Direct-current power transmission fault diagnosis method and system based on multi-modal data analysis

By combining multimodal data analysis and real-time monitoring with filtering algorithms to eliminate electromagnetic interference and dynamically adjusting protection strategies, the accuracy and adaptability of DC transmission fault diagnosis have been solved, thereby improving the stability and safety of the system.

WO2026232799A1PCT designated stage Publication Date: 2026-11-12DALI BUREAU OF ULTRA HIGH VOLTAGE TRANSMISSION CO CHINA SOUTHERN POWER GRID CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2025/140572
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-06
Filing Date
2025-12-06
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing DC transmission fault diagnosis methods rely on a single data source and fixed parameter limits, which makes them prone to misjudgment or omission under different operating conditions. They also cannot fully cover multiple fault types and are susceptible to electromagnetic interference.

Method used

Multimodal data analysis is employed to monitor DC transmission line parameters in real time. By extracting features and matching temporary protection measures, combined with filtering algorithms to eliminate electromagnetic interference, fault diagnosis strategies are dynamically adjusted, and comprehensive fault diagnosis is performed using multimodal data.

Benefits of technology

It improves the accuracy and response speed of fault diagnosis, ensures system stability and security, dynamically adjusts protection strategies to adapt to different fault types, and reduces false positives and false negatives.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2025140572_12112026_PF_FP_ABST
    Figure CN2025140572_12112026_PF_FP_ABST
Patent Text Reader

Abstract

A direct-current power transmission fault diagnosis method and system based on multi-modal data analysis. The method comprises: when an abnormal fluctuation occurs in a direct-current power transmission line, activating a temporary line protection measure, and simultaneously feeding parameters into a data processing center for feature extraction, so as to obtain fault types corresponding to a valid feature; and compiling statistics on associated features of each fault type, so as to obtain a fault diagnosis result. Multi-modal parameters of a direct-current power transmission line are monitored in real time to eliminate electromagnetic interference, so as to improve the accuracy of fault diagnosis and ensure the reliability of data quality. Monitoring and protection strategies are adjusted, so as to achieve optimal operational assurance. By means of dynamic monitoring and tracking analysis of potential faults, the stability and reliability of a system are improved.
Need to check novelty before this filing date? Find Prior Art

Description

A Method and System for Fault Diagnosis of DC Transmission Based on Multimodal Data Analysis Technical Field

[0001] This invention relates to the field of DC transmission line fault diagnosis, specifically to a DC transmission fault diagnosis method and system based on multimodal data analysis. Background Technology

[0002] In modern power systems, high-voltage direct current (HVDC) transmission technology is widely used in power grid construction due to its advantages such as long-distance, low-loss transmission, ability to interconnect asynchronous grids, and flexible power flow control. However, HVDC transmission systems face various faults during operation, such as converter station faults, DC line faults, grounding faults, control system anomalies, and electromagnetic interference. To address these issues, fault diagnosis technology has become crucial for ensuring the safe and stable operation of the power grid.

[0003] Due to the complexity, nonlinearity, and strong electromagnetic interference inherent in DC transmission systems, traditional fault diagnosis methods relying on single data sources and single analysis techniques have significant limitations in terms of accuracy, anti-interference capability, real-time performance, and adaptability. Currently, mainstream HVDC fault diagnosis methods, primarily based on single data sources (such as voltage, current, and power) and single analysis techniques (such as threshold methods, rule matching, and traditional signal processing), still have considerable limitations. Therefore, analyzing multimodal data for fault diagnosis is an important future direction in the field of DC transmission fault diagnosis.

[0004] Existing technology, such as the invention patent with announcement number CN104820158B, is a method for judging DC line breakage faults in flexible DC transmission systems. It detects and calculates the positive and negative DC bus currents, the rate of change of the positive and negative DC bus currents, and the AC-DC power difference. When the above parameters simultaneously meet the following conditions, it is determined that a DC line breakage fault has occurred in the system: ① The absolute value of either the positive or negative DC bus current is less than the current limit; ② The absolute values ​​of the rate of change of the positive and negative DC bus currents are both greater than the current rate of change limit; ③ The absolute value of the difference between AC power and DC power is greater than the power difference limit.

[0005] Existing technology, such as the invention patent with announcement number CN118604530B, is a DC fault location method and system. The method includes: acquiring the first voltage and current signal of the DC line in real time; acquiring the positive and negative current signals of the DC line respectively, and calculating the current between the positive and negative poles of the DC line based on the positive and negative current signals using a component formula; acquiring the second voltage and current signal of the DC line within a preset time period; connecting all the second voltage and current signals within the preset time period sequentially, and performing curve fitting on the formed line to form a fitted line; selecting the third voltage and current signal corresponding to two random time points of the fitted line, and selecting the corresponding matrix formula based on the type of fault signal; and using a distance calculation formula to calculate the location of the DC line fault based on the current value between the positive and negative poles of the DC line, the voltage matrix, and the current matrix.

