A Method and System for Predicting the Entire Life Cycle of Cables Based on Convolutional Neural Networks

Abstract

This invention discloses a method and system for predicting the entire life cycle of cables based on convolutional neural networks, belonging to the field of cable life cycle prediction technology. The method includes: collecting load state data of the target cable, determining the target cable load mode, analyzing the target cable's first state index, analyzing the target cable's second state index, analyzing the target cable's comprehensive life assessment index, collecting the comprehensive life assessment index sequence of the target cable, inputting it into a convolutional neural network model, and outputting the cable's entire life cycle prediction result. This invention can improve the safety and stability of power systems. By accurately analyzing load modes, various state indices, and comprehensive life assessment indicators, it precisely grasps the health status of cables at different operating stages. Using convolutional neural networks, it accurately predicts the entire life cycle of cables, rationally plans cable maintenance, repair, and replacement schedules, avoids power outages caused by cable faults, and improves the utilization efficiency of power system resources.

Description

Technical Field

[0001] This invention relates to the field of cable life cycle prediction technology, and in particular to a method and system for cable life cycle prediction based on convolutional neural networks. Background Technology

[0002] With the rapid advancement of smart grid construction, power cables, as the core carrier of urban underground power transmission networks, directly affect power safety and the continuity of industrial production. Sudden cable failures can not only cause power outages and large-scale economic losses, but also lead to serious accidents such as fires and explosions due to short circuits and leakage. Therefore, accurate prediction of the entire cable lifecycle has become a key requirement for ensuring the safe and stable operation of the power grid.

[0003] For example, the invention patent announcement CN118709571B, which describes a method and system for predicting the remaining life of a cable considering aging trends, includes: acquiring historical load data of the cable to be predicted; processing the historical load data using a trained temperature prediction model to determine the predicted operating temperature of the cable for the next time period; determining the effective lifespan of the cable under the predicted operating temperature, combined with the aging trend of the cable; acquiring the operating time of the cable to be predicted; and determining the actual operating time of the cable under the predicted operating temperature, considering the service life loss of the cable, and combining it with the effective lifespan to obtain the predicted remaining lifespan of the cable.

[0004] For example, the invention patent announcement CN116361679B, which discloses a data-driven intelligent prediction method and system for cable life, includes: firstly, acquiring multi-dimensional time-series data from the cable's historical operation; secondly, obtaining cluster dispersion and optimization degree based on the distance characteristics of data points in iterative clustering; thirdly, obtaining a distance optimization factor for data points based on the distance characteristics between neighboring data points and cluster centers in the clustering space; and fourthly, obtaining a time-series optimization factor for data points based on the distance characteristics between their time-series nearest neighbors and cluster centers in the multi-dimensional time-series data; and finally, obtaining a distance optimization value based on the cluster dispersion, optimization degree, distance optimization factor, and time-series optimization factor, and then improving the iterative clustering based on the distance optimization value.

[0005] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0006] Current cable life prediction methods often rely on certain operating parameters of the cable. However, in actual power system operation scenarios, different degrees of load fluctuations, such as frequent load switching and large voltage jumps, will have different impacts on the cable aging process. If a general method is used to predict the life of all cables, ignoring the differences in load fluctuations faced by different cables in actual operation, the prediction results may deviate significantly from the actual remaining life of the cable, thereby increasing the potential risks of power system operation and even potentially triggering sudden cable failures. Summary of the Invention

[0007] The first aspect of this invention provides a method for predicting the entire lifecycle of cables based on convolutional neural networks, comprising the following steps:

[0008] S1: Within the preset monitoring window, collect the target cable load status data and determine the target cable load mode.

[0009] S2, Analyze the first state index of the target cable based on the target cable load mode determination result.

[0010] S3: Collect target cable life status data, analyze the target cable second status index, and combine the target cable first status index to analyze the target cable comprehensive life assessment index.

[0011] S4 collects the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combines the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into the convolutional neural network model to output the full life cycle prediction results of the cable.

[0012] A second aspect of the present invention provides a cable lifecycle prediction system based on a convolutional neural network, comprising:

[0013] The cable load mode determination module is used to collect target cable load status data within a preset monitoring window and determine the target cable load mode.

[0014] The cable first state index analysis module is used to analyze the first state index of the target cable based on the target cable load mode determination result.

[0015] The cable comprehensive life assessment index analysis module is used to collect target cable life status data, analyze the target cable's second state index, and combine it with the target cable's first state index to analyze the target cable's comprehensive life assessment index.

[0016] The cable life cycle prediction result output module is used to collect the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combine the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into a convolutional neural network model to output the cable life cycle prediction result.

[0017] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0018] 1. The cable life cycle prediction method based on convolutional neural networks provided by this invention can improve the safety and stability of power systems. By accurately analyzing load patterns, various state indicators, and comprehensive life assessment indicators, it can accurately grasp the health status of cables at different operating stages. Utilizing the powerful feature extraction and learning capabilities of convolutional neural networks, it can accurately predict the entire life cycle of cables, which helps to rationally plan cable maintenance, repair, and replacement plans, avoid power outages caused by cable faults, and improve the utilization efficiency of power system resources.

