the past 30 days, including hourly data. , , , Standardized values ​​(following the standardized results from the previous text, range) This forms two sets of time series: and , and ; Correlation calculation: Introducing time series correction coefficient ( (used to balance the near-term and long-term effects of time series data), calculating the grey relational degree of each set of sequences. The formula is: For the first Class of electrical factors in the first Time series values ​​at each time point, For the first Environmental factors in the first Time series values ​​at each time point, For time nodes (values ​​from 1 to 720, corresponding to 30 days × 24 hours). The minimum absolute difference between the two sets of sequences. The maximum absolute difference between the two sets of sequences; Coupling coefficient transformation: transforming the correlation coefficient Normalization to The coupling coefficient is obtained from the interval. , ,Right now ( (This represents the maximum value of the correlation between the corresponding sequences), a larger value indicates a stronger coupling effect between the two sets of factors (e.g., This indicates that the coupling effect between high temperature and partial discharge is extremely strong.

[0009] Step 2.3, Adaptive Weight Allocation Based on Coupling Coefficient: The weights of the four types of factors are dynamically allocated using a combination of basic weights and coupling corrections to ensure that the weights match the strength of the coupling effect. Specific rules are as follows: Basic weight setting: Combining insulation aging mechanism and electrical factor , The core influencing factors have a combined basic weight of 0.6; environmental factors , As a coupling influencing factor, the total basic weight is 0.4, and the basic weight is evenly distributed among the two types of factors. , 0.3 each , 0.2 each); Coupling correction calculation: The weights are adjusted based on the coupling coefficient. The correction formula is as follows: Parameter description: Partial discharge quantity The dynamic weight, 0.3 is The basic weights, For temperature and The coupling coefficient; Parameter description: The tangent of the dielectric loss angle The dynamic weight, 0.3 is The basic weights, For humidity and The coupling coefficient; Parameter description: For temperature The dynamic weight, 0.2 is The basic weights, For temperature and The coupling coefficient; Parameter description: Humidity The dynamic weight, 0.2 is The basic weights, For humidity and The coupling coefficient; Weight normalization calibration: After correction, the four weights are summed. If the sum deviates from 1, normalization is performed to ensure that the total weight sum is 1. Finally, dynamic weights that adapt to the current coupling state are output, providing an accurate basis for subsequent aging index calculation.

[0010] Preferably, the optimized lifetime degradation prediction module is implemented using the following logic: Step S1: Construction of the basic lifespan model: based on the aging index By combining the improved Arrhenius equation (adapted to the aging characteristics of cable insulation materials and optimized for temperature influence coefficients), a basic remaining life model is constructed, as shown in the following formula: ,in Based on remaining lifespan (in years). The rated service life of the cable insulation layer (determined based on the properties of the insulation material, such as PVC cables). Year), The material property coefficient (fitted to 0.015 by multiple insulation aging experiments, unit: ℃⁻¹) The final aging index (range 0 to 1.2) calculated above. The actual value of the ambient temperature (unit: °C, non-standardized value, ranging from -20 °C to 85 °C). It is the natural constant (approximately 2.718); Step S2, Temporal Residual Extraction and Mutation Identification: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: ,in The lifetime reference value is derived from the real-time condition of the cable insulation (by...). (Calculated based on the aging mechanism of insulation materials); the residuals were normalized to... Mapped to Range, excluding extreme outliers (such as those caused by sensor malfunctions). The data points are replaced with the mean of the residuals of the three adjacent nodes, ultimately forming a smoothed residual sequence. This ensures that the residual data accurately reflects the aging fluctuation pattern. Step S3, Dynamic Residual Correction and Lifetime Adjustment: For both normal environmental fluctuations and abrupt aging changes, lifetime adjustment is performed separately to ensure prediction accuracy, as detailed below: Conventional environmental fluctuation correction: A linear residual correction formula is used to adjust the baseline lifetime, adapting to stable fluctuations in environmental parameters such as temperature and humidity. ,in The remaining lifespan (in years) is adjusted for environmental fluctuations. Based on the remaining lifetime, n is the amount of time-series data (here n=720, corresponding to hourly data for the past 3 months). This is the sum of the residuals at 720 time points. This represents the mean of the residuals (reflecting the overall impact of environmental fluctuations on the prediction results).

[0011] Aging mutation correction: For identified aging mutation points, a mutation correction factor is introduced. Adjust lifespan using the following formula: ,in The final revised remaining lifetime (in years). This refers to the lifespan after environmental fluctuation correction. The mutation correction factor (value 1.2-1.5, adaptively adjusted according to the mutation magnitude) is used. hour , hour (Linearly increasing) The influence of sudden changes on lifespan decay is amplified by the coefficient to ensure that the prediction results are close to the actual aging state; at the same time, based on the corrected lifespan data, an aging decay trend curve is plotted, with time as the horizontal axis and remaining lifespan as the vertical axis, to intuitively present the lifespan change pattern. Step S4, Lifetime Warning Threshold Setting: Based on the corrected remaining lifetime Three warning thresholds are set: Level 1 warning is... In 2023, a routine maintenance reminder was issued; a Level 2 warning was given. In [year], it was suggested to strengthen monitoring; Level III warning was [issued]. The system will prompt immediate replacement upon completion of the current lifespan, and output the final lifespan prediction result and warning level.

