A new energy power cable intelligent early warning system and method

Abstract

The application belongs to the technical field of intelligent early warning of power cables, and particularly relates to a new energy power cable intelligent early warning system and method, which comprises the following steps: collecting scene coupling parameters of a target cable at predetermined time intervals; calculating a parameter relative deviation combination and a coupling abnormality coefficient based on specific data in the scene coupling parameters; inputting the coupling abnormality coefficient and the parameter relative deviation combination into a pre-trained Bayesian causal tracing model, and outputting posterior probabilities of each failure mechanism of the target cable through a three-layer association network structure built in the model to lock a failure core cause; and comparing the coupling abnormality coefficient of the target cable with a scene-based early warning threshold to generate hierarchical early warning information containing the failure core cause. The three-layer association network structure built in the pre-trained Bayesian causal tracing model and the posterior probability difference rule can be used to accurately lock the failure core cause and improve the failure troubleshooting efficiency.

Description

Technical Field

[0001] This invention belongs to the field of intelligent early warning technology for power cables, and specifically relates to an intelligent early warning system and method for new energy power cables. Background Technology

[0002] Against the backdrop of global energy transition, new energy power generation systems such as wind power, photovoltaics, and energy storage have become an important part of the power system. Among them, new energy power cables, as the core energy transmission hub between new energy power generation systems (such as wind turbine tower cables, photovoltaic array cables, and energy storage container cables) and the public power grid, directly determine the efficiency of new energy power absorption. If the cable fails and stops operating, it will cause the new energy power generation unit to go offline, resulting in energy waste. It may also cause safety risks such as grid load fluctuations and local power outages, posing a serious threat to grid stability.

[0003] However, wind power, photovoltaic, and energy storage scenarios have unique operating environments and disturbance characteristics, such as mechanical vibration in wind power scenarios, dynamic loads in photovoltaic scenarios, and current surges in energy storage scenarios, resulting in complex cable fault mechanisms. Existing power cable early warning technologies can only identify cable anomalies but cannot pinpoint the core cause of the fault. For example, when the cable sheath temperature exceeds the standard, it is unclear whether it is due to localized heat generation caused by vibration and friction or poor heat dissipation caused by corrosion. The cable laying environment in new energy scenarios is complex. Without identifying the core cause of the fault, blindly troubleshooting will lead to low operation and maintenance efficiency, prolong fault handling time, exacerbate new energy power generation losses and grid security risks, and increase operation and maintenance costs. Summary of the Invention

[0004] In view of the above-mentioned shortcomings in the prior art, the present invention provides an intelligent early warning system and method for new energy power cables to solve the problems in the background art.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A smart early warning method for new energy power cables, applicable to wind power, photovoltaic, and energy storage laying scenarios, includes the following steps:

[0007] S1. Collect scene coupling parameters of the target cable at predetermined time intervals; the scene coupling parameters include wind power scene coupling parameters, photovoltaic scene coupling parameters and energy storage scene coupling parameters;

[0008] S2. Based on the specific data in the scene coupling parameters, calculate the relative deviation of each data point from its normal threshold to obtain the parameter relative deviation combination, and calculate the coupling anomaly coefficient of the target cable by scene adaptability weighting based on the parameter relative deviation combination.

[0009] S3. Input the coupling anomaly coefficient and parameter relative deviation combination into the pre-trained Bayesian causal tracing model, and output the posterior probability of each fault mechanism of the target cable through collaborative inference using the three-layer association network structure built into the model, thereby locking in the core cause of the fault; the three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes, and the three nodes are associated through a conditional probability table.

[0010] S4. Compare the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning threshold includes wind power scenario-based early warning threshold, photovoltaic scenario-based early warning threshold and energy storage scenario-based early warning threshold.

[0011] Furthermore, in step S1, the wind power scenario coupling parameters include at least cable sheath temperature data, cable insulation partial discharge data, cable fixing point vibration acceleration data, and H2S concentration data.

[0012] The photovoltaic scenario coupling parameters include at least cable sheath temperature data, cable insulation partial discharge data, and ambient light intensity data.