[0019] 2. This invention analyzes the first state index of the target cable based on the load mode determination results. By conducting targeted analysis for different load modes, it can accurately analyze the actual aging degree of the cable under various operating conditions. For cables with load disturbances, it can accurately assess the impact of thermo-mechanical alternating stress on their aging. For cables with stable loads, it can accurately measure the thermal-insulation degradation status, thus more accurately reflecting the current health status of the cable and providing a more reliable data foundation for subsequent comprehensive evaluation. In subsequent analysis, it can be more closely aligned with the actual situation of the cable, making the comprehensive evaluation results closer to the true state of the cable and improving the accuracy of cable life cycle prediction.

[0020] 3. By analyzing the comprehensive life assessment indicators of the target cable, this invention can comprehensively and accurately reflect the actual operating status and aging degree of the cable, providing accurate data for subsequent analysis and decision-making, preventing power supply interruptions caused by sudden failures, and improving the reliability and stability of power system operation. Attached Figure Description

[0021] Figure 1 Flowchart of the cable life cycle prediction method based on convolutional neural network provided in the embodiments of this application;

[0022] Figure 2 A schematic diagram of the structure of the cable life cycle prediction system based on convolutional neural networks provided in this application embodiment;

[0023] Figure 3 This is a flowchart illustrating a method for cable condition analysis based on cable load conditions, as described in an embodiment of this application.

[0024] Figure 4 This is a flowchart illustrating the output of the convolutional neural network model involved in the embodiments of this application. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] Reference Figure 1 As shown, the first aspect of the present invention provides a method for predicting the entire life cycle of a cable based on a convolutional neural network, comprising the following steps:

[0027] See Figure 3 The diagram shows a flowchart of a method for cable condition analysis based on cable load status according to an embodiment of this application. The load mode is determined based on the analysis results of the target cable load status data, and targeted data collection and analysis are performed based on the load mode.

[0028] S1: Within the preset monitoring window, collect the target cable load status data and determine the target cable load mode.

[0029] In this embodiment, the target cable load mode is determined, and the specific analysis method is as follows:

[0030] Based on the target cable load status data, analyze the load status indicators of the target cable.

[0031] In a specific embodiment, the load status indicators of the target cable are analyzed, and the specific analysis process is as follows:

[0032] Within the preset monitoring window, the target cable load status data is collected, including the target cable's average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps.

[0033] It should be noted that the average voltage, voltage fluctuation standard deviation, load switching frequency, and number of abrupt changes of the target cable can all be obtained directly from the SCADA system. Among them, the load switching frequency is the total number of times the load equipment connected to the cable starts, stops, and switches within the preset monitoring window, reflecting the frequency of load state changes during system operation. The number of abrupt changes is the number of times the voltage signal changes abruptly (the amplitude exceeds the set threshold) in a short period of time, reflecting the frequency of severe disturbance events caused by the load side.

[0034] Extract the ideal average voltage, reference voltage fluctuation standard deviation, reference load switching frequency, and reference large jump number stored in the database.

[0035] Extract the preset average voltage allocation coefficient, voltage fluctuation standard deviation allocation coefficient, load switching frequency allocation coefficient, and large jump number allocation coefficient from the database.

[0036] It should be noted that the values ​​of the average voltage distribution coefficient, voltage fluctuation standard deviation distribution coefficient, load switching frequency distribution coefficient, and abrupt change number distribution coefficient are all between 0 and 1, and the sum of the average voltage distribution coefficient, voltage fluctuation standard deviation distribution coefficient, load switching frequency distribution coefficient, and abrupt change number distribution coefficient is 1. When using them, the preset values ​​can be directly extracted from the database. For example, the extraction method involves constructing a one-to-one mapping set between the average voltage, voltage fluctuation standard deviation, load switching frequency, and abrupt change number and the average voltage distribution coefficient, voltage fluctuation standard deviation distribution coefficient, load switching frequency distribution coefficient, and abrupt change number distribution coefficient, respectively. When using them, the obtained average voltage, voltage fluctuation standard deviation, load switching frequency, and abrupt change number are input into the corresponding mapping set, thereby extracting the average voltage distribution coefficient, voltage fluctuation standard deviation distribution coefficient, load switching frequency distribution coefficient, and abrupt change number distribution coefficient.

[0037] Analyze the load status indicators of the target cable based on the target cable load status data.

[0038] The load status index of the target cable is a quantitative indicator of the influence of the target cable's average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps on the target cable's load status. The specific analysis process is as follows: the average voltage of the target cable is processed to deviate from its ideal value; the voltage fluctuation standard deviation, load switching frequency, and number of large jumps of the target cable are processed to differentiate from their corresponding reference values; and the results of the deviation processing and differentiation processing are coupled with the corresponding allocation coefficients to obtain the load status index of the target cable.