[0012] Preferably, the specific implementation steps of the temporal residual extraction and mutation identification in step S2 are as follows: Step S2.1, Time Series Data Regularization and Alignment: Collect data from the past 3 months. Time-series data (continuing from the output of the improved multi-dimensional coupled aging assessment module mentioned earlier, sampled at a frequency of 1 time per hour, totaling 720 data points) forms the actual time-series sequence. The subscripts correspond to time nodes (from morning to night); simultaneously, prediction data from the baseline life model for the same period are extracted, based on each time node. The value is used to deduce the predicted basic remaining life at the corresponding time. To form a predicted time series This ensures that the actual data and the predicted data are fully aligned in the time dimension, eliminating the impact of time deviation on residual calculation; Step S2.2, Residual Sequence Extraction and Preprocessing: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: , For the first The residuals (range) at each time point ), For the first The basic remaining lifetime prediction value at each time point. For the first The life reference values ​​for each time point are derived from the real-time state of cable insulation (by...). Calculated based on the aging mechanism of insulation materials (unit: years). The residuals are normalized. Mapped to Range, excluding extreme outliers (such as those caused by sensor malfunctions). The data points are replaced with the mean of the residuals of the three adjacent nodes, ultimately forming a smoothed residual sequence. This ensures that the residual data accurately reflects the aging fluctuation pattern. Step S2.3, Deployment of the Improved Sliding Window Algorithm: An improved sliding window algorithm with a weight decay factor is adopted (to optimize the problem of the traditional fixed window's lag response to sudden changes). The window size is set to 7 days (168 data points), the window sliding step is 1 hour, and a time-series weight decay factor is introduced. (Data weights are gradually reduced from recent to distant data to increase the sensitivity of recent data to mutation identification); the weighted residual mean within each window is calculated as follows: ,in For the current window The weighted residual mean, The time-series weight decay factor is 0.9. Data within the window and the window's end time node The time interval (unit: hours, value from 0 to 167). For the first The residual values ​​at each time point are calculated as follows: the numerator is the sum of weighted residuals within the window, and the denominator is the sum of weights (used for normalization to ensure that the mean range is consistent with the residuals). For the current window, This is the window's termination time point; Step S2.4, Aging Mutation Point Determination and Verification: Constructing mutation determination logic based on the weighted residual mean: The first step is to calculate the difference between the weighted residual mean values ​​of two adjacent windows. ,in For the first The weighted residual mean of each window, For the first The weighted residual mean of each window; The second step sets the mutation threshold to 0.05 (determined through fitting multiple sets of cable aging experiments; this threshold balances the accuracy of mutation identification with the false alarm rate). At that time, it was initially identified as a potential aging mutation point; The third step combines the original... Time series data verification, if the corresponding time node The increase is ≥0.1 (i.e., the aging level crosses a grade trend), and the associated electrical / environmental factors (such as...) Sudden growth If there are abnormal fluctuations (such as sudden increases), then the node is confirmed as an aging mutation point to avoid misjudgment caused by temporary environmental fluctuations. Step S2.5, Mutation Information Marking and Output: For confirmed aging mutation points, mark three core pieces of information: first, the mutation time (accurate to the hour); second, the magnitude of the mutation impact (…). Values ​​and corresponding values The third is mutation-related factors (such as "a sudden increase of 30% in partial discharge, accompanied by a 5°C increase in ambient temperature, triggering aging mutations"), which synchronize the labeled mutation information to subsequent lifetime correction steps, providing a basis for targeted adjustments to lifetime prediction results.

[0013] Compared with the prior art, the beneficial effects of the present invention are: Multi-factor coupling modeling eliminates assessment bias: This invention constructs an electrical factor-environmental factor coupling matrix, quantifies the coupling strength by improving the grey relational algorithm, dynamically allocates weights and introduces coupling correction terms, which solves the limitations of traditional single-parameter assessment or simple weighting, reduces the misjudgment rate of aging level, and improves the accuracy of four-level determination of mild / moderate / severe / near failure.

[0014] Dynamic weights adapt to complex operating conditions: The weights of this invention are adjusted in real time according to the coupling relationship between temperature-partial discharge and humidity-dielectric loss, adapting to the aging characteristics under different operating environments. Compared with fixed weight evaluation, the evaluation accuracy under complex operating conditions is improved, and the latent aging caused by coupling effect can be accurately identified.

[0015] This invention employs linear residual correction to address routine environmental fluctuations. It also introduces a mutation correction coefficient to amplify the attenuation effect of aging mutations, thereby reducing prediction errors and achieving a remaining life prediction deviation of less than 0.5 years, thus providing accurate data support for operation and maintenance decisions.

[0016] Early warning of risk due to mutation identification: This invention uses an improved sliding window algorithm to accurately identify aging mutation points, mark the mutation time, impact magnitude and related factors, and provide early warning of accelerated aging degradation risk 3-6 months in advance, avoiding short circuits and power outages caused by sudden insulation failure. Attached Figure Description

[0017] Fig. 1 This is a schematic diagram of the system structure of the present invention; Fig. 2 This is a schematic diagram of the workflow of the improved multi-dimensional coupled aging assessment module of the present invention; Fig. 3 This is a schematic diagram of the workflow of the optimized lifetime decay prediction module of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figs. 1-3 This invention provides a technical solution: a cable insulation aging degree assessment and remaining life early warning system, comprising a data acquisition module, an improved multi-dimensional coupled aging assessment module, an optimized life decay prediction module, an early warning output module, and a data storage module. The data acquisition module uses a high-frequency partial discharge sensor, a dielectric loss tester, and a temperature and humidity sensor to collect cable operating electrical parameters and environmental parameters in real time, obtaining multi-dimensional parameter data. The sampling frequency is 1 time / hour, and the data transmission delay is ≤10s, ensuring data real-time performance and integrity, providing accurate input data for the two core innovative modules. The cable operating electrical parameters include partial discharge quantity. and dielectric loss tangent Environmental parameters include temperature and humidity The following describes the specific implementation steps of the data acquisition module: Sensor and testing equipment deployment and calibration: Addressing the core needs of cable insulation monitoring, complete precise equipment deployment and initial calibration. ① High-frequency partial discharge sensor: Installed at critical locations prone to partial discharge, such as cable joints and terminations, employing a Rogowski coil structure, with a frequency band set to 1MHz-100MHz, and a calibrated discharge measurement range of 10pC-1000pC, ensuring the capture of weak partial discharge signals. ); ② Dielectric loss tester: connected to the cable conductor and grounding terminal via a high-voltage lead, calibrated to a measurement accuracy of ±0.0001, compatible with cable rated voltage levels (10kV-220kV), accurately acquiring the dielectric loss tangent value ( ); ③ Temperature and humidity sensor: deployed in cable trenches, cable wells, and other cable operating environments, at a distance ≤5cm from the cable insulation layer, with a calibrated temperature measurement range of -20℃ to 85℃ (accuracy ±0.5℃) and a humidity measurement range of 0% to 100%RH (accuracy ±3%RH), simultaneously capturing environmental coupling factors ( , All devices are calibrated together after deployment to eliminate the impact of inherent device errors on the data.