[0006] As can be seen from the above solutions, existing technologies in the field of DC transmission fault diagnosis often rely on fixed parameter limits. However, under different operating conditions (such as load changes, ambient temperature changes, and grid disturbances), fixed parameter limits may lead to misjudgments or missed diagnoses. Furthermore, existing technologies often only investigate single faults, but DC transmission systems have multiple fault types, including DC short circuits, grounding faults, and converter faults. A single parameter combination cannot cover all fault modes. In practical applications, DC transmission systems are easily affected by factors such as electromagnetic interference, and ignoring the influence of these interferences may lead to misjudging certain transient fluctuations as faults. Therefore, when diagnosing DC transmission faults, a comprehensive fault diagnosis should be conducted by considering multi-modal parameters after eliminating the influence of other factors. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a method and system for diagnosing DC transmission faults based on multimodal data analysis. To achieve the above objectives, this invention employs the following technical solution: A method for diagnosing DC transmission faults based on multimodal data analysis, comprising:

[0008] The DC transmission lines are monitored in real time. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated. At the same time, the DC transmission line parameters at the abnormal moment are collected into the data processing center for feature extraction.

[0009] When the feature value of a key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature and the corresponding fault types are matched.

[0010] The correlation characteristics of each fault type are statistically analyzed to obtain the fault diagnosis confidence score of each fault type, and then compared with the fault diagnosis confidence threshold to obtain the fault diagnosis result.

[0011] As a preferred technical solution, the DC transmission line is monitored in real time. When abnormal fluctuations occur in the DC transmission line, temporary line protection measures are activated. The specific process is as follows:

[0012] Real-time acquisition of DC transmission line parameters, including electrical parameters, converter station control parameters, and equipment parameters.

[0013] Electrical parameters include DC voltage, DC current, active power, and harmonic content.

[0014] The converter station control parameters include the firing angle and the arc extinguishing angle.

[0015] Equipment parameters include converter valve temperature, line insulation test score, and grounding current.

[0016] The parameters of the DC transmission line are transmitted to the data processing center for analysis to obtain the abnormal fluctuation risk value. When the abnormal fluctuation risk value exceeds the abnormal fluctuation risk threshold, it is determined that the DC transmission line has experienced abnormal fluctuation. The time when the DC transmission line parameters are acquired is recorded as the abnormal time. Based on the abnormal fluctuation risk value, a mapping and matching is performed with the temporary line protection measures corresponding to each abnormal fluctuation risk value range pre-stored in the data processing center's built-in database to obtain the temporary line protection measures for that abnormal fluctuation risk value, and the temporary line protection is activated. The specific process includes:

[0017] When the abnormal fluctuation risk value is greater than or equal to the highest risk threshold, it is determined that the abnormal fluctuation risk value is in the first interval, and the temporary line protection measures set in the first interval are matched and the temporary line protection is activated.

[0018] When the abnormal fluctuation risk value is less than the highest risk threshold but greater than or equal to the medium risk threshold, the abnormal fluctuation risk value is determined to be in the second range, and is matched with the temporary line protection measures set in the second range and the temporary line protection is activated.

[0019] When the abnormal fluctuation risk value is less than the medium risk threshold but greater than or equal to the abnormal fluctuation risk threshold, the abnormal fluctuation risk value is determined to be in the third interval, and is matched with the temporary line protection measures set in the third interval and the temporary line protection is activated.

[0020] As a preferred technical solution, the parameters of the DC transmission line at abnormal times are imported into the data processing center for feature extraction. The specific process is as follows:

[0021] The parameters of the DC transmission lines at abnormal times are collected into the data processing center and then standardized.

[0022] Each time-domain feature and each frequency-domain feature is extracted separately. After principal component analysis of each time-domain feature and each frequency-domain feature, the key features of the DC transmission line parameters are obtained. The key features are then imported into the fault diagnosis processor built into the data processing center for fault diagnosis analysis.

[0023] As a preferred technical solution, when the feature value of a certain key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature and the corresponding fault types are matched to obtain the valid feature, specifically including:

[0024] After receiving each key feature, the fault diagnosis processor built into the data processing center calculates the feature value of each key feature and extracts the feature threshold of each key feature from the built-in database of the data processing center. It compares the feature value of each key feature with its corresponding feature threshold. When the feature value of a key feature is greater than its corresponding feature threshold, the key feature is determined to be a valid feature, and the fault types corresponding to the valid feature are counted.

[0025] When the feature value of a key feature is less than or equal to its corresponding feature threshold, the key feature is determined to be an invalid feature, and fault diagnosis is performed based on the determination results of each key feature.

[0026] As a preferred technical solution, fault diagnosis is performed based on the determination results of each key feature, specifically including:

[0027] When all key features are invalid, the abnormal moment is recorded, the DC transmission line is switched to pending state, and a short-term monitoring cycle is set for trial operation.

[0028] The DC transmission line is monitored during trial operation to obtain electrical test parameters within a short monitoring period. These parameters are then input into the data processing center for feature extraction. Key features are obtained, and it is determined whether each key feature is valid, thus yielding the trial operation monitoring results of the DC transmission line.

[0029] If the test operation monitoring results of the DC transmission line are all invalid, the DC transmission line is determined to be fault-free and switched to normal operation.

[0030] If the trial operation monitoring results of the DC transmission line show the presence of valid features, then the DC transmission line is determined to have a fault, and the fault types corresponding to each valid feature are statistically analyzed.

[0031] As a preferred technical solution, the correlation characteristics of each fault type are statistically analyzed to obtain the fault diagnosis confidence score for each fault type, specifically including:

[0032] Each effective feature corresponding to each fault type is statistically analyzed into a related feature for each fault type, and the fault index of each related feature is obtained. Based on the fault index of each related feature and the number of related features for each fault type, the coupling process is corrected to obtain the fault diagnosis confidence score of each fault type.