[0013] The coupling parameters for the energy storage scenario include at least cable sheath temperature data, cable insulation partial discharge data, cable charge / discharge current data, state of charge data, and relative humidity data. Further, in step S2, the formula for calculating the coupling anomaly coefficient of the target cable is: , where n is the number of specific data items for the scene coupling parameters; The scene adaptability weight coefficient for the j-th data item; Let be the relative deviation of the j-th data item from its normal threshold. , Let be the deviation between the actual collected value of the j-th data item and its normal threshold. is the normal threshold corresponding to data item j.

[0014] Furthermore, the scenario adaptability weight coefficient The determination method is as follows: a fault contribution matrix for the target cable scenario is constructed based on the historical fault statistics database of new energy power cables. Weights are assigned using the analytic hierarchy process (AHP), and the scenario adaptability weight coefficients are dynamically and iteratively calibrated quarterly based on newly added fault data using the least squares method. ;

[0015] The historical fault statistics database for new energy power cables includes historical fault statistics databases for power cables in wind power scenarios, historical fault statistics databases for power cables in photovoltaic scenarios, and historical fault statistics databases for power cables in energy storage scenarios.

[0016] Further, in step S3, the structured parameter deviation node includes the relative deviation amplitude, deviation duration, and deviation synchronization data item of each data; the scenario operating condition trigger factor node corresponds to the unique operating conditions of the new energy scenario, including the real-time wind speed preset rule for the wind power scenario, the light intensity mutation rate preset rule for the photovoltaic scenario, and the charge and discharge rate preset rule for the energy storage scenario; the cable fault mechanism node includes mechanical wear of the insulation layer, loose conductor joint, electrochemical corrosion of the metal sheath, and thermal aging of the insulation.

[0017] Further, in step S3, the training process of the Bayesian causal tracing model includes: collecting ≥3000 sets of historical fault cases from the historical fault statistics database of new energy power cables according to the target cable scenario to construct a training dataset. Each set of cases includes a combination of relative deviations of parameters, scenario operating condition triggering factors, and cable fault mechanism labels; based on power cable fault mechanism experts, pre-setting the initial values ​​of the prior probability and conditional probability table of the three-layer association network structure; using the expectation-maximization algorithm to optimize the parameters of the conditional probability table; evaluating the model performance through 10-fold cross-validation; and expanding the training data volume for retraining if the preset accuracy of locking the core fault cause is not met.

[0018] Furthermore, in step S3, the posterior probabilities of each fault mechanism of the target cable are output through collaborative inference using a three-layer association network structure, including the following steps:

[0019] The preprocessed relative deviation of parameters is combined with the real-time scene condition triggering factor to form a joint feature of "parameter deviation-condition disturbance".

[0020] Based on the conditional probability table constructed during the training phase of the Bayesian causal tracing model, the parameter deviation part in the joint feature is input, the occurrence probability of each scenario condition triggering factor is queried and output, and the condition triggering factors with a probability ≥ 0.5 are selected as valid items.

[0021] For effective operating condition triggering factors, the conditional probability of each fault mechanism is inferred through the conditional probability table. If there are multiple effective operating condition triggering factors, the average probability of the same fault mechanism under different operating conditions is taken as the intermediate result.

[0022] The risk weight is calculated by introducing a coupling anomaly coefficient, and the intermediate result is multiplied by the risk weight to obtain the corrected initial posterior probability.

[0023] The initial posterior probability is processed using the max-min normalization method, and the final posterior probability is output.

[0024] Further, in step S3, the rule for locking the core cause of the fault is to sort the posterior probabilities of each fault mechanism of each cable in descending order. If the difference between the maximum posterior probability and the second largest posterior probability is ≥15%, then the fault mechanism corresponding to the maximum posterior probability is determined as the core cause; if the difference is <15%, then the combination of the fault mechanisms of the two cables corresponding to the maximum posterior probability and the second largest posterior probability is output as the core cause.

[0025] Furthermore, in step S4, the graded early warning is divided into three levels, including:

[0026] Level 1 warning: When the coupling anomaly coefficient exceeds the scenario-based warning threshold of 1.3, the warning is simultaneously pushed to the dispatch center and maintenance team to initiate the emergency repair process.

[0027] Level 2 warning: When the coupling anomaly coefficient exceeds 1.1 times the scenario warning threshold but is ≤1.3 times the scenario warning threshold, the terminal's audible and visual alarm is triggered, and the information is pushed to the maintenance team for repair.