[0020] Multi-parameter time-division synchronous acquisition: A time-division acquisition + synchronous alignment mode is adopted to balance acquisition accuracy and efficiency. ① Acquisition frequency control: All parameters are acquired at a fixed frequency of 1 time / hour, including partial discharge quantity (…). Continuous sampling was used (sampling time 10s / time), and the peak value was taken as the characteristic value for that hour; the dielectric loss tangent value ( ) Single-precision measurement (measurement time 5s / time) is used to avoid interference from high-frequency sampling on cable operation; temperature ( ),humidity( ① Real-time sampling and averaging (sampling interval 1 second, averaging 60 data points) is used to reduce the impact of instantaneous environmental fluctuations. ② Time-series synchronization marking: A uniform timestamp (accurate to milliseconds) is added to each set of collected data to ensure... , , , The four types of parameters are fully aligned in the time dimension, providing a time-consistent data foundation for subsequent coupling matrix modeling and temporal residual extraction.

[0021] Data preprocessing and outlier removal: The raw data is preprocessed to eliminate interference and outliers. ① Dimensional unification preprocessing: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] , , , Mapped to respectively The interval (connected to the factor standardization step of the improved multidimensional coupled aging assessment module), where Normalized by "actual value / maximum measured value", Normalize by "(actual value + 20) / (85 + 20)". Normalization by "actual value / 100"; ② Outlier identification and correction: using... The criterion for removing outlier data is that a parameter value exceeding the mean ± 3 times the standard deviation is considered an anomaly (e.g., caused by sensor malfunction). (Sudden changes, such as abrupt changes in environmental humidity data) are replaced with the average of two adjacent collection periods, and the anomaly type and time are marked to facilitate subsequent data traceability; ③ Data smoothing: The preprocessed time series data is smoothed using the moving average method (with a window size of 3 collection periods) to reduce random interference and retain the true trend of data change.

[0022] Data Transmission and Real-Time Verification: A low-latency transmission link is constructed to ensure timely data delivery to the core module. ① Transmission Link Setup: A dual-link transmission mode of industrial Ethernet + wireless backup is adopted. The main link transmission delay is controlled to ≤8s, and the backup link (4G / 5G) delay is ≤10s, meeting the system transmission delay requirements. ② Data Integrity Verification: A CRC-32 checksum is added during transmission. The receiving end verifies each set of data. If the verification fails, a retransmission mechanism is triggered to ensure that the data is not lost or tampered with. ③ Data Distribution and Adaptation: After successful verification, the preprocessed standardized data is distributed to the improved multi-dimensional coupled aging assessment module and the optimized lifetime decay prediction module as needed—providing complete four-factor data for the improved multi-dimensional coupled aging assessment module and synchronizing time-series data for the optimized lifetime decay prediction module. Simultaneously, the original data and preprocessing results are retained in the data storage module, forming a closed-loop data acquisition system.