[0039] In a specific embodiment, the load status index of the target cable is represented as follows:

[0040]

[0041] Among them, S load The load condition index of the target cable. σ is the average voltage of the target cable. U f represents the standard deviation of voltage fluctuation in the target cable. swith N is the load switching frequency of the target cable. jump The number of large voltage jumps in the target cable, U0 is the ideal average voltage, (σ U ) vef For reference voltage fluctuation standard deviation, (f swith )vef For the reference load switching frequency, (N) jump ) vef For reference, the number of large voltage jumps is α1, which is the average voltage distribution coefficient, α2, which is the voltage fluctuation standard deviation distribution coefficient, α3, which is the load switching frequency distribution coefficient, and α4, which is the large voltage jump number distribution coefficient.

[0042] It should be noted that the average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps of the target cable are directly interrelated. For example, frequent load switching leads to voltage fluctuations, which in turn increases the voltage fluctuation standard deviation. The more severe the fluctuation, the more events exceed the jump threshold, i.e., the greater the number of large jumps. Each load switching is accompanied by transient changes, which directly increases the number of jumps. If the voltage control is stable, the voltage fluctuation standard deviation is low.

[0043] Extract the preset cable load status index thresholds from the database.

[0044] If the load condition index of the target cable is greater than the load condition index threshold, the load mode of the target cable is recorded as a load disturbance cable.

[0045] If the load condition indicators of the target cable exceed the cable load condition indicator threshold, it indicates that the target cable's operating condition is relatively unstable. Its average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps all exceed the normal range. Frequent load switching, large voltage fluctuations, and numerous large jumps will subject the cable to additional stress. This stress will accelerate the cable's aging process, increase the risk of failure, and affect the cable's service life and performance. Therefore, its load pattern is categorized as a load-disrupted cable.

[0046] If the load condition index of the target cable is less than or equal to the load condition index threshold, the load mode of the target cable is recorded as a load-stable cable.

[0047] If the load condition index of the target cable is less than or equal to the cable load condition index threshold, it indicates that the target cable is operating relatively smoothly. The load equipment connected to the cable does not start, stop, or switch frequently, voltage fluctuations are small, and the number of large jumps is also infrequent. This means that the cable is operating in a relatively stable electrical environment. Under these circumstances, the cable experiences less disturbance, and its aging process is mainly affected by conventional thermal aging and other factors. Its overall operating condition is good; therefore, its load mode is categorized as a load-stable cable.

[0048] S2, Analyze the first state index of the target cable based on the target cable load mode determination result.

[0049] In this embodiment, the first state index of the target cable is analyzed based on the target cable load mode determination result. The specific process is as follows:

[0050] Extract the load state correction factor corresponding to the load state index range of each target cable stored in the database, and map the extracted load state correction factor corresponding to the range of the load state index of the target cable, denoted as the load state correction factor of the target cable.

[0051] It is important to understand that the load condition index of the target cable comprehensively reflects the electrical stability of the cable during operation. The larger the load condition index of the target cable, the greater the disturbances such as load fluctuations and voltage changes that the cable is subjected to, and the more serious the impact on cable aging and lifespan. In order to more accurately measure the degree of aging and remaining lifespan of the cable under different load conditions, the relevant evaluation indexes need to be significantly corrected in subsequent analysis. Therefore, the larger the corresponding load condition correction factor is extracted, the more accurate the subsequent cable condition assessment based on the correction factor will be.

[0052] If the target cable load mode is a load disturbance cable, then the disturbance stress data of the target cable is collected, the disturbance stress score of the target cable is analyzed, and the disturbance stress correction score of the target cable is analyzed in combination with the load state correction factor of the target cable. The disturbance stress correction score of the target cable is recorded as the first state index of the target cable.

[0053] It should be noted that if the target cable load mode is a load disturbance cable, it means that the target cable frequently starts and stops or experiences large voltage fluctuations. In this case, thermo-mechanical alternating stress dominates aging, that is, the cable disturbance stress will affect the subsequent cable life.

[0054] In this embodiment, the disturbance stress correction score of the target cable is analyzed in the following specific process:

[0055] Within a preset time window, disturbance stress data of the target cable are collected, including the temperature rise rate, thermal cycling frequency, strain peak value, and number of start-stop cycles of the target cable.

[0056] It should be noted that the disturbance stress data of the target cable can be directly obtained from the SCADA system. The temperature rise rate refers to the rate at which the cable temperature rises within a preset time window, reflecting the degree of abrupt change in the thermal stress experienced by the cable. The thermal cycling frequency refers to the number of thermal cycles within the preset time window; the temperature rises first and then falls, and exceeding the preset temperature difference in the database is counted as one thermal cycle. The peak strain refers to the maximum mechanical stress experienced by the cable during operation within the preset time window. A large peak strain indicates that the cable is subjected to strong mechanical tension, bending, and other external forces, which, over time, can lead to metal fatigue or sheath cracking. The total number of starts and stops of the load equipment (such as motors, heaters, etc.) connected to the cable within the preset time window is also important. A higher number of starts and stops indicates that the cable is subjected to more frequent current surges and heat-cold cycles, inducing thermal-mechanical stress cycles.