[0028] Level 3 early warning: When the coupling anomaly coefficient exceeds the scenario-based early warning threshold but is ≤1.1 times the scenario-based early warning threshold, the warning will be pushed to the maintenance team for investigation.

[0029] This invention also provides an intelligent early warning system for new energy power cables, used to implement the above-mentioned intelligent early warning method for new energy power cables, comprising the following modules:

[0030] The data acquisition module collects the scene coupling parameters of the target cable at predetermined time intervals; the scene coupling parameters include wind power scene coupling parameters, photovoltaic scene coupling parameters and energy storage scene coupling parameters.

[0031] The parameter processing module calculates the relative deviation of each data point from its normal threshold based on the specific data in the scene coupling parameters, obtains the parameter relative deviation combination, and calculates the coupling anomaly coefficient of the target cable by scene adaptability weighting based on the parameter relative deviation combination.

[0032] The core cause locking module inputs the coupling anomaly coefficient and parameter relative deviation combination into a pre-trained Bayesian causal tracing model. Through collaborative inference using the three-layer association network structure built into the model, it outputs the posterior probability of each fault mechanism of the target cable, thereby locking the core cause of the fault. The three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes. The three nodes are associated with each other through a conditional probability table.

[0033] The early warning output module compares the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning thresholds include wind power scenario-based early warning thresholds, photovoltaic scenario-based early warning thresholds, and energy storage scenario-based early warning thresholds.

[0034] Compared with existing technologies, this invention has the following advantages: Targeting the unique operating conditions of new energy scenarios such as wind power, photovoltaics, and energy storage, this invention utilizes a pre-trained Bayesian causal tracing model with a built-in three-layer relational network structure of "structured parameter deviation nodes – scenario operating condition triggering factor nodes – cable fault mechanism nodes" for collaborative reasoning. This allows the output of the posterior probability of each cable fault mechanism, and precise identification of the core fault cause based on the posterior probability difference rule. This design achieves a complete transformation from anomaly identification to core fault cause identification in new energy power cable faults, effectively solving the key deficiency of existing technologies that can only identify cable anomalies but cannot determine the core fault cause. This improves the efficiency of fault investigation for maintenance personnel in new energy scenarios, shortens the fault handling cycle, and reduces maintenance costs and grid safety risks. Attached Figure Description

[0035] Figure 1 This is a flowchart of an intelligent early warning method for new energy power cables according to the present invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0037] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual images. They should not be construed as limiting the scope of this application. To better illustrate the embodiments of the present invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0038] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present application. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0039] In the description of this invention, unless otherwise explicitly specified and limited, the term "connection" or similar designation indicating a connection between components should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0040] Example One

[0041] like Figure 1 As shown, this invention provides an intelligent early warning method for new energy power cables. It achieves intelligent early warning through a process of scenario-based parameter acquisition, coupled anomaly quantification, Bayesian causal tracing, and hierarchical early warning output. This solves the problem in existing technologies that can only identify anomalies but cannot pinpoint the core causes, leading to low operation and maintenance efficiency. It is applicable to wind power, photovoltaic, and energy storage laying scenarios for new energy power cables, and includes the following steps:

[0042] Step 1: Collect scene coupling parameters of the target cable at predetermined time intervals; the scene coupling parameters include wind power scene coupling parameters, photovoltaic scene coupling parameters and energy storage scene coupling parameters;

[0043] The coupling parameters for the wind power scenario include at least cable sheath temperature data, cable insulation partial discharge data, cable fixing point vibration acceleration data, and H2S concentration data; the coupling parameters for the photovoltaic scenario include at least cable sheath temperature data, cable insulation partial discharge data, and ambient light intensity data; and the coupling parameters for the energy storage scenario include at least cable sheath temperature data, cable insulation partial discharge data, cable charging and discharging current data, state of charge data, and relative humidity data.