[0033] The fault indices for each associated feature are obtained, specifically including:

[0034] The feature difference of each associated feature is obtained by subtracting its feature value from its corresponding feature threshold. Based on the feature difference of each associated feature, the fault index of each effective feature is obtained by mapping and matching it with the fault index corresponding to each feature difference in the built-in database of the data processing center.

[0035] As a preferred technical solution, the fault diagnosis result is obtained by comparing it with the fault diagnosis confidence threshold, specifically including:

[0036] The fault diagnosis confidence score for each fault type is compared with the fault diagnosis confidence threshold. If the fault diagnosis confidence score for a certain fault type is greater than or equal to the fault diagnosis confidence threshold, then the fault type is determined to be a significant fault.

[0037] If the fault diagnosis confidence score of a certain fault type is less than the fault diagnosis confidence threshold, then the fault type is determined to be a potential fault.

[0038] As the preferred technical solution, the fault diagnosis results are obtained, and the specific processing conditions are as follows:

[0039] The fault types that are identified as significant faults in the statistical fault diagnosis results are denoted as each significant fault type. Based on the fault diagnosis confidence score of each significant fault type, the significant fault types are sorted and sent to the data management terminal for early warning.

[0040] If the fault diagnosis results for each fault type are all potential faults, then based on the total number of fault types, the dynamic monitoring cycle is obtained by matching it with the preset dynamic monitoring cycle corresponding to the total number of each fault type in the database built into the data processing center. Dynamic monitoring of the DC transmission line is then performed based on this dynamic monitoring cycle, specifically including:

[0041] Based on the dynamic monitoring cycle, real-time monitoring data of DC transmission lines are acquired within the dynamic monitoring cycle, and trend analysis is performed on the data to obtain the tracking analysis results of DC transmission lines.

[0042] If, during the dynamic monitoring period, the associated characteristic value of a potential fault continues to rise and exceeds its corresponding characteristic threshold, the tracking analysis result of the DC transmission line will be determined as a tracking fault, and the fault type will be determined as a significant fault, and sent to the data management terminal for early warning.

[0043] If the associated characteristic values ​​of each potential fault remain stable and do not exceed their corresponding characteristic thresholds during the dynamic monitoring period, the tracking and analysis results of the DC transmission line will be judged as short-term fluctuations, the dynamic monitoring period will be terminated, and the normal monitoring period will be switched.

[0044] As a preferred technical solution, real-time acquisition of DC transmission line parameters also includes detecting the sensors and removing the influence of electromagnetic interference inside the sensors. The specific processing conditions are as follows:

[0045] Monitor sensor power data during abnormal moments, including voltage change rate, current change rate, and electromagnetic interference spectrum.

[0046] Voltage change rate is used to determine whether it is a high-frequency interference source, current change rate is used to determine whether electromagnetic interference has radiative interference characteristics, and electromagnetic interference spectrum is used to determine the location of the interference source.

[0047] Based on the sensor power data at abnormal times, the data is imported into the data processing center for feature extraction. After confirming the characteristics of electromagnetic interference, the data is mapped and matched with the filtering algorithms corresponding to each feature in the built-in database of the data processing center to obtain the filtering algorithms corresponding to the electromagnetic interference characteristics for electromagnetic interference elimination processing.

[0048] Record the quality data of DC transmission line parameters before and after filtering, including harmonic content, signal-to-noise ratio, and mean square error.

[0049] The filtered DC transmission line parameters are re-imported into the data processing center. The quality data of the DC transmission line parameters before and after filtering are compared to verify whether the electromagnetic interference has been effectively eliminated.

[0050] If the electromagnetic interference is effectively eliminated, the filtered DC transmission line parameters will be further analyzed and processed.

[0051] If the electromagnetic interference is not effectively eliminated, an early warning will be sent to the data management terminal.

[0052] A DC transmission fault diagnosis system based on multimodal data analysis includes:

[0053] The real-time monitoring module is used to monitor DC transmission lines in real time. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated, and the DC transmission line parameters at the abnormal moment are imported into the data processing center for feature extraction.

[0054] The fault type matching module is used to record a key feature as a valid feature and match the fault types corresponding to the valid feature when the feature value of a key feature is greater than its corresponding feature threshold.

[0055] The fault type judgment module is used to statistically analyze the associated features of each fault type, obtain the fault diagnosis confidence score of each fault type, and compare it with the fault diagnosis confidence threshold to obtain the fault diagnosis result.

[0056] Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects:

[0057] (1) This invention provides a DC transmission fault diagnosis method based on multimodal data analysis. By monitoring the multimodal parameters of DC transmission lines in real time, it can significantly improve the response speed to faults and ensure the stability and safety of the system. It provides a comprehensive understanding of the operating status of DC transmission lines from multiple perspectives. Data centers can extract and analyze features from this multimodal data to help accurately diagnose the type and cause of faults.

[0058] (2) This invention monitors the rate of change of voltage, rate of change of current, and electromagnetic interference spectrum, and combines this with the filtering algorithm of the data processing center to eliminate interference, ensuring data quality and preventing electromagnetic interference from misleading fault diagnosis results, thereby improving the accuracy of fault diagnosis. The intelligent filtering algorithm effectively eliminates electromagnetic interference, noise, and other factors in the sensor, ensuring the reliability of data quality. This improves the accuracy of fault diagnosis.