[0057] It should also be noted that the temperature rise rate, thermal cycling frequency, peak strain, and number of start-stop cycles of the target cable are interrelated. A higher temperature rise rate tends to trigger more frequent thermal cycling because a rapid rise in cable temperature makes it easier for the temperature to drop during system operation, thus forming thermal cycling and increasing the frequency of thermal cycling. Simultaneously, frequent start-stop cycles lead to frequent changes in current, causing cable temperature fluctuations, altering the temperature rise rate, and also generating thermal expansion and contraction effects, further increasing the thermal cycling frequency. Moreover, larger peak strain values ​​are usually accompanied by changes in mechanical stress generated during start-stop processes. The more start-stop cycles, the more frequent the mechanical stress cycling of the cable, and the greater the likelihood of high peak strain values. These parameters work together to influence the cable's disturbance stress state and aging process.

[0058] Extract the reference temperature rise rate, reference thermal cycle frequency, reference strain peak value, and reference start-stop count stored in the database.

[0059] Extract the preset temperature rise rate allocation factor, thermal cycle frequency allocation factor, strain peak allocation factor, and start-stop number allocation factor from the database.

[0060] It should be noted that the temperature rise rate allocation factor, thermal cycle frequency allocation factor, strain peak value allocation factor, and start-stop count allocation factor all range from 0 to 1, and the sum of these factors is 1. Pre-defined values ​​can be directly extracted from the database. For example, a one-to-one mapping set can be constructed between the temperature rise rate, thermal cycle frequency, strain peak value, and start-stop count allocation factors. The obtained temperature rise rate, thermal cycle frequency, strain peak value, and start-stop count are then input into the corresponding mapping set to extract these factors.

[0061] The disturbance stress score of the target cable is analyzed based on the disturbance stress data of the target cable.

[0062] The disturbance stress score of the target cable is a quantitative indicator of the influence of the target cable's temperature rise rate, thermal cycling frequency, peak strain, and number of start-stop cycles on the disturbance stress state of the target cable. The specific analysis process is as follows: the temperature rise rate, thermal cycling frequency, peak strain, and number of start-stop cycles of the target cable are differentiated from their corresponding reference values. The results of the differentiation are then coupled with the corresponding allocation factors to obtain the disturbance stress score of the target cable.

[0063] In a specific embodiment, the disturbance stress score of the target cable is expressed as follows:

[0064]

[0065] Among them, S dis R is used to score the disturbance stress of the target cable. T f is the rate of temperature rise of the target cable. th ε is the thermal cycling frequency of the target cable. max N represents the peak strain of the target cable. ss The number of times the target cable is started and stopped, (R) T ) vef For reference temperature rise rate, (f th ) vef For the reference thermal cycling frequency, (ε max ) vef For reference strain peak value, (N) ss ) vef For reference start-stop times, β1 is the temperature rise rate allocation factor, β2 is the thermal cycle frequency allocation factor, β3 is the strain peak allocation factor, and β4 is the start-stop times allocation factor.

[0066] The disturbance stress correction score of the target cable is analyzed based on the disturbance stress score of the target cable and the load state correction factor of the target cable.

[0067] The specific analysis process of the disturbance stress correction score of the target cable is as follows: the disturbance stress score of the target cable is corrected based on the load state correction factor of the target cable to obtain the disturbance stress correction score of the target cable. Specifically, the disturbance stress score of the target cable is multiplied by the load state correction factor of the target cable to obtain the disturbance stress correction score of the target cable.

[0068] If the target cable load mode is a load-stable cable, then obtain the thermal-insulation status data of the target cable, analyze the comprehensive thermal-insulation degradation index of the target cable, and combine the load status correction factor of the target cable to analyze the thermal-insulation degradation correction index of the target cable. The thermal-insulation degradation correction index of the target cable is recorded as the first status index of the target cable.

[0069] It should be noted that if the target cable load mode is a stable load cable, it means that the cable is under constant or slowly changing load for a long time. The cable is mainly thermally aged, and the insulation material gradually deteriorates. In other words, the thermal-insulation state of the cable affects the subsequent life of the cable.

[0070] In a specific embodiment, the thermal-insulation degradation correction index of the target cable is analyzed as follows:

[0071] Acquire thermal-insulation status data of the target cable, including the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable.

[0072] It should be noted that the thermal-insulation status data of the target cable can be obtained directly from the SCADA system. Among them, the running time of the target cable refers to the cumulative running time of the target cable, the conductor resistivity change rate is the rate of change of the resistivity per unit length of the cable conductor relative to the initial value, and the operating voltage deviation rate is the relative deviation between the actual operating voltage and the rated voltage of the cable.

[0073] Extract the reference runtime, reference conductor resistivity change rate, and reference operating voltage offset rate stored in the database.

[0074] Extract the preset runtime allocation factor, conductor resistivity change rate allocation factor, and operating voltage offset rate allocation factor from the database.

