Cable life degradation trend prediction method and system based on internet of things sensing

Abstract

The application provides a cable life degradation trend prediction method and system based on Internet of Things sensing, and belongs to the field of power equipment detection. The method comprises the following steps: synchronously collecting multi-dimensional sensing data of a target cable during operation, and acquiring historical power consumption data of each power distribution area under the jurisdiction of the target cable; inputting the multi-dimensional sensing data into a pre-trained time series prediction model to obtain a preliminary degradation trend curve set representing the change of the cable health state over time; based on the historical power consumption data, the preliminary degradation trend curve set is corrected to obtain a multi-dimensional degradation trend curve set, and the multi-dimensional degradation trend curve set is weighted and fused to generate a cable life degradation trend curve. Through the prediction correction based on the historical power consumption data and the weighted fusion of the multi-dimensional degradation trend curve, the prediction accuracy of the cable life degradation trend is improved.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing, and in particular to a method and system for predicting cable life degradation trends based on Internet of Things (IoT) sensing. Background Technology

[0002] With the rapid development of power systems and the continuous growth of electricity demand, power cables, as an important component of power distribution networks, play a crucial role in ensuring the safe and stable operation of power systems by predicting their operating status and lifespan. Accurately predicting cable lifespan degradation trends can provide a basis for the maintenance and replacement of power equipment, avoiding sudden failures caused by cable aging.

[0003] Currently, cable life prediction mainly involves collecting parameters such as temperature, current, and insulation resistance during cable operation and using machine learning or deep learning algorithms to build predictive models. However, the electrical load of each zone in the power distribution network changes over time, which directly affects the load condition and aging rate of cables. Existing cable life prediction methods lack dynamic correction for the impact of changes in power consumption in different power distribution zones, resulting in insufficient accuracy in cable life prediction. Summary of the Invention

[0004] This invention addresses the technical problem that existing cable life prediction methods do not fully consider the impact of dynamic changes in power distribution load and the characteristics of power consumption cycles, resulting in insufficient accuracy in cable life prediction. It provides a cable life degradation trend prediction method and system based on Internet of Things (IoT) sensing to solve this problem.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a method for predicting the cable life degradation trend based on Internet of Things (IoT) sensing, comprising: synchronously collecting multi-dimensional sensing data of a target cable during operation, and obtaining historical power consumption data of each distribution zone under the target cable; inputting the multi-dimensional sensing data into a pre-trained time-series prediction model to obtain a preliminary degradation trend curve set characterizing the cable's health status over time; based on the historical power consumption data, performing prediction correction on the preliminary degradation trend curve set to obtain a multi-dimensional degradation trend curve set; and performing weighted fusion of the multi-dimensional degradation trend curve set to generate a cable life degradation trend curve; wherein... Based on the historical electricity consumption data, the preliminary decline trend curve set is predicted and corrected to obtain a multidimensional decline trend curve set. This includes: predicting the load growth curve of the target cable based on the historical electricity consumption data, and performing a first prediction correction on the preliminary decline trend curve set based on the load growth curve to obtain a first corrected curve set; performing fitting analysis on the historical electricity consumption data to obtain the electricity consumption cycle characteristic curve of each distribution zone, mapping the electricity consumption cycle characteristic curve to the load enhancement coefficient of each distribution zone, and performing a second load correction on the first corrected curve set to obtain the multidimensional decline trend curve set.

[0007] Secondly, this invention provides a cable life degradation trend prediction system based on Internet of Things (IoT) sensing, comprising: a data synchronization acquisition module for synchronously acquiring multi-dimensional sensing data of the target cable during operation and obtaining historical power consumption data of each distribution zone under the target cable; a time-series prediction analysis module for inputting the multi-dimensional sensing data into a pre-trained time-series prediction model to obtain a preliminary degradation trend curve set characterizing the cable health status over time; a historical data correction module for predictively correcting the preliminary degradation trend curve set based on the historical power consumption data to obtain a multi-dimensional degradation trend curve set; and a weighted fusion processing module for weighted fusion of the multi-dimensional degradation trend curve set to generate a cable life degradation trend curve.

[0008] The beneficial effects of this invention are:

[0009] Multi-dimensional sensor data of the target cable during operation are collected synchronously, and historical power consumption data of each distribution zone under the target cable are obtained to provide comprehensive basic data support for cable degradation trend prediction. The multi-dimensional sensor data is input into a pre-trained time series prediction model to obtain a preliminary set of degradation trend curves representing the cable health status over time, and a basic prediction framework for cable degradation trend is initially established. Based on historical power consumption data, the preliminary degradation trend curve set is predicted and corrected to obtain a multi-dimensional degradation trend curve set. By introducing historical power consumption data, the preliminary prediction results are dynamically corrected to make the predicted trend more consistent with the changes in the actual operating environment of the cable. The multi-dimensional degradation trend curve set is weighted and fused to generate the cable life degradation trend curve. Through the weighted fusion processing of multi-dimensional data, the influence of various dimensional factors is comprehensively considered to finally obtain an accurate and reliable prediction result of cable life degradation trend.

[0010] The above technical solution effectively integrates multi-dimensional sensor data with historical electricity consumption data, solving the problem that existing cable life prediction methods lack dynamic correction of the impact of changes in electricity consumption in power distribution zones, and achieving the technical effect of accurately predicting the trend of cable life degradation. Attached Figure Description

[0011] Figure 1 A flowchart illustrating the cable life degradation trend prediction method based on Internet of Things sensing provided by this invention.

[0012] Figure 2 This is a schematic diagram of the cable life degradation trend prediction system based on Internet of Things sensing provided by the present invention.