[0023] Improved Multi-Dimensional Coupled Aging Assessment Module: Employing an improved weighted adaptive coupling algorithm, this module models the coupling relationships of multi-dimensional parameter data transmitted from the data acquisition module, adaptively calculates weights, and performs coupled aging index calculations. It quantifies the degree of insulation aging and outputs four aging levels: mild, moderate, severe, and near failure. This module fundamentally addresses the significant assessment bias caused by reliance on a single parameter and neglect of the coupling effects of multiple factors in traditional assessments. Specific implementation steps are as follows: Step 1: Acquisition and Standardization of Evaluation Factors: Obtain four core evaluation factors from the data acquisition module, namely, partial discharge quantity. Dielectric loss tangent Ambient temperature Ambient humidity All standardized to Intervals, eliminating the influence of dimensions; where Reflects the degree of internal defects in the insulation. Reflects the degradation of the insulating dielectric properties. , As an environmental coupling factor; Step 2, Coupling Relationship Modeling and Adaptive Weight Calculation: Construct the "Electrical Factor-Environmental Factor" coupling matrix, and calculate the coupling strength between factors using an improved grey relational analysis algorithm, letting... and Coupling coefficient , and Coupling coefficient , and The coupling coefficient is Coupling coefficient range The larger the value, the stronger the coupling effect; the weights are dynamically allocated based on the coupling strength: the basic weight of the electrical factor is 0.6, the basic weight of the environmental factor is 0.4, and then adjusted according to the coupling coefficient: , , , This ensures that the weights dynamically adapt to the coupling relationship; It should be noted that the core of this step is to address the problem of fixed factor weights and neglected coupling effects in traditional assessments. Through dynamic modeling and weight adjustment, the assessment results are made to better reflect the actual scenario of multiple factors interacting. The specific implementation steps are as follows: Step 2.1: Construct the electrical factor-environmental factor coupling matrix: Based on the obtained four types of core factors, according to electrical factors (partial discharge quantity)... Dielectric loss tangent - Environmental factors (temperature) ,humidity The interaction logic of ) is used to construct a coupling matrix and clarify the coupling relationships that need to be quantified as " and " and (The coupling effect of these two sets of factors accounts for over 80% of the influence on insulation aging, while the influence of other cross-relationships is negligible), the matrix form is as follows: Coupling coefficients correlate electrical factors with environmental factors, forming a core aging impact mechanism; the first set of couplings. Partial discharge quantity temperature High temperature amplification The second set of couplings is affected by the breakdown of the insulation. Dielectric loss tangent humidity High humidity exacerbates The rise in temperature accelerates insulation aging. Step 2.2: Calculate the coupling coefficient using the improved grey relational analysis algorithm: An improved grey relational analysis algorithm (introducing a factor time-series correlation correction term to address the problem of insufficient adaptation to dynamic data in traditional algorithms) is used to calculate the coupling coefficient of the two sets of coupling relationships. (Coupling coefficient between temperature and partial discharge quantity) (Coupling coefficient between humidity and dielectric loss tangent), specific operation: Data preprocessing: Extract time series data for the past 30 days, including hourly data. , , , Standardized values ​​(following the standardized results from the previous text, range) This forms two sets of time series: and , and ; Correlation calculation: Introducing time series correction coefficient ( (used to balance the near-term and long-term effects of time series data), calculating the grey relational degree of each set of sequences. The formula is: For the first Class of electrical factors in the first Time series values ​​at each time point, For the first Environmental factors in the first Time series values ​​at each time point, For time nodes (values ​​from 1 to 720, corresponding to 30 days × 24 hours). The minimum absolute difference between the two sets of sequences. The maximum absolute difference between the two sets of sequences; Coupling coefficient transformation: transforming the correlation coefficient Normalization to The coupling coefficient is obtained from the interval. , ,Right now ( (This represents the maximum value of the correlation between the corresponding sequences), a larger value indicates a stronger coupling effect between the two sets of factors (e.g., This indicates that the coupling effect between high temperature and partial discharge is extremely strong.

[0024] Step 2.3, Adaptive Weight Allocation Based on Coupling Coefficient: The weights of the four types of factors are dynamically allocated using a combination of basic weights and coupling corrections to ensure that the weights match the strength of the coupling effect. Specific rules are as follows: Basic weight setting: Combining insulation aging mechanism and electrical factor , The core influencing factors have a combined basic weight of 0.6; environmental factors , As a coupling influencing factor, the total basic weight is 0.4, and the basic weight is evenly distributed among the two types of factors. , 0.3 each , 0.2 each); Coupling correction calculation: The weights are adjusted based on the coupling coefficient. The correction formula is as follows: Parameter description: Partial discharge quantity The dynamic weight, 0.3 is The basic weights, For temperature and The coupling coefficient; Parameter description: The tangent of the dielectric loss angle The dynamic weight, 0.3 is The basic weights, For humidity and The coupling coefficient; Parameter description: For temperature The dynamic weight, 0.2 is The basic weights, For temperature and The coupling coefficient; Parameter description: Humidity The dynamic weight, 0.2 is The basic weights, For humidity and The coupling coefficient; Weight normalization calibration: After correction, the four weights are summed. If the sum deviates from 1, normalization is performed to ensure that the total weight sum is 1. Finally, dynamic weights that adapt to the current coupling state are output, providing an accurate basis for subsequent aging index calculation.