[0044] In the implementation, the data is processed at predetermined time intervals based on the characteristics of the new energy scenario where the target cable is located to ensure the timeliness and representativeness of the data. For example, in wind power scenarios, due to frequent mechanical vibrations and rapid changes in operating conditions, the time interval can be set to 5 minutes; in photovoltaic scenarios, which are greatly affected by changes in sunlight, the time interval can be set to 10 minutes; and in energy storage scenarios, which fluctuate with the charging and discharging cycle, the time interval can be set to 8 minutes. Cable sheath temperature data and partial discharge data of cable insulation layer are used to monitor the basic condition of the cable. These can be collected using PT100 platinum resistance sensors and ultra-high frequency (UHF) sensors, respectively, to reflect the cable's heat dissipation and heating status and to determine the integrity of the insulation layer. The coupling parameter collection for each scenario is involved. The vibration acceleration data of cable fixing points, which is unique to wind power scenarios, is a typical way to capture the wear effect of mechanical vibration on cables and can be collected using piezoelectric acceleration sensors. H2S concentration data monitors the erosion of cables by corrosive gases and can be collected using electrochemical H2S sensors. The ambient light intensity data specific to the scenario is related to the impact of photovoltaic load changes on cables and can be collected using photoresistor sensors. The cable charging and discharging current data specific to the energy storage scenario reflects current surges and can be collected using Hall current sensors. The state of charge data is related to the energy storage operating load and can be collected through the energy storage BMS system. The relative humidity data monitors the corrosion risk of humid environments on cable insulation and can be collected using capacitive humidity sensors. By collecting the above scenario data in a targeted manner, the key status information of cable operation under different scenarios can be comprehensively captured, laying a data foundation for subsequent anomaly analysis.

[0045] Step two: Based on the specific data in the scenario coupling parameters, calculate the relative deviation of each data point from its normal threshold to obtain the parameter relative deviation combination. The normal threshold of each data point needs to be preset according to the scenario characteristics. For example, the normal threshold for cable sheath temperature is set to 40℃ in the wind power scenario, 38℃ in the photovoltaic scenario, and 42℃ in the energy storage scenario. Then, based on the parameter relative deviation combination, calculate the coupling anomaly coefficient of the target cable by weighting it with scenario adaptability weights. This coefficient can comprehensively reflect the overall anomaly degree of the cable.

[0046] The formula for calculating the coupling anomaly coefficient of the target cable is: , where n is the number of specific data items for the scene coupling parameters. For example, in a wind power scenario, the scene coupling parameters are: collected cable sheath temperature data, cable insulation partial discharge data, cable fixing point vibration acceleration data, and H2S concentration data, which are four items. For example, in a wind power scenario, the scenario-appropriate weighting coefficient is used for the j-th data item. These are the weighting coefficients for cable sheath temperature data, cable insulation partial discharge data, cable fixing point vibration acceleration data, and H2S concentration data, respectively. Let be the relative deviation of the j-th data item from its normal threshold. , Let be the deviation between the actual collected value of the j-th data item and its normal threshold. The normal thresholds for item j are to be set manually. For example, in a wind power scenario, the normal threshold for cable sheath temperature is set to 40℃, the normal threshold for partial discharge of cable insulation is set to 5pC, the normal threshold for vibration acceleration at cable fixing point is set to 0.5g, and the normal threshold for H2S concentration is set to 10ppm.

[0047] Among them, the scene adaptability weight coefficient The determination method involves constructing a fault contribution matrix for target cable scenarios based on a historical fault statistics database of new energy power cables. This database includes historical fault statistics databases for wind power, photovoltaic, and energy storage power cables, storing historical fault data for cables in their respective scenarios, with no fewer than 3000 fault records. Each record must include information such as the parameter status at the time of the fault and the fault mechanism. Then, the analytic hierarchy process (AHP) is used to assign weights to the scenario coupling parameters. For example, in the wind power scenario, the weight of cable sheath temperature data is set to 0.2, the weight of partial discharge data in the cable insulation layer is set to 0.25, the weight of vibration acceleration at the cable fixing point is set to 0.3, and the weight of H2S concentration data is set to 0.25. Furthermore, to adapt to changes in fault patterns, the scenario adaptability weight coefficients need to be dynamically iteratively calibrated quarterly based on newly added fault data using the least squares method. .