[0013] In the attached diagram, the components represented by each number are as follows:

[0014] The system includes a data synchronization acquisition module 11, a time series prediction and analysis module 12, a historical data correction module 13, and a weighted fusion processing module 14. Detailed Implementation

[0015] 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.

[0016] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0017] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0018] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for predicting the cable life degradation trend based on Internet of Things (IoT) sensing, including:

[0019] S1. Synchronously collect multi-dimensional sensor data during the operation of the target cable, and obtain historical power consumption data of each power distribution zone under the target cable.

[0020] Specifically, the operational status of the target cable is monitored in real time. Multiple sensors deployed at key nodes of the target cable simultaneously collect multi-dimensional sensor data reflecting its operational condition. This multi-dimensional sensor data includes, but is not limited to, key physical parameters that effectively characterize cable degradation trends, such as temperature, current, voltage, insulation resistance, partial discharge, vibration, and ambient humidity. This sensor data reflects the real-time operational status and health level of the target cable from different dimensions, providing fundamental data support for subsequent lifespan prediction.

[0021] Simultaneously, historical electricity consumption data for each distribution zone under the target cable is acquired. Specifically, firstly, by analyzing the distribution network topology of the target cable, the distribution and load characteristics of the multiple distribution zones served by the target cable are determined. Then, historical electricity consumption data for each distribution zone within a preset time window is extracted from the power management system, including information such as electricity consumption trends, peak-valley electricity consumption characteristics, and load fluctuation patterns. This historical electricity consumption data can reflect the changing patterns of the load carried by the target cable and the development trend of electricity demand, providing a data foundation for subsequent prediction and correction based on load characteristics.

[0022] By simultaneously acquiring multi-dimensional sensor data and historical power consumption data, it is possible to establish a correlation between cable operating status and load demand, providing a data foundation for building a more accurate prediction of cable life degradation trends.

[0023] S2. Input the multi-dimensional sensing data into the pre-trained time-series prediction model to obtain a set of preliminary degradation trend curves characterizing the cable health status over time.

[0024] Specifically, the collected multi-dimensional sensor data is used as input and fed into a pre-trained time-series prediction model for processing and analysis. The time-series prediction model employs a multi-channel parallel architecture, with corresponding time-series prediction channels set up for different sensor data dimensions. Specifically, the time-series prediction model includes an input allocation layer, multiple time-series prediction channels, and an output layer. The input allocation layer is responsible for distributing the multi-dimensional sensor data to the corresponding time-series prediction channels according to their dimension type, achieving dimensional parallel processing. Each time-series prediction channel specifically processes a specific set of sensor data dimensions; for example, the temperature channel specifically processes temperature sensor data, the current channel specifically processes current sensor data, and the voltage channel specifically processes voltage sensor data. Through this differentiated dimensional configuration, each time-series prediction channel can specifically model and predict the impact mechanism of specific physical parameters on cable degradation.

[0025] Each time-series prediction channel independently predicts the temporal evolution of the cable's health status based on its corresponding sensor data dimension, outputting a preliminary degradation trend curve reflecting the influence of that dimension. Therefore, if sensor data from N dimensions are collected, N preliminary degradation trend curves will be generated accordingly, each reflecting the degradation characteristics of the target cable under the individual influence of the corresponding physical factor.

[0026] Subsequently, the output layer performs time axis alignment and format unification on multiple preliminary decline trend curves from various time series prediction channels to form a complete set of preliminary decline trend curves, providing a multi-dimensional analytical basis for subsequent prediction corrections.

[0027] S3. Based on the historical electricity consumption data, predict and correct the preliminary decline trend curve set to obtain a multidimensional decline trend curve set.

[0028] Specifically, the preliminary degradation trend curve set is a time-series prediction result based on current multi-dimensional sensor data, reflecting the degradation pattern of the target cable under existing operating conditions. However, the actual degradation process of the cable is also affected by load changes and power consumption patterns. With the development of power distribution networks, the load carried by cables usually shows an increasing trend, which accelerates the aging process of cables; at the same time, the power consumption cycle characteristics of different power distribution zones also affect the fatigue degradation of cables.

[0029] Therefore, historical electricity consumption data from each distribution zone is used to correct the initial degradation trend curve set. By analyzing the load development patterns and electricity consumption behavior characteristics contained in the historical electricity consumption data, the initial degradation trend curves are adjusted and optimized accordingly to better reflect the degradation characteristics of cables under actual operating conditions. After correction, a multi-dimensional degradation trend curve set considering load impact and electricity consumption characteristics is obtained, providing a more accurate and reliable prediction basis for subsequent weighted fusion.

[0030] S4. Perform weighted fusion on the set of multidimensional degradation trend curves to generate cable life degradation trend curves.

[0031] Specifically, the multidimensional decay trend curve set contains multiple decay prediction curves from different sensor data dimensions, each reflecting the decay characteristics of the cable from a specific physical perspective. Since different sensor data dimensions exhibit varying sensitivity and prediction reliability to cable decay, direct averaging would lead to decreased prediction accuracy. Therefore, a weighted fusion approach is adopted, assigning corresponding weights based on the reliability of the prediction results for each dimension. Specifically, firstly, the prediction performance of each time-series prediction channel in the time-series prediction model is evaluated, and the prediction confidence level for each channel is obtained. The prediction confidence level reflects the reliability and contribution weight of each sensor data dimension in cable decay prediction. Then, using the prediction confidence level as the fusion weight, a weighted average is calculated for each curve in the multidimensional decay trend curve set.

[0032] By employing weighted fusion based on prediction confidence, the predictive information from various sensor data dimensions can be fully utilized, while avoiding the adverse effects of low-reliability predictions on the final result. The resulting cable life degradation trend curve integrates multi-dimensional sensor information and load characteristics, exhibiting higher prediction accuracy and stability, and providing reliable technical support for cable maintenance decisions.