[0025] Step 3, Calculation of Coupled Aging Index: The aging index is calculated using an improved coupled weighted summation formula. : At the same time, a coupling correction term is introduced. Final aging index The range of values ​​is ; Step 4, Aging Level Determination: According to Aging is classified into four levels: mild aging (0-0.3), moderate aging (0.3-0.6), severe aging (0.6-0.9), and near-failure (0.9-1.2). The system outputs precise aging levels and core influencing factors (e.g., "Moderate aging, mainly affected by the coupling effect of high temperature and partial discharge"). Optimized Lifespan Degradation Prediction Module: Employing an improved temporal residual correction algorithm, and based on the aging level output by the improved multi-dimensional coupled aging assessment module, an improved Arrhenius basic lifespan model is constructed. Through temporal residual extraction, abrupt change identification, and dynamic correction, the remaining lifespan and degradation trend of the cable are predicted. This module fundamentally addresses the insufficient accuracy issues caused by environmental interference and abrupt aging changes in traditional prediction methods. The specific implementation logic is as follows: Step S1: Construction of the basic lifespan model: based on the aging index By combining the improved Arrhenius equation (adapted to the aging characteristics of cable insulation materials and optimized for temperature influence coefficients), a basic remaining life model is constructed, as shown in the following formula: ,in Based on remaining lifespan (in years). The rated service life of the cable insulation layer (determined based on the properties of the insulation material, such as PVC cables). Year), The material property coefficient (fitted to 0.015 by multiple insulation aging experiments, unit: ℃⁻¹) The final aging index (range 0 to 1.2) calculated above. The actual value of the ambient temperature (unit: °C, non-standardized value, ranging from -20 °C to 85 °C). It is the natural constant (approximately 2.718); Step S2, Temporal Residual Extraction and Mutation Identification: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: ,in The lifetime reference value is derived from the real-time condition of the cable insulation (by...). (Calculated based on the aging mechanism of insulation materials); the residuals were normalized to... Mapped to Range, excluding extreme outliers (such as those caused by sensor malfunctions). The data points are replaced with the mean of the residuals of the three adjacent nodes, ultimately forming a smoothed residual sequence. To ensure that the residual data accurately reflects the aging fluctuation pattern, the specific implementation steps are as follows: Step S2.1, Time Series Data Regularization and Alignment: Collect data from the past 3 months. Time-series data (continuing from the output of the improved multi-dimensional coupled aging assessment module mentioned earlier, sampled at a frequency of 1 time per hour, totaling 720 data points) forms the actual time-series sequence. The subscripts correspond to time nodes (from morning to night); simultaneously, prediction data from the baseline life model for the same period are extracted, based on each time node. The value is used to deduce the predicted basic remaining life at the corresponding time. To form a predicted time series This ensures that the actual data and the predicted data are fully aligned in the time dimension, eliminating the impact of time deviation on residual calculation; Step S2.2, Residual Sequence Extraction and Preprocessing: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: , For the first The residuals (range) at each time point ), For the first The basic remaining lifetime prediction value at each time point. For the first The life reference values ​​for each time point are derived from the real-time state of cable insulation (by...). Calculated based on the aging mechanism of insulation materials (unit: years). The residuals are normalized. Mapped to Range, excluding extreme outliers (such as those caused by sensor malfunctions). The data points are replaced with the mean of the residuals of the three adjacent nodes, ultimately forming a smoothed residual sequence. This ensures that the residual data accurately reflects the aging fluctuation pattern. Step S2.3, Deployment of the Improved Sliding Window Algorithm: An improved sliding window algorithm with a weight decay factor is adopted (to optimize the problem of the traditional fixed window's lag response to sudden changes). The window size is set to 7 days (168 data points), the window sliding step is 1 hour, and a time-series weight decay factor is introduced. (Data weights are gradually reduced from recent to distant data to increase the sensitivity of recent data to mutation identification); the weighted residual mean within each window is calculated as follows: ,in For the current window The weighted residual mean, The time-series weight decay factor is 0.9. Data within the window and the window's end time node The time interval (unit: hours, value from 0 to 167). For the first The residual values ​​at each time point are calculated as follows: the numerator is the sum of weighted residuals within the window, and the denominator is the sum of weights (used for normalization to ensure that the mean range is consistent with the residuals). For the current window, This is the window's termination time point; Step S2.4, Aging Mutation Point Determination and Verification: Constructing mutation determination logic based on the weighted residual mean: The first step is to calculate the difference between the weighted residual mean values ​​of two adjacent windows. ,in For the first The weighted residual mean of each window, For the first The weighted residual mean of each window; The second step sets the mutation threshold to 0.05 (determined through fitting multiple sets of cable aging experiments; this threshold balances the accuracy of mutation identification with the false alarm rate). At that time, it was initially identified as a potential aging mutation point; The third step combines the original... Time series data verification, if the corresponding time node The increase is ≥0.1 (i.e., the aging level crosses a grade trend), and the associated electrical / environmental factors (such as...) Sudden growth If there are abnormal fluctuations (such as sudden increases), then the node is confirmed as an aging mutation point to avoid misjudgment caused by temporary environmental fluctuations. Step S2.5, Mutation Information Marking and Output: For confirmed aging mutation points, mark three core pieces of information: first, the mutation time (accurate to the hour); second, the magnitude of the mutation impact (…). Values ​​and corresponding values The third is mutation-related factors (such as "a sudden increase of 30% in partial discharge, accompanied by a 5°C increase in ambient temperature, triggering aging mutations"), which synchronize the labeled mutation information to subsequent lifetime correction steps, providing a basis for targeted adjustments to lifetime prediction results.

[0026] Step S3, Dynamic Residual Correction and Lifetime Adjustment: For both normal environmental fluctuations and abrupt aging changes, lifetime adjustment is performed separately to ensure prediction accuracy, as detailed below: Conventional environmental fluctuation correction: A linear residual correction formula is used to adjust the baseline lifetime, adapting to stable fluctuations in environmental parameters such as temperature and humidity. ,in The remaining lifespan (in years) is adjusted for environmental fluctuations. Based on the remaining lifetime, n is the amount of time-series data (here n=720, corresponding to hourly data for the past 3 months). This is the sum of the residuals at 720 time points. This represents the mean of the residuals (reflecting the overall impact of environmental fluctuations on the prediction results).

[0027] Aging mutation correction: For identified aging mutation points, a mutation correction factor is introduced. Adjust lifespan using the following formula: ,in The final revised remaining lifetime (in years). This refers to the lifespan after environmental fluctuation correction. The mutation correction factor (value 1.2-1.5, adaptively adjusted according to the mutation magnitude) is used. hour , hour (Linearly increasing) The influence of sudden changes on lifespan decay is amplified by the coefficient to ensure that the prediction results are close to the actual aging state; at the same time, based on the corrected lifespan data, an aging decay trend curve is plotted, with time as the horizontal axis and remaining lifespan as the vertical axis, to intuitively present the lifespan change pattern. Step S4, Lifetime Warning Threshold Setting: Based on the corrected remaining lifetime Three warning thresholds are set: Level 1 warning is... In 2023, a routine maintenance reminder was issued; a Level 2 warning was given. In [year], it was suggested to strengthen monitoring; Level III warning was [issued]. The system will prompt immediate replacement upon completion of the current lifespan, and output the final lifespan prediction result and warning level.