[0059] (3) This invention adjusts the monitoring and protection strategies according to different fault types and diagnostic confidence scores, enabling the system to have strong adaptability. Matching different risk ranges with temporary protection measures can dynamically adjust the protection level, thereby achieving optimal operational assurance. Through dynamic monitoring and tracking analysis of potential faults, the system can identify and respond to impending faults in advance, thereby improving the stability and reliability of the system. A dynamic monitoring cycle is introduced to continuously track and analyze potential faults. When the associated characteristic values ​​of certain potential faults continue to rise and exceed the characteristic threshold, the system will adjust the diagnostic strategy in real time, determine the fault type as a significant fault, and issue an early warning. Through this dynamic adjustment mechanism, the system can continuously optimize the accuracy of fault diagnosis and take corresponding protection measures in a timely manner, avoiding omissions and misjudgments caused by static judgment.

[0060] Of course, any product implementing this invention does not necessarily need to achieve all of the above advantages at the same time. Attached Figure Description

[0061] Figure 1 is a schematic diagram of the method flow of the present invention.

[0062] Figure 2 is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] In the description of this invention, it should be understood that the terms "opening", "upper", "lower", "thickness", "top", "middle", "length", "inner", "around", etc., which indicate orientation or positional relationship, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting this invention.

[0065] Please refer to Figure 1. This embodiment of the invention provides a DC transmission fault diagnosis method based on multimodal data analysis, specifically including:

[0066] The DC transmission lines are monitored in real time. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated. At the same time, the DC transmission line parameters at the abnormal moment are collected into the data processing center for feature extraction.

[0067] Real-time monitoring of DC transmission lines is conducted. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated. The specific process is as follows:

[0068] Real-time acquisition of DC transmission line parameters, including electrical parameters, converter station control parameters, and equipment parameters.

[0069] Electrical parameters include DC voltage, DC current, active power, and harmonic content.

[0070] DC voltage refers to the voltage difference across a DC power transmission line and is a fundamental parameter for power transmission. The magnitude of the DC voltage directly determines the flow of current and the transmission capacity. Higher voltage helps improve transmission efficiency, but excessively high voltage may damage equipment.

[0071] Direct current (DC) refers to the amount of current flowing along a DC transmission line, representing the actual current flowing through the circuit under a certain voltage. The magnitude of the current directly affects the load and power transmission capacity of the DC transmission line. Excessive current can lead to line overload and even burn out equipment.

[0072] Active power refers to the actual electrical energy transmitted, representing the energy conversion between current and voltage, and is the actual usable electricity. It reflects the workload of DC transmission lines and the efficiency of energy use. Too low active power indicates low system efficiency, while too high power may lead to system overload and affect equipment safety.

[0073] Harmonic content refers to the distortion of current or voltage waveforms in DC transmission lines, especially when using power electronic equipment (such as converter stations). Ideally, the voltage and current waveforms in a DC transmission line are sinusoidal, but due to nonlinear loads or the use of converter equipment, the current and voltage waveforms may deviate from the sine wave, a phenomenon known as harmonics. High harmonic content can lead to overheating of electrical equipment, reduced efficiency, and decreased system stability.

[0074] The converter station control parameters include the firing angle and the arc extinguishing angle.

[0075] In a DC transmission system, the converter station is a key device that converts alternating current (AC) into direct current (DC) or vice versa.

[0076] The firing angle refers to the angle relative to the cycle of alternating current when the thyristor trigger signal occurs in the converter station. The converter station uses thyristors to control the conduction and cutoff of current, and the firing angle controls the conduction time of the thyristor.

[0077] By adjusting the firing angle, the on-time of the current can be controlled, thereby affecting the magnitude and direction of the current and regulating the flow of transmitted power. The firing angle directly affects the converter's output voltage; changing the firing angle can adjust the magnitude and fluctuation range of the DC voltage. Adjusting the firing angle can also alter the amount of active power transmitted. A larger firing angle (i.e., a delayed firing time) reduces the current's on-time, thus reducing output power. Conversely, a smaller firing angle increases power output.

[0078] The arc-extinguishing angle refers to the angle relative to the alternating current cycle when the thyristor in the converter station is turned off. It determines the moment when the thyristor switches from the on state to the off state.

[0079] The purpose of the arc-extinguishing angle is to ensure that the thyristor can be completely turned off at the correct time of the AC cycle, avoiding problems such as "reverse breakdown" or inability to switch states caused by reverse current.

[0080] An appropriate arc-extinguishing angle ensures that the converter station can effectively "extinguish the arc," preventing the current from being turned off too early or too late, thereby avoiding unstable operation of the converter.

[0081] If the arc-extinguishing angle is not set properly, the thyristor may not be able to shut off effectively, resulting in overcurrent or overvoltage, which may affect the safe operation of the system.

[0082] During AC / DC conversion, the arc extinguishing angle affects the timing of current flow switching, ensuring that the converter station can smoothly perform power direction conversion or zero current crossover.

[0083] Equipment parameters include converter valve temperature, line insulation test score, and grounding current.

[0084] The converter valve temperature refers to the operating temperature of the converter valve in the converter station. During operation, the converter valve generates heat due to the flow of current, and temperature variations can affect its performance and safety. Excessively high temperatures can damage converter valve components or shorten their lifespan. Overheating can also affect the converter valve's conductivity, potentially leading to equipment overload or malfunction.

[0085] Line insulation test scoring is the result of evaluating the insulation performance of transmission lines, determining their health status by detecting the insulation resistance. This score is derived by measuring the insulation resistance value and other relevant parameters of the line, reflecting the presence of aging, cracks, or other defects affecting insulation performance. Insulation performance is one of the key factors ensuring the safe operation of DC transmission lines. Low insulation resistance values ​​indicate poor insulation effectiveness, potentially posing risks of leakage current, short circuits, or equipment damage.