[0033] Furthermore, multi-dimensional sensor data of the target cable during operation are collected synchronously, and historical power consumption data of each distribution zone under the target cable are obtained, including:

[0034] S11. Obtain the power distribution topology information downstream of the target cable;

[0035] S12. Based on the power distribution topology information, determine the multiple power distribution zones under the target cable;

[0036] S13. Acquire the multi-dimensional sensing data of the target cable, and combine it with the preset acquisition window to acquire the historical power consumption data of multiple power distribution zones.

[0037] In one feasible implementation, the first step is to obtain the distribution topology information downstream of the target cable. This distribution topology information describes the structural relationships of the distribution network to which the target cable connects, including the branching of distribution lines, the connection relationships of distribution equipment at each level, the distribution locations of load nodes, and electrical connection paths, among other network topology data. This distribution topology information can be obtained from the power system's distribution automation system, geographic information system, or distribution management system, providing an accurate network structure foundation for subsequent determination of distribution zone division.

[0038] Then, based on the acquired power distribution topology information, multiple power distribution zones under the target cable are determined. Specifically, based on the branch structure and load distribution characteristics of the power distribution network, the downstream area served by the target cable is divided into several relatively independent power distribution zones according to electrical connections and load characteristics. Each power distribution zone corresponds to the power supply range of one or more distribution transformers, and has relatively independent power consumption characteristics and load variation patterns, which facilitates differentiated power consumption data analysis.

[0039] Subsequently, multiple sensor devices deployed at key monitoring points of the target cable are used to acquire multi-dimensional sensing data of the cable in real time. Simultaneously, combined with a pre-defined acquisition window, historical electricity consumption data for each distribution zone within a specified time range is extracted from the power management system or electricity consumption information collection system. The acquisition window defines the time span, sampling interval, and data granularity of the historical data, ensuring that the collected electricity consumption data has sufficient time coverage and statistical representativeness, providing reliable data support for subsequent load analysis and trend prediction.

[0040] Through the above steps, a clear correspondence was established between the target cable and each power distribution zone it serves, and the synchronous collection of sensor data and power consumption data was realized, providing a complete data foundation for subsequent correction of the degradation trend based on load characteristics.

[0041] Furthermore, the multi-dimensional sensing data is input into a pre-trained time-series prediction model to obtain a preliminary set of degradation trend curves characterizing the cable's health status over time. Prior to this, the following steps are taken:

[0042] S201. Collect a sample dataset containing historical multi-dimensional sensor data and corresponding cable health status labels;

[0043] S202. Based on the dimension combination of the historical multi-dimensional sensor data in the sample dataset, divide the sample dataset into multiple sample data groups;

[0044] S203. Construct and train multiple time-series prediction channels based on the multiple sets of sample data;

[0045] S204. Connect multiple time series prediction channels in parallel to the same input allocation layer and output layer to obtain the time series prediction model.

[0046] In a preferred embodiment, firstly, a sample dataset containing historical multi-dimensional sensor data and corresponding cable health status labels is collected. The sample dataset is constructed by collecting a large amount of historical monitoring data of cables under different operating conditions. The historical multi-dimensional sensor data includes cable operating parameters collected by various sensors at different time points, while the cable health status labels represent the actual health level or degree of degradation of the cable at the corresponding time. This sample dataset provides sufficient learning samples and supervision information for training the time-series prediction model.

[0047] Then, based on the dimensional combinations of historical multi-dimensional sensor data in the sample dataset, the sample dataset is divided into multiple sample data groups. Specifically, according to different combinations of sensor data dimensions, such as temperature dimension group, current dimension group, insulation resistance dimension group, etc., the original sample dataset is decomposed into several subsets. Each sample data group contains sensor data with a specific combination of dimensions and its corresponding health status label, providing a training data foundation for subsequently constructing specialized time-series prediction channels.

[0048] Subsequently, multiple time-series prediction channels were constructed and trained based on various sample data sets. Each time-series prediction channel specifically processes a particular sample data set and is trained using a time-series analysis algorithm suitable for that dimension's characteristics. Through this specialized training approach, each time-series prediction channel can deeply learn the mapping relationship between a specific sensor data dimension and cable degradation, forming a dedicated prediction model for different physical mechanisms. For example, the temperature sensing channel specifically learns the relationship between cable temperature changes and thermal aging degradation, the insulation resistance channel focuses on modeling the correspondence between insulation performance degradation and cable life attenuation, and the partial discharge channel emphasizes analyzing the correlation between insulation defect development and the evolution of cable health status. Each time-series prediction channel selects an appropriate time-series analysis algorithm, such as LSTM, GRU, or Transformer deep learning models, based on the characteristics of the sensor data dimension it processes, to achieve accurate modeling and prediction of cable degradation patterns under that dimension.

[0049] Subsequently, the trained time-series prediction channels are connected in parallel to the same input allocation layer and output layer to form a complete time-series prediction model. The input allocation layer is responsible for distributing the multi-dimensional input sensor data to the corresponding time-series prediction channels according to their dimensional features, while the output layer is responsible for integrating the prediction results from each channel and performing format unification processing. Through this parallel architecture design, the time-series prediction model can simultaneously leverage the prediction capabilities of multiple specialized channels, achieving multi-dimensional parallel prediction of cable degradation trends.

[0050] Furthermore, based on the multiple sets of sample data, multiple time-series prediction channels are constructed and trained respectively. Prior to this, the process also includes:

[0051] S2031. Based on the preset sample capacity constraint, traverse multiple sample data groups to perform capacity verification.