[0028] Early warning output module: Based on the aging level of the improved multi-dimensional coupled aging assessment module and the lifespan results of the optimized lifespan decay prediction module, it outputs three levels of early warning information according to preset thresholds (Level 1: Remaining lifespan ≥ 5 years, prompting routine maintenance; Level 2: 2 years ≤ Remaining lifespan < 5 years, prompting enhanced monitoring; Level 3: Remaining lifespan < 2 years, prompting immediate replacement). The early warning information is pushed through multiple channels such as pop-up windows on the maintenance platform, SMS, and audible and visual alarms, and core influencing factors are displayed simultaneously to support maintenance decisions. The data storage module retains multi-dimensional parameter data collected by the data acquisition module, evaluation and prediction results from the improved multi-dimensional coupled aging assessment module and the optimized lifespan decline prediction module, as well as full-process operation data. The retention period is ≥5 years. This provides historical data support for the coupling coefficient and correction parameter iterative optimization of the improved multi-dimensional coupled aging assessment module and the optimized lifespan decline prediction module, while ensuring the traceability of the system's operation trajectory. A brief description of this module's operation is provided below: Data Classification, Reception, and Verification: A data receiving interface is established, synchronously connecting to the data acquisition module, the improved multi-dimensional coupled aging assessment module, the optimized lifespan decline prediction module, and the early warning output module. Data is received according to "data source + data type." ① Classification Rules: Data is divided into three categories—raw acquired data (including raw sensor signals, standardized data, etc.) and standardized data. / / / Parameters and timing markers), core module operation data (coupling matrix, coupling coefficients) / Dynamic weights Values, residual sequences, correction coefficients ① Results and operational data (aging level, remaining life prediction value, early warning record, module operation log); ② Receiving verification: Perform integrity verification (check data fields and timestamp consistency) and legality verification (remove data with format errors and abnormal check codes) on each batch of received data. If the verification passes, the storage process is triggered. If it fails, the corresponding module is notified to request retransmission to ensure that the stored data is accurate and reliable.

[0029] Tiered Storage and Format Standardization: A tiered storage architecture of "edge caching + cloud archiving" is adopted to balance real-time access and long-term retention needs. ① Edge Caching Layer: Stores high-frequency time-series data from the past three months (such as hourly collected parameters, etc.). The system employs Redis caching format to ensure that the time-series data extraction latency of the optimized lifetime decay prediction module is ≤2s, adapting to the real-time computation requirements of the sliding window algorithm; ② Cloud archiving layer: All data is archived for a long period of ≥5 years, using encrypted Parquet format for compressed storage (compression ratio ≥5:1), and a directory structure is established according to "year-month-day". The system associates unique identifiers of the data (including device number, timestamp, and data type code). At the same time, core computational parameters (coupling coefficient, correction coefficient) are indexed separately for easy querying and retrieval.

[0030] Data Association Labeling and Version Management: Add multi-dimensional association tags to each batch of data to build a full-link association relationship of "data-operation-result". ① Association Labeling: Bind the original collected data with the core module operation data and result data at the corresponding time (e.g., a certain hour). / Data and corresponding coupling coefficients , ① Correlate the values ​​and remaining life prediction results, and simultaneously label the associated cable equipment number, sensor location, and module operating version; ② Version management: For parameter iterations (such as updating the fitting value of the coupling coefficient and optimizing the adjustment range of the correction coefficient β) of the improved multi-dimensional coupled aging assessment module and the optimized life decay prediction module, retain the parameter version according to the iteration timestamp, record the reason for parameter change, the values ​​before and after the change, and the corresponding assessment / prediction effect, and ensure that parameter iteration is traceable.

[0031] Iterative support for on-demand data push: A data call interface is built to push historical data on demand based on the iterative needs of the two innovative modules. ① For the improved multi-dimensional coupled aging assessment module: Coupling coefficients and dynamic weights under different operating conditions are pushed periodically. The values ​​correspond to data that support coupling matrix optimization and timing correction coefficients. ① Iterative fitting with the weighted baseline values ​​improves the accuracy of coupled evaluation; ② For the optimized lifetime decay prediction module: push residual sequences, mutation records, and correction coefficients for the past 1-5 years. The prediction error data provides data support for adjusting the sliding window size, calibrating the mutation threshold, and optimizing the lifetime correction formula, enabling adaptive upgrades of the module.

[0032] Traceability Service and Security Management: Provides end-to-end data traceability while ensuring data storage security. ① Traceability Service: Supports querying corresponding data by device number, time range, and data type, and can trace the generation chain of a certain aging level / lifespan prediction result (from raw data collection → coupled calculation → ...). ① Calculation → Lifetime Correction): Provides a basis for maintenance personnel to troubleshoot anomalies and verify module reliability; ② Security Control: Encrypt stored data (transmission uses SSL protocol, storage uses AES-256 encryption), set operation permission levels (maintenance personnel can only query, administrators can export backups), and perform data backup and integrity verification regularly (monthly) to prevent data loss or tampering and ensure data security and compliance.

[0033] This invention relates to the field of cable operation and maintenance, aiming to solve the problems of traditional cable insulation assessment neglecting multi-factor coupling and low accuracy of lifespan prediction, and to provide a precise early warning system. This invention includes modules for data acquisition, improved multi-dimensional coupled aging assessment, optimized lifespan decay prediction, early warning output, and data storage. It collects multi-dimensional electrical and environmental parameters through sensors, quantifies the aging level through coupled modeling and dynamic weight calculation, predicts the remaining lifespan based on an improved Arrhenius model and time-series residual correction, and outputs three levels of early warning through multiple channels. This invention improves the accuracy of aging assessment and lifespan prediction, supports cable operation and maintenance decisions, reduces fault risks, and is applicable to power system cable insulation condition monitoring scenarios.