[0086] In this embodiment of the invention, the insulation performance of the transmission line is tested by transmitting a high-voltage DC voltage to the transmission line, specifically including:

[0087] Connect the high-voltage end of the high-voltage generator to the circuit under test, and ground the other end. Connect the leakage current meter in series in the circuit. Gradually increase the voltage according to the standard (e.g., 1.5-2 times the rated voltage) to avoid impact that could damage the insulation. Record the leakage current as the voltage changes. If there is an abnormal increase, i.e., the single increase does not exceed the preset increase threshold but the total increase exceeds the preset increase threshold, there may be an insulation defect. Continuously apply the test voltage according to the duration standard and observe whether the leakage current is stable. If the leakage current is small and stable, i.e., the increase of the leakage current changes with the voltage and remains within the preset increase threshold, it indicates good insulation performance. If the leakage current gradually increases, i.e., the leakage current increases slowly with the voltage until it exceeds the preset increase threshold at a certain voltage, it indicates possible moisture or aging problems. If the leakage current suddenly increases, i.e., the single increase of the leakage current exceeds the preset increase threshold, it indicates possible insulation damage or partial discharge. Gradually decrease the voltage to avoid sudden changes in the electric field that could lead to breakdown.

[0088] Grounding current refers to the current flowing to the ground through a grounding device. Grounding in DC transmission lines is a method of protecting systems and equipment from electric shock and other electrical faults. When a fault occurs in the system (such as insulation failure or current leakage), grounding current flows to the ground to protect people and equipment from electrical hazards. The magnitude of the grounding current can reflect whether a grounding fault or current leakage exists in the DC transmission line. An excessively large grounding current usually indicates a fault in a part of the system, requiring immediate inspection and repair.

[0089] The parameters of the DC transmission line are transmitted to the data processing center. The DC transmission line verification parameters are extracted from the data processing center's built-in database, and the abnormal fluctuation risk values ​​are obtained through processing and analysis. Specifically, these include:

[0090] ;

[0091] in, This represents the risk value for abnormal fluctuations. DC voltage It is direct current. Active power Harmonic content, For trigger angle, To eliminate the arc angle, For the temperature of the converter valve, For grounding current, Scoring for line insulation testing, To verify the DC voltage, To verify the DC current, To verify active power, To verify the harmonic content, To verify the trigger angle, To verify the arc extinguishing angle, To verify the temperature of the converter valve, To verify the grounding current, The threshold for scoring line insulation tests. For electrical parameter weighting factors, The weighting factor for the control parameters of the converter station. For equipment parameter weighting factors, For voltage parameter unification factor, For the current parameter uniformity factor, This is a unified factor for active power parameters. A unified factor for harmonic content parameters. For temperature parameter uniformity factor, This is a unified factor for the scoring parameters of line insulation testing.

[0092] It should be noted that the weighting factors for electrical parameters, converter station control parameters, and equipment parameters all range from 0 to 1, and satisfy the following conditions: The electrical parameter weighting factor is a pre-stored influence factor of electrical parameters in the built-in database of the data processing center, representing the degree to which electrical parameters affect the abnormal fluctuation risk value. The converter station control parameter weighting factor is a pre-stored influence factor of converter station control parameters in the built-in database of the data processing center, representing the degree to which converter station control parameters affect the abnormal fluctuation risk value. The equipment parameter weighting factor is a pre-stored influence factor of equipment parameters in the built-in database of the data processing center, representing the degree to which equipment parameters affect the abnormal fluctuation risk value. When in use, it is directly extracted from the built-in database of the data processing center. For example, the electrical parameters, converter station control parameters, and equipment parameters of the power system DC transmission parameters are input into the preset mapping set in the built-in database of the data processing center to obtain the electrical parameter weighting factor, converter station control parameter weighting factor, and equipment parameter weighting factor of the power system DC transmission parameters. The corresponding mapping relationship is one-to-one.

[0093] It should also be noted that there is a close correlation between electrical parameters, converter station control parameters, and equipment parameters. DC voltage, DC current, and active power are the core operating parameters of a DC transmission system, and their changes are directly affected by the converter station control parameters. Increasing the firing angle will decrease the DC voltage, possibly reduce the DC current, and decrease the active power. Decreasing the firing angle will increase the DC voltage, possibly increase the DC current, and increase the active power. The arc-extinguishing angle affects the shut-off state of the converter valve, and thus affects the stability of the DC voltage. If the arc-extinguishing angle is too small, the converter valve may not be able to shut off completely, leading to abnormal current fluctuations and affecting power quality. The larger the DC current, the longer the converter valve conduction time, and the higher the temperature, which may lead to converter valve overheating. Excessively high converter valve temperatures will reduce the reliability of the converter valve and may even cause damage, resulting in abnormal voltage and current fluctuations. High harmonic content may be caused by abnormal operation of the converter station converter valve (such as improper firing angle or arc-extinguishing angle control), which will lead to additional current fluctuations and affect equipment safety. Degraded insulation performance (lower line insulation test score) leads to voltage waveform distortion and exacerbates harmonic effects, causing abnormal equipment heating or electromagnetic interference. Excessive grounding current may indicate insulation damage, causing current leakage and affecting the stability of DC current and voltage. If the DC current remains excessively high for a prolonged period, the converter valve temperature will continue to rise, potentially leading to insulation aging and consequently affecting line insulation performance (lower line insulation test score). Poor insulation performance results in leakage current, increasing ground leakage current. If the grounding current continues to increase, it may threaten the safe operation of equipment and even affect the operational stability of the converter station.