[0052] S2032. Based on the capacity verification results, extract the corresponding undercapacity sample data group and match and determine the equivalent sample data group corresponding to the undercapacity sample data group, wherein the sensor data dimension combination of the undercapacity sample data group is a proper subset of the sensor data dimension combination of the equivalent sample data group.

[0053] S2033. Iteratively and randomly extract sample data items from the equivalent sample data group, and add them to the corresponding insufficient sample data group after dimensional clipping, until the insufficient sample data group satisfies the sample capacity constraint.

[0054] S2034. Traverse multiple under-capacity sample data groups until all of them meet the sample capacity constraint.

[0055] In a preferred embodiment, firstly, based on a preset sample capacity constraint, multiple sample data sets are traversed for capacity verification. The sample capacity constraint defines the minimum number of samples required for effective training of each sample data set. The actual number of samples in each sample data set is checked one by one and compared with the sample capacity constraint to obtain a capacity verification result indicating whether each sample data set meets the capacity requirement. Then, based on the capacity verification result, insufficient sample data sets are identified and extracted, and a corresponding equivalent sample data set is determined for each insufficient sample data set. The selection principle for the equivalent sample data set is that its sensor data dimension combination includes all dimensions of the insufficient sample data set; that is, the sensor data dimension combination of the insufficient sample data set is a proper subset of the sensor data dimension combination of the equivalent sample data set. This inclusion relationship ensures that the equivalent sample data set can provide compatible supplementary samples for the insufficient sample data set through dimension pruning.

[0056] Subsequently, sample data items are iteratively and randomly extracted from the equivalent sample data set. The extracted sample data items are then dimensionally pruned, retaining only the data dimensions that match the dimension combination of the under-supplied sample data set. The processed sample data items are then added to the corresponding under-supplied sample data set. This process continues until the number of samples in the under-supplied sample data set reaches the preset sample capacity constraint.

[0057] The process iterates through all identified under-supplied sample data sets, repeating the sample supplementation process in step S2033 until all sample data sets meet the sample capacity constraint. Through sample balancing, it ensures that each time-series prediction channel obtains sufficient and consistent quality training samples during training, thereby guaranteeing the effectiveness of model training and the balance of prediction performance across channels.

[0058] Furthermore, the multi-dimensional sensing data is input into a pre-trained time-series prediction model to obtain a preliminary set of degradation trend curves characterizing the cable's health status over time, including:

[0059] S21. Input the multi-dimensional sensing data into the input allocation layer of the time-series prediction model;

[0060] S22. The input allocation layer inputs the multi-dimensional sensing data into multiple time-series prediction channels of the time-series prediction model to obtain multiple preliminary decline trend curves, wherein the input dimensions of the multiple time-series prediction channels are configured differently.

[0061] S23. The output layer of the time series prediction model aligns multiple preliminary decline trend curves and outputs them as the set of preliminary decline trend curves.

[0062] In a preferred embodiment, firstly, multi-dimensional sensor data is input to the input allocation layer of the time-series prediction model. The input allocation layer, serving as the data access interface for the time-series prediction model, receives complete multi-dimensional sensor data, including temperature data, current data, voltage data, insulation resistance data, partial discharge data, vibration data, and ambient humidity data. It then performs format normalization and preprocessing operations on the input data to prepare for subsequent dimension allocation processing.

[0063] Then, the input allocation layer, according to a preset dimensional allocation strategy, inputs the multi-dimensional sensor data into multiple time-series prediction channels of the time-series prediction model according to different dimensional combinations. Since each time-series prediction channel uses a differentiated input dimensional configuration—for example, the temperature channel only receives temperature-related data, the current channel specifically processes current-related data, and the insulation resistance channel specifically analyzes insulation resistance data—the input allocation layer needs to accurately allocate the corresponding sensor data dimensions to the appropriate prediction channels. After receiving their specific sensor data, each time-series prediction channel, based on its dimension-specific prediction capabilities acquired during training, outputs a corresponding preliminary decay trend curve.

[0064] Subsequently, the output layer of the time series forecasting model performs unified processing on the preliminary decline trend curves from multiple time series forecasting channels. Since the forecast results of each channel may differ in terms of time axis scale, numerical range, etc., the output layer first performs time axis alignment and numerical standardization on multiple preliminary decline trend curves to ensure that each curve has a consistent time base and comparable numerical scale. Then, the processed curves are integrated and output as a complete set of preliminary decline trend curves.

[0065] Through the aforementioned hierarchical processing mechanism, the time-series prediction model achieves parallel analysis and specialized prediction of multi-dimensional sensor data, obtaining a preliminary set of decay trend curves reflecting the decay characteristics of different physical dimensions, providing a multi-dimensional prediction basis for subsequent prediction correction.

[0066] Furthermore, based on the historical electricity consumption data, the preliminary decline trend curve set is predicted and corrected to obtain a multi-dimensional decline trend curve set, including:

[0067] S31. Based on the historical electricity consumption data, predict the load growth curve of the target cable, and perform a first prediction correction on the initial decline trend curve set based on the load growth curve to obtain a first correction curve set.

[0068] S32. Based on the historical electricity consumption data, fit and analyze to obtain the electricity consumption cycle characteristic curve of each distribution zone, and map the electricity consumption cycle characteristic curve to the load enhancement coefficient of each distribution zone. Perform a second load correction on the first correction curve set to obtain the multidimensional decline trend curve set.