[0048] Step 3: Input the coupling anomaly coefficient and parameter relative deviation combination into the pre-trained Bayesian causal tracing model. Through the model's built-in three-layer association network structure, collaborative inference outputs the posterior probability of each fault mechanism of the target cable, thus identifying the core cause of the fault. The three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes, which are linked through conditional probability tables. The structured parameter deviation nodes include the relative deviation amplitude, deviation duration, and deviation synchronization data items for each data point. The scenario condition triggering factor nodes correspond to specific operating conditions in new energy scenarios, including wind power scenarios. The system includes preset rules for wind speed, such as determining a trigger if the wind speed exceeds 12 m / s, or if the H2S concentration exceeds 15 ppm for 10 minutes; preset rules for the change rate of light intensity in photovoltaic scenarios, such as determining a trigger if the increase is >500 W / m² within 10 minutes; and preset rules for the charge / discharge rate in energy storage scenarios, such as determining a trigger if the charge / discharge rate exceeds 1.2C, or if the state of charge (SOC) is below 10% or above 90% for 5 minutes, or if the relative humidity exceeds 80% for 15 minutes. The cable fault mechanism nodes include mechanical wear of the insulation layer, loose conductor joints, electrochemical corrosion of the metal sheath, and thermal aging of the insulation.

[0049] The training of the Bayesian causal tracing model is carried out by collecting no less than 3,000 sets of historical fault cases from the historical fault statistics database of new energy power cables according to the target cable scenario to build a training dataset. Each set of cases includes a combination of relative deviations of parameters, a scenario condition triggering factor, and a cable fault mechanism label. For example, in a set of historical fault cases in the wind power scenario, the combination of relative deviations of parameters is: relative deviation of cable sheath temperature 0.125, relative deviation of partial discharge of cable insulation layer 0.08, relative deviation of vibration acceleration of cable fixing point 0.15, relative deviation of H2S concentration data 0.1, scenario condition triggering factor is "real-time wind speed exceeds 12m / s", and cable fault mechanism label is "mechanical wear of insulation layer". Before training, experts in power cable fault mechanisms need to be invited to pre-set initial values ​​for the prior and conditional probability tables of the three-layer association network structure based on their professional experience. For example, in the wind power scenario, the initial conditional probability of "real-time wind speed exceeding 12 m / s" corresponding to "relative deviation of vibration acceleration at cable fixing point of 0.15" is set to 0.6, and the initial conditional probability of "insulation layer mechanical wear" corresponding to "real-time wind speed exceeding 12 m / s" is set to 0.7. Then, the expectation-maximization algorithm is used to optimize the parameters of the conditional probability tables, enabling the Bayesian causal attribution model to more accurately associate parameters, operating conditions, and fault mechanisms. Simultaneously, 10-fold cross-validation is used to evaluate the model's performance. If the model's accuracy in identifying the core causes of faults does not reach the preset standard, such as 92%, the amount of training data needs to be expanded, such as increasing it to 4000 sets, and retraining is required until the accuracy reaches the target, ensuring the model's reliability.

[0050] Then, the specific operation of outputting the posterior probability of each fault mechanism of the target cable through collaborative inference using a three-layer association network structure is as follows: First, the coupling anomaly coefficient and parameter relative deviation combination are preprocessed, including removing abnormal data with collection errors and standardizing the deviation data to ensure the validity of the input data. Next, the preprocessed parameter relative deviation combination is associated with the real-time scenario operating condition triggering factor to form a "parameter deviation-operating condition disturbance" joint feature, which can simultaneously reflect the cable's abnormal state and operating condition impact. Subsequently, based on the conditional probability table constructed during the training phase of the Bayesian causal tracing model, the parameter deviation part of the joint feature is input, and the occurrence probability of each scenario operating condition triggering factor is queried and output. Operating condition triggering factors with a probability ≥ 0.5 are selected as valid items. For example, in the wind power scenario, after inputting the parameter deviation part, the probability of "real-time wind speed exceeding 12m / s" is 0.7, and the probability of "H2S concentration exceeding 15ppm for 10 minutes" is 0.5. 5. Both are ≥0.5, and both are considered valid items. For valid operating condition triggering factors, the conditional probability of each fault mechanism is inferred through a conditional probability table. If there are multiple valid operating condition triggering factors, the average probability of the same fault mechanism under different valid operating conditions should be taken as an intermediate result. For example, in the wind power scenario above, the conditional probability of "real-time wind speed exceeding 12m / s" corresponding to "mechanical wear of insulation layer" is 0.7, and the conditional probability of "H2S concentration exceeding 15ppm for 10 minutes" is 0.6. The average of the two is 0.65, which is taken as the intermediate result of "mechanical wear of insulation layer". Then, the coupling anomaly coefficient is introduced to calculate the risk weight. The calculation of the risk weight needs to be calibrated according to the characteristics of the scenario. For example, in the wind power scenario, the risk weight coefficient is set to 10, that is, risk weight = coupling anomaly coefficient × weight coefficient. If the coupling anomaly coefficient is 0.13, then the risk weight = 0.13 × 10 = 1.3. The intermediate result is multiplied by the risk weight to obtain the corrected initial posterior probability, such as "mechanical wear of insulation layer". The initial posterior probability is 0.65 × 1.3 = 0.845. Finally, the maximum-minimum normalization method is used to process the initial posterior probabilities of all fault mechanisms to eliminate the influence of dimensions and output the final posterior probability, which can intuitively reflect the probability of each fault mechanism.