[0094] It should also be noted that, in this embodiment of the invention, the voltage parameter unification factor, current parameter unification factor, active power parameter unification factor, harmonic content parameter unification factor, temperature parameter unification factor, and line insulation test scoring parameter unification factor are adjustment coefficients used to eliminate mismatches between different parameters, enabling multiple parameters with different physical units or types to perform linear calculations.

[0095] When the abnormal fluctuation risk value exceeds the abnormal fluctuation risk threshold, it is determined that an abnormal fluctuation has occurred in the DC transmission line. The time when the DC transmission line parameters are acquired is recorded as the abnormal time. Based on the abnormal fluctuation risk value, a mapping and matching is performed with the temporary line protection measures corresponding to each abnormal fluctuation risk value range pre-stored in the built-in database of the data processing center to obtain the temporary line protection measures for that abnormal fluctuation risk value. The temporary line protection is then activated. In this embodiment of the invention, the temporary line protection measures include switching to a backup line, reclosing control, and dynamic load adjustment, specifically including:

[0096] When the abnormal fluctuation risk value is greater than or equal to the highest risk threshold, it is determined that the abnormal fluctuation risk value is in the first interval, and is matched with the temporary line protection measures set in the first interval and the temporary line protection is activated. In this embodiment of the invention, the temporary line protection measure of switching to the backup route is adopted.

[0097] When the abnormal fluctuation risk value is less than the highest risk threshold and greater than or equal to the medium risk threshold, it is determined that the abnormal fluctuation risk value is in the second interval, and is matched with the temporary line protection measures set in the second interval and the temporary line protection is activated. In this embodiment of the invention, the temporary line protection measures using reclosing control are adopted.

[0098] When the abnormal fluctuation risk value is less than the medium risk threshold but greater than the abnormal fluctuation risk threshold, it is determined that the abnormal fluctuation risk value is in the third interval, and is matched with the temporary line protection measures set in the third interval and the temporary line protection is activated. In this embodiment of the invention, the temporary line protection measures using dynamic load adjustment are adopted.

[0099] It should be noted that in DC transmission lines, temporary line protection measures are protective mechanisms implemented quickly in the event of abnormal fluctuations to ensure stable system operation and prevent wider-ranging faults. Switching to a backup line refers to the system automatically or manually switching to a pre-designated backup power transmission line. This is a protective measure that prevents the fault from spreading to other parts by changing the direction of power flow.

[0100] Reclosing control refers to the process of automatically reclosing power lines after a short-term outage when a power line experiences a brief fault. Specifically, after a line fault occurs, the system first disconnects the line, then waits a few seconds (usually hundreds of milliseconds to a few seconds). If the fault has disappeared, it automatically recloses the circuit to restore power. Some faults may be temporary, such as contact problems caused by external factors like lightning or wind blowing tree branches. Reclosing can effectively resolve these temporary faults without requiring a prolonged power outage. Through reclosing control, power supply can be quickly restored, reducing outage time and impact on users.

[0101] Dynamic load regulation refers to reducing system pressure by controlling the load of a DC transmission line when abnormal fluctuations occur. It typically involves adjusting the DC transmission line based on load reduction values ​​to decrease the supply of some load and maintain system stability. When a DC transmission line faces overload risk, dynamic load regulation can alleviate pressure by temporarily reducing some load, preventing the DC transmission line from collapsing.

[0102] It should be noted that, in this embodiment of the invention, the abnormal fluctuation risk value is obtained by subtracting the abnormal fluctuation risk threshold from the abnormal fluctuation risk value. This abnormal fluctuation difference is then mapped and matched with the load reduction values ​​corresponding to the abnormal fluctuation differences pre-stored in the built-in database of the data processing center to obtain the load reduction value of the DC transmission line.

[0103] Real-time acquisition of DC transmission line parameters also includes sensor detection and removal of internal electromagnetic interference from the sensors. Specific processing conditions are as follows:

[0104] Monitor sensor power data during abnormal moments, including voltage change rate, current change rate, and electromagnetic interference spectrum.

[0105] Voltage change rate is used to determine whether it is a high-frequency interference source, current change rate is used to determine whether electromagnetic interference has radiative interference characteristics, and electromagnetic interference spectrum is used to determine the location of the interference source.

[0106] Based on sensor power data at abnormal times, the data is imported into the data processing center for feature extraction. After confirming the characteristics of electromagnetic interference, it is mapped and matched with the preset filtering and denoising methods corresponding to each feature in the data processing center's built-in database to obtain the filtering algorithm corresponding to the electromagnetic interference characteristics for electromagnetic interference elimination processing. The specific process is as follows:

[0107] Monitoring the voltage and current change rates of sensors during abnormal moments is crucial. The voltage change rate reflects the speed of voltage fluctuations. A rapid voltage change rate can indicate the presence of high-frequency electromagnetic interference sources. Generally, high-frequency interference sources cause rapid voltage changes, manifesting as rapid voltage fluctuations.

[0108] The rate of change of current reflects how quickly the current changes. By monitoring the rate of change of current, it is possible to identify whether current fluctuations exhibit radiated interference characteristics. For example, electromagnetic interference can cause sudden current changes, manifesting as abrupt current fluctuations. Radiated interference typically exhibits strong instantaneous current changes.