[0069] In a preferred embodiment, firstly, the load growth curve of the target cable is predicted based on historical electricity consumption data, and a first prediction correction is made to the preliminary degradation trend curve set based on this load growth curve. Specifically, by analyzing historical electricity consumption data of each distribution zone, the load development patterns and growth trends are identified, a load growth prediction model is established, and a load growth curve reflecting future load changes of the target cable is obtained. Since increased load accelerates the aging process of the cable, the impact of load growth on the cable degradation rate is quantified as a correction parameter, and each curve in the preliminary degradation trend curve set is scaled or adjusted accordingly on the time axis to obtain a first corrected curve set that considers the impact of load growth.

[0070] Then, based on historical electricity consumption data, the electricity consumption cycle characteristic curve for each distribution zone is obtained through fitting analysis. This curve is then mapped to a load enhancement coefficient for each distribution zone, and a second load correction is applied to the first set of correction curves. Specifically, the periodic variation patterns in the historical electricity consumption data of each distribution zone are analyzed, and a electricity consumption cycle characteristic curve reflecting the characteristics of electricity consumption behavior is obtained through data fitting. Then, feature parameters of the electricity consumption cycle characteristic curve, such as periodicity, fluctuation amplitude, and frequency of change, are extracted. These feature parameters are then converted into corresponding load enhancement coefficients using a pre-trained mapping model. This load enhancement coefficient reflects the cumulative effect of periodic load changes on the cable degradation process. Afterwards, the first set of correction curves is further adjusted using the load enhancement coefficient to obtain a multi-dimensional degradation trend curve set that simultaneously considers the effects of load growth and periodic characteristics.

[0071] Through the aforementioned dual correction mechanism, the load development patterns and electricity consumption behavior characteristics contained in historical electricity consumption data are fully integrated into the decline trend prediction, thereby improving the adaptability and accuracy of the prediction results to the actual operating environment.

[0072] Furthermore, based on the historical electricity consumption data, the load growth curve of the target cable is predicted, and the initial decline trend curve set is corrected according to the load growth curve to obtain a first corrected curve set, including:

[0073] S311. Traverse multiple power distribution zones corresponding to the target cable, combine the corresponding historical power consumption data to predict load growth, and obtain the zone expansion curve, wherein the zone expansion curve is a relative value curve.

[0074] S312. Using the real-time zoning capacity ratio of multiple power distribution zones as weights, weight the expansion curves of multiple zones to obtain the load growth curve of the target cable.

[0075] S313. Calculate the integral of the load growth curve to obtain the load growth acceleration factor, and scale and correct the time axis of the initial decline trend curve set based on the load growth acceleration factor to generate the first corrected curve set.

[0076] In a preferred embodiment, firstly, multiple distribution zones corresponding to the target cable are traversed, and load growth prediction is performed based on the historical electricity consumption data of each zone to obtain the zone expansion curve. Specifically, for each distribution zone, its historical electricity consumption data is arranged in a time series to construct a training sample set with historical load data as input and future load values ​​as output. Subsequently, a time-series prediction algorithm is used to train the model on the training sample set to learn the temporal pattern and growth pattern of load changes in the zone. After training, the load growth prediction model is used to predict the future load development of each zone, and the prediction results are converted into relative values ​​based on the current load level to generate the zone expansion curve. This zone expansion curve represents the relationship between the growth rate of the load capacity of each distribution zone relative to the current moment and the change over time.

[0077] Subsequently, using the real-time capacity ratio of multiple distribution zones as weights, the expansion curves of multiple zones are weighted and fused to obtain the load growth curve of the target cable. Since different distribution zones contribute differently to the total load of the target cable, corresponding weighting coefficients are determined based on the proportion of each zone's real-time capacity to the total capacity of the target cable. Then, the expansion curves of each zone are weighted and averaged according to their respective weighting coefficients to obtain the load growth curve reflecting the overall load growth trend of the target cable.

[0078] Next, the integral value of the load growth curve is calculated to obtain the load growth acceleration factor, and the time axis of the initial degradation trend curve set is scaled and corrected based on this factor. The integral value of the load growth curve reflects the cumulative accelerating effect of load growth on cable aging, i.e., the load growth acceleration factor. This acceleration factor is used to scale and adjust the time axis of each curve in the initial degradation trend curve set. Specifically, the original time axis is compressed according to the acceleration factor, causing the degradation process to accelerate in time, thereby generating the first corrected curve set that considers the impact of load growth.

[0079] Through the above processing, load prediction at the zone level and degradation correction at the cable level are achieved, accurately quantifying the impact of load growth on cable life and incorporating it into the prediction model.

[0080] Furthermore, based on the historical electricity consumption data, a fitting analysis is performed to obtain the electricity consumption cycle characteristic curve for each distribution zone, and the electricity consumption cycle characteristic curve is mapped to the load enhancement coefficient for each distribution zone. A second load correction is then applied to the first correction curve set to obtain the multidimensional decline trend curve set, including:

[0081] S321. Taking each power distribution zone as the analysis object, traverse the historical electricity consumption data to perform fitting analysis and obtain multiple electricity consumption cycle characteristic curves.

[0082] S322. Calculate the behavioral entropy indicators of multiple electricity consumption cycle characteristic curves respectively, wherein the behavioral entropy indicators include at least time entropy and information entropy;

[0083] S323. Input the behavioral entropy indicator into the pre-trained entropy behavior correction model to obtain the load enhancement coefficient;

[0084] S324. Adjust the trend slope of the first modified curve set according to the load enhancement coefficient to obtain the multidimensional decline trend curve set.

[0085] In a preferred embodiment, firstly, taking each distribution zone as the analysis object, the corresponding historical electricity consumption data is traversed for fitting analysis to obtain multiple electricity consumption cycle characteristic curves. Specifically, the historical electricity consumption data of each distribution zone is analyzed periodically to identify electricity consumption patterns at different time scales, such as daily, weekly, and monthly cycles. Through curve fitting algorithms, various periodic electricity consumption patterns are modeled into corresponding electricity consumption cycle characteristic curves, and each distribution zone can obtain multiple characteristic curves reflecting different cycle characteristics.