[0075] It should be noted that the values ​​of the runtime allocation factor, conductor resistivity change rate allocation factor, and operating voltage offset rate allocation factor are all between 0 and 1, and the sum of the runtime allocation factor, conductor resistivity change rate allocation factor, and operating voltage offset rate allocation factor is 1. When using these factors, pre-defined values ​​can be directly extracted from the database. For example, a one-to-one mapping set can be constructed between the runtime, conductor resistivity change rate, and operating voltage offset rate and the runtime allocation factor, respectively. When using these factors, the obtained runtime, conductor resistivity change rate, and operating voltage offset rate are input into the corresponding mapping set to extract the runtime allocation factor, conductor resistivity change rate allocation factor, and operating voltage offset rate allocation factor.

[0076] Based on the thermal-insulation status data of the target cable, the comprehensive index of thermal-insulation degradation of the target cable is analyzed.

[0077] The thermal-insulation degradation comprehensive index of the target cable is a quantitative indicator of the degree of influence of the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable on the thermal-insulation degradation state of the target cable. The specific analysis process is as follows: the absolute values ​​of the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable are differentiated from their corresponding reference values. The results of the differentiation are then coupled with the corresponding allocation factors to obtain the thermal-insulation degradation comprehensive index of the target cable.

[0078] In a specific embodiment, the comprehensive thermal-insulation degradation index of the target cable is expressed as follows:

[0079]

[0080] Among them, I deg R is the thermal-insulation degradation composite index of the target cable, t is the operating time of the target cable, and |R| is the thermal-insulation degradation composite index.ρ | represents the rate of change of conductor resistivity of the target cable, R U t represents the operating voltage deviation rate of the target cable. vef For reference runtime, (R ρ ) vef As a reference conductor resistivity change rate, (R) U ) vef The reference operating voltage offset rate is defined by γ1, which is the operating duration allocation factor, γ2, which is the conductor resistivity change rate allocation factor, and γ3, which is the operating voltage offset rate allocation factor.

[0081] It should be noted that the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable are interrelated. As the operating time increases, the cable conductor is subjected to long-term effects such as current heating and chemical corrosion, which often leads to an increase in the conductor resistivity change rate. This is because the conductor material gradually deteriorates, increasing resistance. The operating voltage deviation rate also affects this. If the operating voltage is consistently high, it will increase conductor heating, accelerate material aging, and make the resistivity change rate rise faster. Conversely, if the operating voltage is consistently low, although heating is reduced, it may affect the stability of cable performance and also affect conductor resistivity change to some extent. In addition, as the operating time increases, the cable insulation performance decreases, making it more susceptible to voltage fluctuations, and the operating voltage deviation rate may change more frequently. These parameters work together to affect the thermal-insulation degradation state of the cable.

[0082] Based on the thermal-insulation degradation comprehensive index of the target cable and the load condition correction factor of the target cable, the thermal-insulation degradation correction index of the target cable is analyzed.

[0083] The specific analysis process of the thermal-insulation degradation correction index of the target cable is as follows: the thermal-insulation degradation comprehensive index of the target cable is corrected based on the load state correction factor of the target cable to obtain the thermal-insulation degradation correction index of the target cable. Specifically, it is expressed as: the result of multiplying the thermal-insulation degradation comprehensive index of the target cable by the load state correction factor of the target cable is used as the thermal-insulation degradation correction index of the target cable.

[0084] S3: Collect target cable life status data, analyze the target cable second status index, and combine the target cable first status index to analyze the target cable comprehensive life assessment index.

[0085] In this embodiment, the comprehensive life assessment indicators of the target cable are analyzed. The specific analysis process is as follows:

[0086] Analysis of the second state index of the target cable based on the life status data of the target cable.

[0087] In a specific embodiment, the second state index of the target cable is analyzed, and the specific analysis steps are as follows:

[0088] Collect target cable life status data within a preset time window.

[0089] The target cable life status data includes the target cable's maximum discharge quantity, discharge frequency, and dielectric loss angle change rate.

[0090] It should be noted that the maximum discharge quantity, discharge frequency, and dielectric loss angle change rate of the target cable can be obtained directly from the SCADA system. The maximum discharge quantity refers to the maximum charge of a single partial discharge within a preset time window, the discharge frequency refers to the number of discharges within the preset time window, and the dielectric loss angle change rate refers to the rate of change of the loss tangent of the cable insulation dielectric over time.

[0091] It should be noted that the maximum discharge quantity, discharge frequency, and dielectric loss angle change rate of the target cable are interrelated. When defects occur in the internal insulation of the cable, an increase in discharge frequency means that local discharge activity at the insulation defect becomes more frequent. Each discharge consumes a certain amount of energy, leading to an increase in dielectric loss and thus a larger dielectric loss angle change rate. Simultaneously, frequent discharges may gradually expand the scope and severity of the insulation defect, allowing subsequent discharges to release more energy and resulting in a larger maximum discharge quantity. Conversely, a larger maximum discharge quantity also indicates a more severe insulation defect, which often encourages discharges to occur more frequently, thereby increasing the discharge frequency, further exacerbating dielectric loss, and causing the dielectric loss angle change rate to continue to rise.