[0051] Furthermore, the identification of the core cause of the fault follows these rules: The final posterior probabilities of each cable fault mechanism are sorted in descending order. If the difference between the highest and second-highest posterior probabilities is ≥15%, the fault mechanism corresponding to the highest posterior probability is identified as the core cause. For example, in a wind power scenario, after sorting, the posterior probability of "insulation layer mechanical wear" is 0.85, and "metal sheath electrochemical corrosion" is 0.68, with a difference of 17% ≥ 15%, so the core cause is "insulation layer mechanical wear." If the difference between the highest and second-highest posterior probabilities is <15%, the combination of the two cable fault mechanisms corresponding to the highest and second-highest posterior probabilities is output as the core cause. For example, in a wind power scenario, the posterior probability of "insulation layer mechanical wear" is 0.82, and "conductor joint loosening" is 0.71, with a difference of 11% < 15%, so the core cause is "insulation layer mechanical wear + conductor joint loosening." This rule allows for precise identification of the core cause of the fault, avoiding blind troubleshooting during maintenance and improving maintenance efficiency.

[0052] Step four: Compare the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning threshold includes wind power scenario-based early warning threshold, photovoltaic scenario-based early warning threshold and energy storage scenario-based early warning threshold.

[0053] This process requires first determining scenario-based early warning thresholds. The thresholds for different scenarios are manually set based on their fault characteristics and historical data. For example, wind power scenarios have a high risk of mechanical vibration, so the scenario-based early warning threshold can be set to 0.15; photovoltaic scenarios have relatively stable loads, so the scenario-based early warning threshold can be set to 0.12; and energy storage scenarios have large current surges, so the scenario-based early warning threshold can be set to 0.18. Early warning levels are divided into three levels, with the triggering conditions and response measures for each level as follows: Level 1 is the highest level, triggered when the coupling anomaly coefficient exceeds 1.3 times the scenario-based early warning threshold. At this time, the system will simultaneously push early warning information containing the core causes of the fault to the power grid dispatch center and the on-site maintenance team, and initiate emergency repair procedures. The dispatch center can adjust the power grid load in a timely manner to avoid load fluctuations, and the maintenance team must arrive on-site within a specified time, such as within 1 hour, to carry out emergency repairs and minimize downtime. Level 2 is the medium level, triggered when the coupling anomaly coefficient exceeds 1.1 times the scenario-based early warning threshold. When the threshold is reached but not exceeding 1.3 times the scenario-based warning threshold, an audible and visual alarm will be triggered on the field terminal, such as a flashing red light on the control cabinet accompanied by a buzzer, reminding maintenance personnel to pay attention to the cable status in time to prevent the fault from developing further. The warning information containing the core cause of the fault will also be pushed to the maintenance team for repair. The third-level warning is the lowest level. When the coupling anomaly coefficient exceeds the scenario-based warning threshold but does not exceed 1.1 times the scenario-based warning threshold, a warning information containing the core cause of the fault will be pushed to the maintenance team to guide maintenance personnel to conduct targeted investigations and eliminate potential fault risks in advance.

[0054] Example Two This invention also provides an intelligent early warning system for new energy power cables, used to implement the above-mentioned intelligent early warning method for new energy power cables, comprising the following modules:

[0055] The data acquisition module collects the scene coupling parameters of the target cable at predetermined time intervals; the scene coupling parameters include wind power scene coupling parameters, photovoltaic scene coupling parameters and energy storage scene coupling parameters.