[0034] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A cable insulation aging degree assessment and remaining life early warning system, characterized in that, The system includes a data acquisition module, an improved multi-dimensional coupled aging assessment module, an optimized lifespan decay prediction module, an early warning output module, and a data storage module. The data acquisition module uses a high-frequency partial discharge sensor, a dielectric loss tester, and temperature and humidity sensors to collect real-time electrical and environmental parameters of the cable, obtaining multi-dimensional parameter data. The sampling frequency is once per hour, and the data transmission delay is ≤10 seconds, ensuring data real-time performance and integrity. This provides accurate input data for the two core innovative modules. The cable's electrical parameters include partial discharge levels. and dielectric loss tangent Environmental parameters include temperature and humidity ; The improved multi-dimensional coupled aging assessment module adopts an improved weighted adaptive coupling algorithm to model the coupling relationship, adaptively calculate the weights, and calculate the coupled aging index of the multi-dimensional parameter data transmitted by the data acquisition module. It quantifies the degree of insulation aging and outputs four aging levels: mild, moderate, severe, and near failure. The optimized lifespan degradation prediction module employs an improved time-series residual correction algorithm. Based on the aging level output by the improved multi-dimensional coupled aging assessment module, it constructs an improved Arrhenius basic lifespan model. Through time-series residual extraction, mutation identification, and dynamic correction, it predicts the remaining service life and degradation trend of the cable. The early warning output module outputs three levels of early warning information based on the aging level of the improved multi-dimensional coupled aging assessment module and the lifespan result of the optimized lifespan decay prediction module, according to a preset threshold. The information is pushed through multiple channels, including pop-up windows on the operation and maintenance platform, SMS messages, and audible and visual alarms, and core influencing factors are displayed simultaneously to support operation and maintenance decisions. The data storage module retains the multi-dimensional parameter data collected by the data acquisition module, the evaluation and prediction results of the improved multi-dimensional coupled aging assessment module and the optimized life decay prediction module, as well as the full-process operation data. The retention period is ≥5 years. This provides historical data support for the coupling coefficient and correction parameter iterative optimization of the improved multi-dimensional coupled aging assessment module and the optimized life decay prediction module, while ensuring that the system operation trajectory is traceable.

2. The cable insulation aging degree assessment and remaining life early warning system according to claim 1, characterized in that: The specific implementation steps of the improved multi-dimensional coupled aging assessment module are as follows: Step 1: Acquisition and Standardization of Evaluation Factors: Obtain four core evaluation factors from the data acquisition module, namely, partial discharge quantity. Dielectric loss tangent Ambient temperature Ambient humidity All standardized to Intervals, eliminating the influence of dimensions; where Reflects the degree of internal defects in the insulation. Reflects the degradation of the insulating dielectric properties. , As an environmental coupling factor; Step 2, Coupling Relationship Modeling and Adaptive Weight Calculation: Construct the "Electrical Factor-Environmental Factor" coupling matrix, and calculate the coupling strength between factors using an improved grey relational analysis algorithm, letting... and Coupling coefficient , and Coupling coefficient , and The coupling coefficient is Coupling coefficient range The larger the value, the stronger the coupling effect; The weights are dynamically allocated based on coupling strength: the basic weight for the electrical factor is 0.6, and the basic weight for the environmental factor is 0.4, then adjusted according to the coupling coefficient. , , , This ensures that the weights dynamically adapt to the coupling relationship; Step 3, Calculation of Coupled Aging Index: The aging index is calculated using an improved coupled weighted summation formula. : At the same time, a coupling correction term is introduced. Final aging index The range of values ​​is ; Step 4, Aging Level Determination: According to The aging process is divided into four levels: mild aging (0-0.3), moderate aging (0.3-0.6), severe aging (0.6-0.9), and near-failure (0.9-1.2), providing precise aging levels and core influencing factors.

3. The cable insulation aging degree assessment and remaining life early warning system according to claim 2, characterized in that: The specific implementation steps for step 2, coupling relationship modeling and adaptive weight calculation, are as follows: Step 2.1: Construct the electrical factor-environmental factor coupling matrix: Based on the obtained four types of core factors, construct the coupling matrix according to the interaction logic of electrical factors and environmental factors, and clarify the coupling relationships that need to be quantified as " and "" and "In the coupling matrix, the coupling coefficients are related to electrical factors and environmental factors, which are the core aging impact mechanisms. The first group of couplings..." Partial discharge quantity temperature High temperature amplification The second set of couplings is affected by the breakdown of the insulation. Dielectric loss tangent humidity High humidity exacerbates The rise in temperature accelerates insulation aging. Step 2.2: Calculate the coupling coefficient using the improved grey relational analysis algorithm: The improved grey relational analysis algorithm is used to calculate the coupling coefficient between the two sets of coupled relationships. , , specific operations: Data preprocessing: Extract time series data for the past 30 days, including hourly data. , , , Standardized values ​​result in two sets of time series: and , and ; Correlation calculation: Introducing time series correction coefficient Calculate the grey relational degree for each sequence. The formula is: For the first Class of electrical factors in the first Time series values ​​at each time point, For the first Environmental factors in the first Time series values ​​at each time point, As a time node, The minimum absolute difference between the two sets of sequences. The maximum absolute difference between the two sets of sequences; Coupling coefficient transformation: transforming the correlation coefficient Normalization to The coupling coefficient is obtained from the interval. , ,Right now The larger the value, the stronger the coupling effect between the two sets of factors; Step 2.3, Adaptive Weight Allocation Based on Coupling Coefficient: The weights of the four types of factors are dynamically allocated using a combination of basic weights and coupling corrections to ensure that the weights match the strength of the coupling effect. Specific rules are as follows: Basic weight setting: Combining insulation aging mechanism and electrical factor , As the core influencing factor, the total basic weight is 0.6; Environmental factors , As a coupling influencing factor, the total basic weight is 0.4, and the basic weight is evenly distributed among the two types of factors. Coupling correction calculation: The weights are adjusted based on the coupling coefficient. The correction formula is as follows: Parameter description: Partial discharge quantity The dynamic weight, 0.3 is The basic weights, For temperature and The coupling coefficient; Parameter description: The tangent of the dielectric loss angle The dynamic weight, 0.3 is The basic weights, For humidity and The coupling coefficient; Parameter description: For temperature The dynamic weight, 0.2 is The basic weights, For temperature and The coupling coefficient; Parameter description: Humidity The dynamic weight, 0.2 is The basic weights, For humidity and The coupling coefficient; Weight normalization calibration: After correction, the four weights are summed. If the sum deviates from 1, normalization is performed to ensure that the total weight sum is 1. Finally, dynamic weights that adapt to the current coupling state are output, providing an accurate basis for subsequent aging index calculation.