[0092] Extract the reference maximum discharge quantity, reference discharge frequency, and reference dielectric loss angle change rate stored in the database.

[0093] Extract the preset maximum discharge quantity allocation coefficient, discharge frequency allocation coefficient, and dielectric loss angle change rate allocation coefficient from the database.

[0094] It should be noted that the values ​​of the maximum discharge quantity allocation coefficient, discharge frequency allocation coefficient, and dielectric loss angle change rate allocation coefficient are all between 0 and 1, and the sum of the maximum discharge quantity allocation coefficient, discharge frequency allocation coefficient, and dielectric loss angle change rate allocation coefficient is 1. When using them, the pre-set values ​​can be directly extracted from the database. The specific extraction method is as follows: construct a one-to-one mapping set between the maximum discharge quantity, discharge frequency, and dielectric loss angle change rate and the maximum discharge quantity allocation coefficient, discharge frequency allocation coefficient, and dielectric loss angle change rate allocation coefficient, respectively. When using them, input the obtained maximum discharge quantity, discharge frequency, and dielectric loss angle change rate into the corresponding mapping set, thereby extracting the maximum discharge quantity allocation coefficient, discharge frequency allocation coefficient, and dielectric loss angle change rate allocation coefficient.

[0095] Analysis of the second state index of the target cable based on the life status data of the target cable.

[0096] The second state index of the target cable is a quantitative data on the degree of influence of the target cable's maximum discharge quantity, discharge frequency, and dielectric loss angle change rate on the target cable's state. The specific analysis process is as follows: the reference value of the target cable's maximum discharge quantity is compared with the maximum discharge quantity, the discharge frequency and dielectric loss angle change rate of the target cable are compared, and the results of the comparison are coupled with the corresponding allocation coefficients to obtain the second state index of the target cable.

[0097] In a specific embodiment, the second state index of the target cable is specifically represented as follows:

[0098]

[0099] Among them, S it E is the second state indicator of the target cable. ins f is the maximum discharge quantity of the target cable. PD Let δ be the discharge frequency of the target cable, and δ be the rate of change of the dielectric loss angle of the target cable. ins ) vef For reference to the maximum discharge amount, (f PD ) vef As a reference discharge frequency, δ vef As a reference, the rate of change of the medium loss angle, This is the maximum discharge distribution coefficient. This is the discharge frequency allocation coefficient. This is the coefficient for the rate of change of the dielectric loss angle.

[0100] Extract the first state correction index corresponding to the first state index interval of each target cable stored in the database, and map the extracted first state correction index corresponding to the interval where the first state index of the target cable is located, and denot it as the first state correction index of the target cable.

[0101] It's important to understand that a larger first-state index for the target cable indicates a more significant impact of aging and lifespan factors on the cable's current condition. For example, a larger first-state index for a cable under load disturbance signifies a higher degree of damage from the disturbance stress; conversely, for a cable under stable load, it indicates relatively severe thermal-insulation degradation. To more accurately assess the overall lifespan of the cable, when analyzing the comprehensive life assessment indices, for cables more severely affected, the subsequent correction of the second-state indices based on the first-state correction indices requires a more significant adjustment to the assessment results. Therefore, the larger the extracted first-state correction index for the target cable, the more accurately the comprehensive lifespan assessment indices can reflect the actual condition of the cable, thereby improving the accuracy of the cable's full lifespan prediction.

[0102] The comprehensive life assessment index of the target cable is analyzed based on the second state index and the first state correction index of the target cable.

[0103] The specific analysis method for the comprehensive life assessment index of the target cable is as follows: the second state index of the target cable is corrected based on the first state correction index of the target cable to obtain the comprehensive life assessment index of the target cable. Specifically, the comprehensive life assessment index of the target cable is obtained by multiplying the second state index of the target cable by the first state correction index of the target cable.

[0104] See Figure 4 The diagram shown is a flowchart of the output of the convolutional neural network model involved in this application embodiment. Specifically, it involves: setting the collection interval and number of times for comprehensive life assessment indicators; forming a sequence by collecting indicators multiple times according to the set parameters; simultaneously acquiring the time features corresponding to these indicators; integrating and reshaping the indicator sequence and time features into a time series input tensor; and inputting this tensor into the convolutional neural network model. The model extracts local features through convolutional layers, reduces data dimensionality through pooling layers, and finally outputs the prediction result through a fully connected layer, thereby obtaining the prediction of the cable's entire life cycle.

[0105] S4 collects the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combines the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into the convolutional neural network model to output the full life cycle prediction results of the cable.

[0106] In this embodiment, the specific analysis steps for predicting the full life cycle of the output cable are as follows:

[0107] A1 collects and analyzes the comprehensive life assessment index of the target cable multiple times at fixed time intervals to obtain the comprehensive life assessment index sequence of the target cable. The time interval and number of collections are preset in the database.