[0056] The parameter processing module calculates the relative deviation of each data point from its normal threshold based on the specific data in the scene coupling parameters, obtains the parameter relative deviation combination, and calculates the coupling anomaly coefficient of the target cable by scene adaptability weighting based on the parameter relative deviation combination.

[0057] The core cause locking module inputs the coupling anomaly coefficient and parameter relative deviation combination into a pre-trained Bayesian causal tracing model. Through collaborative inference using the three-layer association network structure built into the model, it outputs the posterior probability of each fault mechanism of the target cable, thereby locking the core cause of the fault. The three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes. The three nodes are associated with each other through a conditional probability table.

[0058] The early warning output module compares the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning thresholds include wind power scenario-based early warning thresholds, photovoltaic scenario-based early warning thresholds, and energy storage scenario-based early warning thresholds.

[0059] The data acquisition module, parameter processing module, core cause locking module, and early warning output module described above correspond to the steps in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. These modules, as part of the system, can be executed in a computer system, such as a set of computer-executable instructions.

[0060] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0061] The above are merely embodiments of the present invention. The circuits, electronic components, and modules involved are all prior art, fully achievable by those skilled in the art, and require no further explanation. The scope of protection in this application does not involve improvements to the software and methods. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all prior art in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A smart early warning method for new energy power cables, characterized in that, The following steps are applicable to the laying of new energy power cables in wind power, photovoltaic, and energy storage scenarios: S1. Collect scene coupling parameters of the target cable at predetermined time intervals; the scene coupling parameters include wind power scene coupling parameters, photovoltaic scene coupling parameters and energy storage scene coupling parameters; S2. Based on the specific data in the scene coupling parameters, calculate the relative deviation of each data point from its normal threshold to obtain the parameter relative deviation combination, and calculate the coupling anomaly coefficient of the target cable by scene adaptability weighting based on the parameter relative deviation combination. S3. Input the coupling anomaly coefficient and parameter relative deviation combination into the pre-trained Bayesian causal tracing model, and output the posterior probability of each fault mechanism of the target cable through collaborative inference using the three-layer association network structure built into the model, thereby locking in the core cause of the fault; the three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes, and the three nodes are associated through a conditional probability table. S4. Compare the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning threshold includes wind power scenario-based early warning threshold, photovoltaic scenario-based early warning threshold and energy storage scenario-based early warning threshold.

2. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S1, the wind power scenario coupling parameters include at least cable sheath temperature data, cable insulation partial discharge data, cable fixing point vibration acceleration data, and H2S concentration data. The photovoltaic scenario coupling parameters include at least cable sheath temperature data, cable insulation partial discharge data, and ambient light intensity data. The coupling parameters for the energy storage scenario include at least cable sheath temperature data, cable insulation partial discharge data, cable charging and discharging current data, state of charge data, and relative humidity data.

3. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S2, the formula for calculating the coupling anomaly coefficient of the target cable is: Where n is the number of specific data items for the scene coupling parameters; The scene adaptability weight coefficient for the j-th data item; Let be the relative deviation of the j-th data item from its normal threshold. , Let be the deviation between the actual collected value of the j-th data item and its normal threshold. is the normal threshold corresponding to data item j.

4. The intelligent early warning method for new energy power cables as described in claim 3, characterized in that, The scene adaptability weight coefficient The determination method is as follows: a fault contribution matrix for the target cable scenario is constructed based on the historical fault statistics database of new energy power cables. Weights are assigned using the analytic hierarchy process (AHP), and the scenario adaptability weight coefficients are dynamically and iteratively calibrated quarterly based on newly added fault data using the least squares method. ; The historical fault statistics database for new energy power cables includes historical fault statistics databases for power cables in wind power scenarios, historical fault statistics databases for power cables in photovoltaic scenarios, and historical fault statistics databases for power cables in energy storage scenarios.

5. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S3, the structured parameter deviation node includes the relative deviation amplitude, deviation duration, and deviation synchronization data item of each data item; the scenario operating condition trigger factor node corresponds to the unique operating conditions of the new energy scenario, including the real-time wind speed preset rule for the wind power scenario, the light intensity mutation rate preset rule for the photovoltaic scenario, and the charge and discharge rate preset rule for the energy storage scenario; the cable fault mechanism node includes mechanical wear of the insulation layer, loose conductor joint, electrochemical corrosion of the metal sheath, and thermal aging of the insulation.

6. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S3, the training process of the Bayesian causal tracing model includes collecting ≥3000 sets of historical fault cases from the historical fault statistics database of new energy power cables according to the target cable scenario to construct a training dataset. Each set of cases includes a combination of relative parameter deviations, scenario operating conditions triggering factors, and cable fault mechanism labels. Based on the expert on power cable fault mechanism, the initial values ​​of the prior probability and conditional probability tables of the three-layer correlation network structure are preset; The conditional probability table parameters are optimized using the expectation-maximization algorithm, and the model performance is evaluated using 10-fold cross-validation. If the model does not meet the preset accuracy for identifying the core cause of the fault, the amount of training data is expanded and the model is retrained.

7. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S3, the posterior probabilities of each fault mechanism of the target cable are output through collaborative inference using a three-layer association network structure, including the following steps: The preprocessed parameter relative deviations are combined and associated with real-time scene condition triggering factors to form a joint feature of "parameter deviation - condition disturbance". Based on the conditional probability table constructed during the training phase of the Bayesian causal tracing model, the parameter deviation part in the joint feature is input, the occurrence probability of each scenario condition triggering factor is queried and output, and the condition triggering factors with a probability ≥ 0.5 are selected as valid items. For effective operating condition triggering factors, the conditional probability of each fault mechanism is inferred through the conditional probability table. If there are multiple effective operating condition triggering factors, the average probability of the same fault mechanism under different operating conditions is taken as the intermediate result. The risk weight is calculated by introducing a coupling anomaly coefficient, and the intermediate result is multiplied by the risk weight to obtain the corrected initial posterior probability. The initial posterior probability is processed using the max-min normalization method, and the final posterior probability is output.

8. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S3, the rule for locking the core cause of the fault is to sort the posterior probabilities of each fault mechanism of each cable in descending order. If the difference between the maximum posterior probability and the second largest posterior probability is ≥15%, then the fault mechanism corresponding to the maximum posterior probability is determined as the core cause; if the difference is <15%, then the combination of the fault mechanisms of the two cables corresponding to the maximum posterior probability and the second largest posterior probability is output as the core cause.

9. The intelligent early warning method for new energy power cables as described in claim 1, characterized in that, In step S4, the graded early warning is divided into three levels, including: Level 1 warning: When the coupling anomaly coefficient exceeds the scenario-based warning threshold of 1.3, the warning is simultaneously pushed to the dispatch center and maintenance team to initiate the emergency repair process. Level 2 warning: When the coupling anomaly coefficient exceeds 1.1 times the scenario warning threshold but is ≤1.3 times the scenario warning threshold, the terminal's audible and visual alarm is triggered, and the information is pushed to the maintenance team for repair. Level 3 early warning: When the coupling anomaly coefficient exceeds the scenario-based early warning threshold but is ≤1.1 times the scenario-based early warning threshold, the warning will be pushed to the maintenance team for investigation.

10. A smart early warning system for new energy power cables, used to implement the smart early warning method for new energy power cables as described in any one of claims 1-9, characterized in that, Includes the following modules: The data acquisition module collects scene coupling parameters of the target cable at predetermined time intervals; The scenario coupling parameters include wind power scenario coupling parameters, photovoltaic scenario coupling parameters, and energy storage scenario coupling parameters; The parameter processing module calculates the relative deviation of each data point from its normal threshold based on the specific data in the scene coupling parameters, obtains the parameter relative deviation combination, and calculates the coupling anomaly coefficient of the target cable by scene adaptability weighting based on the parameter relative deviation combination. The core cause locking module inputs the coupling anomaly coefficient and parameter relative deviation combination into a pre-trained Bayesian causal tracing model. Through collaborative inference using the three-layer association network structure built into the model, it outputs the posterior probability of each fault mechanism of the target cable, thereby locking the core cause of the fault. The three-layer association network structure includes structured parameter deviation nodes, scenario condition triggering factor nodes, and cable fault mechanism nodes. The three nodes are associated with each other through a conditional probability table. The early warning output module compares the coupling anomaly coefficient of the target cable with the scenario-based early warning threshold to generate hierarchical early warning information containing the core causes of the fault. The scenario-based early warning thresholds include wind power scenario-based early warning thresholds, photovoltaic scenario-based early warning thresholds, and energy storage scenario-based early warning thresholds.

Images