[0086] Then, behavioral entropy indicators for multiple electricity consumption cycle characteristic curves are calculated separately. These behavioral entropy indicators include at least temporal entropy and information entropy. Specifically, temporal entropy is calculated based on the time-series data of the electricity consumption cycle characteristic curves. By analyzing the numerical change patterns between adjacent time points, the conditional probability distribution of the sequence is calculated to obtain an entropy value reflecting the complexity of the time series. Information entropy is calculated based on the numerical distribution characteristics of the electricity consumption cycle characteristic curves. The curve values ​​are discretized according to certain intervals, the probability distribution of each interval is statistically analyzed, and the information entropy value reflecting the uniformity of data distribution is calculated using the Shannon entropy formula. Through the above entropy calculation process, the temporal complexity and distribution characteristics of each electricity consumption cycle characteristic curve are quantified into corresponding entropy indicator values. The temporal entropy and information entropy are then combined to form a behavioral entropy indicator characterizing the electricity consumption behavior of the distribution zone.

[0087] Subsequently, the behavioral entropy indicator is input into a pre-trained entropy behavior correction model to obtain the corresponding load enhancement coefficient. The construction process of the entropy behavior correction model is as follows: A large amount of historical electricity consumption data and corresponding actual cable degradation data from different distribution zones are collected. Behavioral entropy indicators for each zone are extracted as input features, and the ratio of the actual degradation rate to the basic predicted degradation rate is calculated as the output label to construct a training sample set. Then, a machine learning regression algorithm is used to train this sample set, learning the nonlinear mapping relationship between the behavioral entropy indicator and the degradation acceleration effect, thus obtaining the entropy behavior correction model. In practical applications, the behavioral entropy indicator vector, composed of the time-series entropy and information entropy calculated for the current distribution zone, is input into the trained entropy behavior correction model. The model outputs the corresponding load enhancement coefficient based on the learned mapping relationship. This load enhancement coefficient reflects the degree of influence of specific electricity consumption cycle characteristics on the cable degradation process.

[0088] Subsequently, the trend slope of the first modified curve set is adjusted based on the obtained load enhancement coefficient to obtain a multidimensional decay trend curve set. Specifically, the load enhancement coefficient is used as a slope correction factor to amplify or reduce the decay trend slope of each curve in the first modified curve set accordingly, reflecting the impact of power consumption cycle characteristics on the cable decay rate. The adjusted curve set simultaneously considers the dual effects of load growth and power consumption cycle characteristics, forming the final multidimensional decay trend curve set.

[0089] Through the above-mentioned electricity consumption behavior analysis and correction mechanism based on entropy theory, the impact of complex electricity consumption patterns has been accurately quantified and effectively integrated, further improving the accuracy of cable degradation trend prediction.

[0090] Furthermore, the weighted fusion of the multidimensional degradation trend curve set to generate the cable life degradation trend curve also includes:

[0091] S41. Perform prediction performance verification on the time series prediction model and obtain the prediction confidence of multiple time series prediction channels;

[0092] S42. Using the prediction confidence level as the fusion weight, the multidimensional degradation trend curve set is weighted and fused to obtain the cable life degradation trend curve.

[0093] In a preferred embodiment, firstly, the prediction performance of the time-series prediction model is validated, and the prediction confidence scores of multiple time-series prediction channels are obtained. Specifically, the prediction accuracy of each time-series prediction channel is evaluated using a validation dataset. The prediction performance level of each channel is quantified by calculating error indices between the predicted and actual values, such as root mean square error and mean absolute error. Based on the performance evaluation results, a corresponding prediction confidence score is assigned to each time-series prediction channel. This prediction confidence score reflects the reliability and accuracy of the corresponding sensor data dimension in cable degradation prediction.

[0094] Then, using the prediction confidence levels of each time-series prediction channel as fusion weights, the multi-dimensional degradation trend curve set is weighted and fused to obtain the cable life degradation trend curve. Specifically, each degradation trend curve is calculated by weighting its corresponding prediction confidence level; curves with higher confidence levels receive greater weight during the fusion process, while curves with lower confidence levels have relatively smaller weights. This confidence-based weighted fusion strategy fully utilizes information from high-reliability prediction channels while reducing the impact of low-reliability prediction results on the final result. Ultimately, the output is a cable life degradation trend curve that integrates prediction information from multiple sensor data dimensions and considers the reliability differences across these dimensions. This cable life degradation trend curve exhibits higher prediction accuracy and stability.

[0095] Through the aforementioned intelligent fusion mechanism based on prediction confidence, the multi-dimensional prediction results are optimized and integrated, ensuring the accuracy and reliability of the final decline trend prediction.

[0096] Example 2, as Figure 2 As shown, based on the same inventive concept as the cable life degradation trend prediction method based on IoT sensing provided in Embodiment 1, this embodiment of the invention also provides a cable life degradation trend prediction system based on IoT sensing, including:

[0097] The data synchronization acquisition module 11 is used to synchronously acquire multi-dimensional sensor data during the operation of the target cable and obtain historical power consumption data of each power distribution zone under the target cable.

[0098] The time-series prediction analysis module 12 is used to input the multi-dimensional sensing data into a pre-trained time-series prediction model to obtain a set of preliminary degradation trend curves characterizing the cable health status over time.

[0099] The historical data correction module 13 is used to predict and correct the preliminary decline trend curve set based on the historical electricity consumption data, and obtain a multidimensional decline trend curve set.

[0100] The weighted fusion processing module 14 is used to perform weighted fusion on the multidimensional degradation trend curve set to generate a cable life degradation trend curve.