[0108] A2. Obtain the collection period corresponding to the comprehensive life assessment index sequence of the target cable, and denote it as the time feature of the comprehensive life assessment index of the target cable. Integrate the comprehensive life assessment index sequence of the target cable and the time feature of the comprehensive life assessment index of the target cable, and use the time feature as an additional channel to splice it with the original index sequence. Reshape the integrated data into a tensor form suitable for the convolutional neural network model, thereby obtaining the time series input tensor.

[0109] A3 inputs the time series input tensor into the convolutional neural network model. The convolutional neural network model extracts and learns features from the input data through convolutional layers, pooling layers, and fully connected layers, and outputs the prediction results of the cable's entire life cycle.

[0110] Convolutional layers extract local features by sliding convolutional kernels of different sizes across the input tensor. Pooling layers reduce data dimensionality and computational cost while preserving important features. Fully connected layers integrate the extracted features and output a prediction of the cable's entire lifecycle.

[0111] It should be noted that the full life cycle prediction results for cables include the cable's remaining lifespan and failure risk level.

[0112] See Figure 2 As shown, a second aspect of the present invention provides a cable lifecycle prediction system based on a convolutional neural network, comprising:

[0113] The cable load mode determination module is used to collect target cable load status data within a preset monitoring window and determine the target cable load mode.

[0114] The cable first state index analysis module is used to analyze the first state index of the target cable based on the target cable load mode determination result.

[0115] The cable comprehensive life assessment index analysis module is used to collect target cable life status data, analyze the target cable's second state index, and combine it with the target cable's first state index to analyze the target cable's comprehensive life assessment index.

[0116] The cable life cycle prediction result output module is used to collect the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combine the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into a convolutional neural network model to output the cable life cycle prediction result.

[0117] 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.

[0118] 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 processor, 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 and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] 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.

[0120] 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.

[0121] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0122] 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 the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for predicting the entire lifecycle of cables based on convolutional neural networks, characterized in that, Includes the following steps: S1. Within the preset monitoring window, collect the target cable load status data and determine the target cable load mode. S2, Analyze the first state index of the target cable based on the target cable load mode determination result; S3, collect the target cable life status data, analyze the target cable second status index, and combine the target cable first status index to analyze the target cable comprehensive life assessment index. S4 collects the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combines the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into the convolutional neural network model to output the full life cycle prediction results of the cable. The specific analysis method for determining the target cable load mode is as follows: Based on the target cable load status data, analyze the load status indicators of the target cable; Extract the preset cable load status index thresholds from the database; If the load condition index of the target cable is greater than the load condition index threshold, the load mode of the target cable is recorded as a load disturbance cable. If the load condition index of the target cable is less than or equal to the load condition index threshold, the load mode of the target cable is recorded as a load-stable cable. The analysis process for the load condition indicators of the target cable is as follows: Within the preset monitoring window, collect target cable load status data, including the target cable's average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps; Analyze the load status indicators of the target cable based on the target cable load status data; The load status index of the target cable is a quantitative indicator of the influence of the target cable's average voltage, voltage fluctuation standard deviation, load switching frequency, and number of large jumps on the load status of the target cable. The specific analysis process is as follows: the average voltage of the target cable is processed to deviate from its ideal value; the voltage fluctuation standard deviation, load switching frequency, and number of large jumps of the target cable are processed to differentiate from their corresponding reference values; and the results of the deviation processing and differentiation processing are coupled with the corresponding allocation coefficients to obtain the load status index of the target cable. The specific process for analyzing the first state index of the target cable based on the target cable load mode determination result is as follows: Extract the load state correction factor corresponding to the load state index interval of each target cable stored in the database, and map the extracted load state correction factor corresponding to the interval of the load state index of the target cable, which is denoted as the load state correction factor of the target cable. If the target cable load mode is a load disturbance cable, then the disturbance stress data of the target cable is collected, the disturbance stress score of the target cable is analyzed, and the disturbance stress correction score of the target cable is analyzed in combination with the load state correction factor of the target cable. The disturbance stress correction score of the target cable is recorded as the first state index of the target cable. If the target cable load mode is a load-stable cable, then obtain the thermal-insulation status data of the target cable, analyze the comprehensive thermal-insulation degradation index of the target cable, and combine the load status correction factor of the target cable to analyze the thermal-insulation degradation correction index of the target cable. The thermal-insulation degradation correction index of the target cable is recorded as the first status index of the target cable. The comprehensive life assessment indicators of the target cable are analyzed, and the specific analysis process is as follows: Analysis of the second state index of the target cable based on the target cable life status data; Extract the first state correction index corresponding to the first state index interval of each target cable stored in the database, and map the extracted first state correction index corresponding to the interval where the first state index of the target cable is located, and denot it as the first state correction index of the target cable. The comprehensive life assessment index of the target cable is analyzed based on the second state index and the first state correction index of the target cable. The specific analysis method for the comprehensive life assessment index of the target cable is as follows: the second state index of the target cable is corrected based on the first state correction index of the target cable, thereby obtaining the comprehensive life assessment index of the target cable.