[0109] Abnormal high-frequency components in the spectrum can be used to identify the characteristics and location of electromagnetic interference sources. By comparing spectral characteristics, the frequency range, intensity, and location of the interference source can be determined.

[0110] Based on voltage change rate, current change rate, and electromagnetic interference spectrum, the data is fed into the data processing center for analysis. The analysis process includes: if the voltage change rate is large and a significant high-frequency signal appears in the spectrum, this indicates the presence of an external or internal high-frequency electromagnetic interference source, such as the switching action of electrical equipment or lightning. If the current change rate shows a significant abrupt change and the spectrum displays a strong interference signal, this is caused by a radiated interference source (such as a transformer or converter). Spectrum analysis can determine the frequency range of the electromagnetic interference source and further analyze its origin. For example, interference in a specific frequency band may be related to specific equipment (such as high-frequency equipment or switching equipment), helping to locate the interference source.

[0111] Import the data into the data processing center to extract time-domain and frequency-domain features. Electromagnetic interference features are clustered using pattern recognition (such as K-means clustering). In this embodiment, these clusters include power frequency interference, high-frequency switching noise, harmonic interference, and arc interference. The extracted electromagnetic interference features are matched with denoising filters in the data processing center's built-in database to obtain the corresponding filtering algorithms. These algorithms are then applied to denoise the sensor data.

[0112] Record the quality data of DC transmission line parameters before and after filtering, including harmonic content, signal-to-noise ratio, and mean square error.

[0113] The filtered DC transmission line parameters are re-imported into the data processing center. The quality data of the DC transmission line parameters before and after filtering are compared to verify whether the electromagnetic interference has been effectively eliminated. Specifically, this includes:

[0114] The quality data of the filtered DC transmission line parameters, including harmonic content, signal-to-noise ratio, and mean square error, are compared with the quality data of the DC transmission line parameters before filtering to obtain the electromagnetic interference elimination evaluation value of the DC transmission line parameters. The specific processing conditions are as follows:

[0115] ;

[0116] in, The electromagnetic interference elimination evaluation values ​​for DC transmission line parameters are as follows: The harmonic content after filtering. This represents the signal-to-noise ratio after filtering. This represents the mean square error after filtering. The harmonic content before filtering. The signal-to-noise ratio before filtering. This represents the mean square error before filtering.

[0117] The electromagnetic interference elimination assessment value based on the DC transmission line parameters is compared with the preset electromagnetic interference elimination assessment threshold. If the electromagnetic interference elimination assessment value of the DC transmission line parameters is greater than or equal to the electromagnetic interference elimination assessment threshold, it is determined that the electromagnetic interference has not been effectively eliminated. If the electromagnetic interference elimination assessment value of the DC transmission line parameters is less than the electromagnetic interference elimination assessment threshold, it is determined that the electromagnetic interference has been effectively eliminated.

[0118] If the electromagnetic interference is effectively eliminated, the filtered DC transmission line parameters will be further analyzed and processed.

[0119] If the electromagnetic interference is not effectively eliminated, an early warning will be sent to the data management terminal.

[0120] Filtering algorithms include bandpass filtering, wavelet transform denoising, adaptive filtering, and morphological filtering. The parameters of the DC transmission line at abnormal times are imported into the data processing center for feature extraction. The specific process is as follows:

[0121] The parameters of the DC transmission lines at abnormal times are collected into the data processing center and then standardized.

[0122] Each time-domain feature and each frequency-domain feature is extracted separately. After principal component analysis of each time-domain feature and each frequency-domain feature, the key features of the DC transmission line parameters are obtained. The key features are then imported into the fault diagnosis processor built into the data processing center for fault diagnosis analysis.

[0123] It should be noted that the purpose of standardization is to transform data with different dimensions and ranges into a uniform scale for subsequent analysis. A common standardization method is to subtract the mean from each feature value and divide by the standard deviation, so that the data has zero mean and unit variance.

[0124] Time-domain features refer to the features extracted from time-series data of the original signal, including mean, variance, peak value, skewness, kurtosis, etc. These features can be used to obtain information about the signal's volatility, trend, and periodicity.

[0125] Frequency domain features are characteristics obtained by performing frequency domain analysis on a signal (such as Fast Fourier Transform, FFT), including the signal's spectrum, the intensity of frequency components, harmonic components, and bandwidth. These features help identify periodic patterns and interference in the signal.

[0126] Principal component analysis (PCA) is a commonly used dimensionality reduction technique. It projects the original features into a new space through linear transformation, generating new "principal components." These principal components are linear combinations of the data and have the largest variance. The goal of PCA is to reduce the dimensionality of the data while preserving its key information.

[0127] Principal component analysis yields the key components that constitute the parameters of a DC transmission line. These key features effectively reflect the overall behavior of the DC transmission line and help detect potential faults or anomalies.

[0128] When the feature value of a key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature and the corresponding fault types are matched.

[0129] When the feature value of a key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature, and the corresponding fault types are matched to obtain the valid feature, specifically including:

[0130] After receiving each key feature, the fault diagnosis processor built into the data processing center calculates the feature values ​​of each key feature, specifically including:

[0131] Key features in this embodiment of the invention include, but are not limited to, voltage surge, voltage drop, DC current surge, DC current drop, DC bus potential drift, IGBT temperature rise, enhanced partial discharge, and IGBT temperature rise.