[0101] Furthermore, the execution steps of the data synchronization acquisition module 11 include:

[0102] Obtain the power distribution topology information downstream of the target cable;

[0103] Based on the power distribution topology information, determine multiple power distribution zones under the target cable;

[0104] The multi-dimensional sensing data of the target cable is acquired, and the historical power consumption data of multiple power distribution zones is acquired in combination with the preset acquisition window.

[0105] Furthermore, the execution steps of the time series prediction analysis module 12 include:

[0106] Collect a sample dataset containing historical multi-dimensional sensor data and corresponding cable health status labels;

[0107] Based on the dimensional combination of the historical multi-dimensional sensor data in the sample dataset, the sample dataset is divided into multiple sample data groups;

[0108] Based on the multiple sets of sample data, multiple time-series prediction channels are constructed and trained respectively;

[0109] Multiple time series prediction channels are connected in parallel to the same input allocation layer and output layer to obtain the time series prediction model.

[0110] Furthermore, the execution steps of the time series prediction analysis module 12 also include:

[0111] Based on the preset sample capacity constraint, the capacity is verified by traversing multiple sample data groups.

[0112] Based on the capacity verification results, the corresponding undercapacity sample data group is extracted, and the equivalent sample data group corresponding to the undercapacity sample data group is determined. The sensor data dimension combination of the undercapacity sample data group is a proper subset of the sensor data dimension combination of the equivalent sample data group.

[0113] Iteratively and randomly extract sample data items from the equivalent sample data group, and add them to the corresponding insufficient sample data group after dimensional clipping, until the insufficient sample data group satisfies the sample capacity constraint.

[0114] Iterate through multiple under-capacity sample data sets until all of them meet the sample capacity constraint.

[0115] Furthermore, the execution steps of the time series prediction analysis module 12 also include:

[0116] The multi-dimensional sensing data is input into the input allocation layer of the time-series prediction model;

[0117] The input allocation layer inputs the multi-dimensional sensing data into multiple time-series prediction channels of the time-series prediction model to obtain multiple preliminary decline trend curves, wherein the input dimensions of the multiple time-series prediction channels are configured differently.

[0118] The output layer of the time series prediction model aligns multiple preliminary decline trend curves and outputs the set of preliminary decline trend curves.

[0119] Furthermore, the execution steps of the historical data correction module 13 include:

[0120] Based on the historical electricity consumption data, predict the load growth curve of the target cable, and perform a first prediction correction on the initial decline trend curve set based on the load growth curve to obtain a first corrected curve set.

[0121] Based on the historical electricity consumption data, the electricity consumption cycle characteristic curve of each distribution zone is obtained through fitting analysis, and the electricity consumption cycle characteristic curve is mapped to the load enhancement coefficient of each distribution zone. The first set of correction curves is then subjected to a second load correction to obtain the multidimensional decline trend curve set.

[0122] Furthermore, the execution steps of the historical data correction module 13 also include:

[0123] The system iterates through multiple power distribution zones corresponding to the target cable, combines the corresponding historical power consumption data to predict load growth, and obtains the zone expansion curve, wherein the zone expansion curve is a relative value curve.

[0124] Using the real-time capacity ratio of multiple power distribution zones as weights, the load growth curve of the target cable is obtained by weighting the expansion curves of multiple zones.

[0125] Calculate the integral of the load growth curve to obtain the load growth acceleration factor, and scale and correct the time axis of the initial decline trend curve set based on the load growth acceleration factor to generate the first corrected curve set.

[0126] Furthermore, the execution steps of the historical data correction module 13 also include:

[0127] Taking each power distribution zone as the analysis object, the historical electricity consumption data is traversed for fitting analysis to obtain multiple electricity consumption cycle characteristic curves;

[0128] Calculate the behavioral entropy indicators of multiple electricity consumption cycle characteristic curves respectively, wherein the behavioral entropy indicators include at least time entropy and information entropy;

[0129] Input the behavioral entropy indicator into the pre-trained entropy behavior correction model to obtain the load enhancement coefficient;

[0130] The trend slope of the first modified curve set is adjusted according to the load enhancement coefficient to obtain the multidimensional decline trend curve set.

[0131] Furthermore, the execution steps of the weighted fusion processing module 14 include:

[0132] The prediction performance of the time series prediction model is verified, and the prediction confidence of multiple time series prediction channels is obtained.

[0133] Using the prediction confidence level as the fusion weight, the multidimensional degradation trend curve set is weighted and fused to obtain the cable life degradation trend curve.