2. The cable lifecycle prediction method based on convolutional neural networks as described in claim 1, characterized in that: The disturbance stress correction score for the target cable is analyzed in the following process: Within a preset time window, collect disturbance stress data of the target cable, including the temperature rise rate, thermal cycling frequency, peak strain, and number of start-stop cycles of the target cable. Analysis of the disturbance stress data of the target cable to score the disturbance stress. The disturbance stress score of the target cable is a quantitative index of the influence of the temperature rise rate, thermal cycling frequency, strain peak value, and number of start-stop cycles of the target cable on the disturbance stress state of the target cable. The specific analysis process is as follows: the temperature rise rate, thermal cycling frequency, strain peak value, and number of start-stop cycles of the target cable are differentiated from their corresponding reference values, and the results of the differentiation are coupled with the corresponding allocation factors to obtain the disturbance stress score of the target cable. Based on the disturbance stress score of the target cable and the load state correction factor of the target cable, analyze the disturbance stress correction score of the target cable. The specific analysis process for the disturbance stress correction score of the target cable is as follows: the disturbance stress score of the target cable is corrected based on the load state correction factor of the target cable, thereby obtaining the disturbance stress correction score of the target cable.

3. The cable lifecycle prediction method based on convolutional neural networks as described in claim 1, characterized in that: The thermal-insulation degradation correction index of the target cable is analyzed in the following process: Acquire thermal-insulation status data of the target cable, including the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable; Based on the thermal-insulation status data of the target cable, the comprehensive index of thermal-insulation degradation of the target cable is analyzed. The thermal-insulation degradation comprehensive index of the target cable is a quantitative indicator of the degree of influence of the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable on the thermal-insulation degradation state of the target cable. The specific analysis process is as follows: the absolute values ​​of the operating time, conductor resistivity change rate, and operating voltage deviation rate of the target cable are differentiated from their corresponding reference values, and the results of the differentiation are coupled with the corresponding allocation factors to obtain the thermal-insulation degradation comprehensive index of the target cable. Based on the thermal-insulation degradation comprehensive index of the target cable and the load condition correction factor of the target cable, the thermal-insulation degradation correction index of the target cable is analyzed. The specific analysis process of the thermal-insulation degradation correction index of the target cable is as follows: the thermal-insulation degradation comprehensive index of the target cable is corrected based on the load state correction factor of the target cable, thereby obtaining the thermal-insulation degradation correction index of the target cable.

4. The cable lifecycle prediction method based on convolutional neural networks as described in claim 1, characterized in that: The specific analysis steps for the second state index of the target cable are as follows: Collect target cable life status data within a preset time window; The target cable life status data includes the target cable's maximum discharge amount, discharge frequency, and dielectric loss angle change rate. Analysis of the second state index of the target cable based on the target cable life status data; The second state index of the target cable is a quantitative data on the degree of influence of the target cable's maximum discharge quantity, discharge frequency, and dielectric loss angle change rate on the target cable's state. The specific analysis process is as follows: the reference value of the target cable's maximum discharge quantity is compared with the maximum discharge quantity, the discharge frequency and dielectric loss angle change rate of the target cable are compared, and the results of the comparison are coupled with the corresponding allocation coefficients to obtain the second state index of the target cable.

5. The cable lifecycle prediction method based on convolutional neural networks as described in claim 1, characterized in that: The specific analysis steps for predicting the full life cycle of the output cable are as follows: A1. Collect and analyze the comprehensive life assessment index of the target cable multiple times at fixed time intervals to obtain the comprehensive life assessment index sequence of the target cable. A2. Obtain the collection period corresponding to the comprehensive life assessment index sequence of the target cable, and denote it as the time feature of the comprehensive life assessment index of the target cable. Integrate the comprehensive life assessment index sequence of the target cable and the time feature of the comprehensive life assessment index of the target cable, and use the time feature as an additional channel to splice it with the original index sequence. Reshape the integrated data into a tensor form suitable for the convolutional neural network model, thereby obtaining the time series input tensor. A3 inputs the time series input tensor into the convolutional neural network model. The convolutional neural network model extracts and learns features from the input data through convolutional layers, pooling layers, and fully connected layers, and outputs the prediction results of the cable's entire life cycle.

6. A system applying the cable lifecycle prediction method based on convolutional neural networks as described in any one of claims 1-5, characterized in that, include: The cable load mode determination module is used to collect target cable load status data within a preset monitoring window and determine the target cable load mode. The cable first state index analysis module is used to analyze the first state index of the target cable based on the target cable load mode determination result. The cable comprehensive life assessment index analysis module is used to collect life status data of the target cable, analyze the second state index of the target cable, and combine the first state index of the target cable to analyze the comprehensive life assessment index of the target cable. The cable life cycle prediction result output module is used to collect the comprehensive life assessment index sequence of the target cable multiple times at fixed time intervals, and combine the time characteristics of the comprehensive life assessment index of the target cable to construct a time series input tensor, which is then input into a convolutional neural network model to output the cable life cycle prediction result.

Images