4. The cable insulation aging degree assessment and remaining life early warning system according to claim 1, characterized in that: The specific implementation logic of the optimized lifetime decay prediction module is as follows: Step S1: Construction of the basic lifespan model: based on the aging index By combining the improved Arrhenius equation, a basic remaining lifetime model is constructed, as shown in the following formula: ,in Based on remaining lifespan, The rated service life of the cable insulation layer (determined based on the properties of the insulation material, such as PVC cables). Year), For material property coefficients, The final aging index calculated above, This is the actual value of the ambient temperature. It is a natural constant; Step S2, Temporal Residual Extraction and Mutation Identification: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: ,in This is a life reference value derived from the real-time condition of cable insulation; the residual is normalized. Mapped to The interval is used to remove extreme outliers, ultimately resulting in a smoothed residual sequence. This ensures that the residual data accurately reflects the aging fluctuation pattern. Step S3, Dynamic Residual Correction and Lifetime Adjustment: For both normal environmental fluctuations and abrupt aging changes, lifetime adjustment is performed separately to ensure prediction accuracy, as detailed below: Conventional environmental fluctuation correction: A linear residual correction formula is used to adjust the baseline lifetime, adapting to stable fluctuations in environmental parameters such as temperature and humidity. ,in This is the remaining lifespan after adjustment for environmental fluctuations. Given the remaining lifetime, n represents the amount of time-series data. This is the sum of the residuals at 720 time points. The mean of the residuals; Aging mutation correction: For identified aging mutation points, a mutation correction factor is introduced. Adjust lifespan using the following formula: ,in This is the final revised remaining lifetime. This refers to the lifespan after environmental fluctuation correction. The mutation correction coefficient amplifies the impact of mutations on lifespan decline, ensuring that the prediction results closely match the actual aging state. At the same time, based on the corrected lifespan data, an aging decline trend curve is plotted, with time as the horizontal axis and remaining lifespan as the vertical axis, to intuitively present the lifespan change pattern. Step S4, Lifetime Warning Threshold Setting: Based on the corrected remaining lifetime Three warning thresholds are set: Level 1 warning is... In 2023, a routine maintenance reminder was issued; a Level 2 warning was given. In [year], it was suggested to strengthen monitoring; Level III warning was [issued]. The system will prompt immediate replacement upon completion of the current lifespan, and output the final lifespan prediction result and warning level.

5. The cable insulation aging degree assessment and remaining life early warning system according to claim 4, characterized in that: The specific implementation steps of the temporal residual extraction and mutation identification in step S2 are as follows: Step S2.1, Time Series Data Regularization and Alignment: Collect data from the past 3 months. Time series data, forming actual time series sequences The subscripts correspond to time nodes; simultaneously, the prediction data from the baseline life model for the same period are extracted, based on each time node. The value is used to deduce the predicted basic remaining life at the corresponding time. To form a predicted time series This ensures that the actual data and the predicted data are fully aligned in the time dimension, eliminating the impact of time deviation on residual calculation; Step S2.2, Residual Sequence Extraction and Preprocessing: For each time node Calculate the residual between the basic predicted value and the theoretically derived actual life reference value. The formula is: , For the first The residuals at each time point For the first The basic remaining lifetime prediction value at each time point. For the first The life reference value derived from the real-time state of cable insulation at each time point; The residuals are normalized. Mapped to The interval is used to remove extreme outliers, ultimately resulting in a smoothed residual sequence. This ensures that the residual data accurately reflects the aging fluctuation pattern. Step S2.3, Deployment of the improved sliding window algorithm: An improved sliding window algorithm with a weight decay factor is adopted, setting the window size to 7 days and the window sliding step size to 1 hour, while introducing a time-series weight decay factor. The weighted residual mean within each window is calculated as follows: ,in For the current window The weighted residual mean, This is the time-series weight decay factor. Data within the window and the window's end time node The time interval, For the first The residual values ​​at each time point, with the numerator being the sum of the weighted residuals within the window and the denominator being the sum of the weights. For the current window, This is the window's termination time point; Step S2.4, Aging Mutation Point Determination and Verification: Constructing mutation determination logic based on the weighted residual mean: The first step is to calculate the difference between the weighted residual mean values ​​of two adjacent windows. ,in For the first The weighted residual mean of each window, For the first The weighted residual mean of each window; The second step is to set the mutation threshold to 0.

05. At that time, it was initially identified as a potential aging mutation point; The third step combines the original... Time series data verification, if the corresponding time node If the increase is ≥0.1 and the associated electrical / environmental factors show abnormal fluctuations, then the node is confirmed as an aging mutation point to avoid misjudgment caused by temporary environmental fluctuations. Step S2.5, Mutation Information Marking and Output: For the confirmed aging mutation points, mark three core pieces of information: the mutation time, the mutation impact magnitude, and the mutation correlation factor. Synchronize the marked mutation information to the subsequent lifetime correction steps to provide a basis for targeted adjustment of lifetime prediction results.

Images