[0134] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0135] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0137] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0139] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0140] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for predicting cable life degradation trends based on Internet of Things (IoT) sensing, characterized in that, include: Simultaneously collect multi-dimensional sensor data during the operation of the target cable, and obtain historical power consumption data of each power distribution zone under the target cable; The multi-dimensional sensing data is input into a pre-trained time-series prediction model to obtain a set of preliminary degradation trend curves characterizing the cable health status over time. Based on the historical electricity consumption data, the preliminary decline trend curve set is predicted and corrected to obtain a multidimensional decline trend curve set. The multidimensional degradation trend curve set is weighted and fused to generate a cable life degradation trend curve; Specifically, based on the historical electricity consumption data, the preliminary decline trend curve set is predicted and corrected to obtain a multi-dimensional decline trend curve set, including: Based on the historical electricity consumption data, predict the load growth curve of the target cable, and perform a first prediction correction on the initial decline trend curve set based on the load growth curve to obtain a first correction curve set. Based on the historical electricity consumption data, the electricity consumption cycle characteristic curve of each distribution zone is obtained by fitting analysis, and the electricity consumption cycle characteristic curve is mapped to the load enhancement coefficient of each distribution zone. The first correction curve set is then subjected to a second load correction to obtain the multidimensional decline trend curve set. Specifically, based on the historical electricity consumption data, the load growth curve of the target cable is predicted, and the initial decline trend curve set is corrected based on the load growth curve to obtain a first corrected curve set, including: The system iterates through multiple power distribution zones corresponding to the target cable, combines the corresponding historical power consumption data to predict load growth, and obtains the zone expansion curve, wherein the zone expansion curve is a relative value curve. Using the real-time capacity ratio of multiple power distribution zones as weights, the load growth curve of the target cable is obtained by weighting the expansion curves of multiple zones. Calculate the integral of the load growth curve to obtain the load growth acceleration factor, and scale and correct the time axis of the initial decline trend curve set based on the load growth acceleration factor to generate the first corrected curve set. Specifically, based on the historical electricity consumption data, a power consumption cycle characteristic curve for each distribution zone is obtained through fitting analysis, and the power consumption cycle characteristic curve is mapped to a load enhancement coefficient for each distribution zone. A second load correction is then applied to the first correction curve set to obtain the multidimensional decline trend curve set, including: Taking each power distribution zone as the analysis object, the historical electricity consumption data is traversed for fitting analysis to obtain multiple electricity consumption cycle characteristic curves; Calculate the behavioral entropy indicators of multiple electricity consumption cycle characteristic curves respectively, wherein the behavioral entropy indicators include at least time entropy and information entropy; Input the behavioral entropy indicator into the pre-trained entropy behavior correction model to obtain the load enhancement coefficient; The trend slope of the first modified curve set is adjusted according to the load enhancement coefficient to obtain the multidimensional decline trend curve set.

2. The cable life degradation trend prediction method based on IoT sensing as described in claim 1, characterized in that, Synchronously collect multi-dimensional sensor data during the operation of the target cable, and obtain historical power consumption data for each power distribution zone under the target cable, including: Obtain the power distribution topology information downstream of the target cable; Based on the power distribution topology information, determine multiple power distribution zones under the target cable; The multi-dimensional sensing data of the target cable is acquired, and the historical power consumption data of multiple power distribution zones is acquired in combination with the preset acquisition window.

3. The method for predicting cable life degradation trends based on IoT sensing as described in claim 1, characterized in that, The multi-dimensional sensing data is input into a pre-trained time-series prediction model to obtain a preliminary set of degradation trend curves characterizing the cable's health status over time. Prior to this, the model includes: Collect a sample dataset containing historical multi-dimensional sensor data and corresponding cable health status labels; Based on the dimensional combination of the historical multi-dimensional sensor data in the sample dataset, the sample dataset is divided into multiple sample data groups; Based on the multiple sets of sample data, multiple time-series prediction channels are constructed and trained respectively; Multiple time series prediction channels are connected in parallel to the same input allocation layer and output layer to obtain the time series prediction model.

4. The cable life degradation trend prediction method based on IoT sensing as described in claim 3, characterized in that, Based on the multiple sets of sample data, multiple time-series prediction channels are constructed and trained respectively. Prior to this, the process also includes: Based on the preset sample capacity constraint, the capacity is verified by traversing multiple sample data groups. Based on the capacity verification results, the corresponding undercapacity sample data group is extracted, and the equivalent sample data group corresponding to the undercapacity sample data group is determined. The sensor data dimension combination of the undercapacity sample data group is a proper subset of the sensor data dimension combination of the equivalent sample data group. Iteratively and randomly extract sample data items from the equivalent sample data group, and add them to the corresponding insufficient sample data group after dimensional clipping, until the insufficient sample data group satisfies the sample capacity constraint. Iterate through multiple under-capacity sample data sets until all of them meet the sample capacity constraint.

5. The cable life degradation trend prediction method based on IoT sensing as described in claim 3, characterized in that, The multi-dimensional sensing data is input into a pre-trained time-series prediction model to obtain a preliminary set of degradation trend curves characterizing the cable's health status over time, including: The multi-dimensional sensing data is input into the input allocation layer of the time-series prediction model; The input allocation layer inputs the multi-dimensional sensing data into multiple time-series prediction channels of the time-series prediction model to obtain multiple preliminary decline trend curves, wherein the input dimensions of the multiple time-series prediction channels are configured differently. The output layer of the time series prediction model aligns multiple preliminary decline trend curves and outputs the set of preliminary decline trend curves.

6. The method for predicting cable life degradation trend based on IoT sensing as described in claim 1, characterized in that, The process of weighted fusion of the multidimensional degradation trend curve set to generate cable life degradation trend curves also includes: The prediction performance of the time series prediction model is verified, and the prediction confidence of multiple time series prediction channels is obtained. Using the prediction confidence level as the fusion weight, the multidimensional degradation trend curve set is weighted and fused to obtain the cable life degradation trend curve.

7. A cable life degradation trend prediction system based on Internet of Things (IoT) sensing, characterized in that, A cable life degradation trend prediction method based on Internet of Things sensing as described in any one of claims 1 to 6, comprising: The data synchronization acquisition module is used to synchronously acquire multi-dimensional sensor data during the operation of the target cable and obtain historical power consumption data of each power distribution zone under the target cable. The time-series prediction analysis module is used to input the multi-dimensional sensing data into a pre-trained time-series prediction model to obtain a set of preliminary degradation trend curves characterizing the cable health status over time. The historical data correction module is used to predict and correct the preliminary decline trend curve set based on the historical electricity consumption data, and obtain a multi-dimensional decline trend curve set. The weighted fusion processing module is used to perform weighted fusion on the multidimensional degradation trend curve set to generate cable life degradation trend curves.

Images