[0132] The mean, root mean square value, peak-to-peak value, kurtosis, and skewness of each feature data under each key feature are extracted for time-domain feature calculation to obtain the time-domain feature value of each key feature.

[0133] By performing a Fourier transform on the signal, the intensity of each frequency component in the signal is calculated, and frequency domain features are calculated to obtain the frequency domain feature values ​​of each key feature.

[0134] By combining the time-domain and frequency-domain feature values ​​of each key feature, the feature values ​​of each key feature are calculated in the fault diagnosis processor built into the data processing center.

[0135] It should be noted that methods for synthesizing the time-domain and frequency-domain feature values ​​of various key features include feature normalization, feature fusion, and weighted fusion. In this embodiment of the invention, weighted fusion is selected to comprehensively process the time-domain and frequency-domain feature values ​​of each key feature. Deep learning is performed using the neural network of the data processing center to automatically obtain the weights of the time-domain and frequency-domain feature values ​​of each key feature. The feature value of a certain key feature is calculated, and the specific process includes:

[0136] ;

[0137] ;

[0138] in, This is the initial matrix of the time-domain features of this key feature. This is the initial matrix of the frequency domain features of this key feature. The mean of this key feature is used to measure the average level of the signal. RMS is the root mean square value of this key feature, which is a representative indicator of the signal energy. This represents the peak-to-peak value of this key feature. Used to describe the maximum amplitude range. For the kurtosis of this key feature, Skewness is used to measure the degree of abrupt changes in a signal and detect impulse characteristics. Skewness is the skewness of this key feature and is used to describe the symmetry of the signal. The dominant frequency of this key feature is used to characterize the frequency component with the highest energy in the signal. The spectral energy of this key feature is used to characterize the total energy of the signal in the frequency domain. The harmonic content of this key feature is used to analyze the amplitude ratio of the fault characteristic frequency. The PSD is the power spectral density of this key feature, used to describe the power distribution of the signal at different frequencies. The SN is the spectral entropy of this key feature, used to measure whether the energy distribution of the spectrum is uniform.

[0139] After normalization, the normalized matrix is ​​obtained, which includes:

[0140] ;

[0141] ;

[0142] in, This is the time-domain feature normalization matrix of this key feature. This is the frequency domain feature normalization matrix of this key feature. This is the normalized mean of the key feature. This is the normalized root mean square value of the key feature. This represents the normalized peak value of this key feature. This is the normalized kurtosis of the key feature. This represents the normalized skewness of the key feature. This is the normalized principal frequency of this key feature. This is the normalized spectral energy of this key feature. This represents the normalized harmonic content of this key feature. This is the normalized power spectral density of this key feature. This is the normalized spectral entropy of this key feature.

[0143] Deep learning training using neural networks automatically calculates the optimal weights for time-domain and frequency-domain features:

[0144] ;

[0145] The neural network is trained using a multilayer perceptron (MLP), and the weights are determined by optimizing the objective function. NN stands for Neural Network Model. For time-domain feature weights, For frequency domain feature weights.

[0146] ;

[0147] in, This is the fusion feature matrix of the key feature. This is the time-domain feature normalization matrix of this key feature. This is the frequency domain feature normalization matrix of this key feature. For time-domain feature weights, For frequency domain feature weights.

[0148] Based on the fusion feature matrix of this key feature, the feature values ​​of this key feature are calculated, specifically including:

[0149] ;

[0150] in, This is the fusion feature matrix of the key feature. The feature value of this key feature. It is the identity matrix. It is a determinant.

[0151] Solve the above equation to obtain the eigenvalues ​​of this key feature.

[0152] The feature thresholds of each key feature are extracted from the built-in database of the data processing center. The feature values ​​of each key feature are compared with their corresponding feature thresholds. When the feature value of a key feature is greater than its corresponding feature threshold, the key feature is determined to be a valid feature. The fault types corresponding to the valid feature are extracted and counted from the built-in database of the data processing center.

[0153] In this embodiment of the invention, the correspondence between fault types and valid features is shown in Table 1 below.

[0154] Table 1. Correspondence between Fault Type and Valid Feature

[0155]

[0156] As shown in Table 1, each fault type typically corresponds to one or more valid features.

[0157] When the feature value of a key feature is less than or equal to its corresponding feature threshold, the key feature is determined to be an invalid feature, and fault diagnosis is performed based on the determination results of each key feature.

[0158] Fault diagnosis is performed based on the determination results of each key feature, specifically including:

[0159] When the feature values ​​of each key feature are all less than their corresponding feature thresholds, i.e., all are invalid features, the abnormal moment is recorded, the DC transmission line is switched to a pending state, and a short-term monitoring cycle is set for trial operation.

[0160] The DC transmission line is monitored during trial operation to obtain electrical test parameters within a short monitoring period. These parameters are then input into the data processing center for feature extraction. Key features are obtained, and it is determined whether each key feature is valid, thus yielding the trial operation monitoring results of the DC transmission line.

[0161] The electrical test parameters for trial operation during the short-term monitoring period specifically include the DC voltage, DC current, active power, and harmonic content during the trial operation during the short-term monitoring period.

[0162] If the test operation monitoring results of the DC transmission line are all invalid, the DC transmission line is determined to be fault-free and switched to normal operation.

[0163] If the trial operation monitoring results of the DC transmission line show the presence of valid features, then the DC transmission line is determined to have a fault, and the fault types corresponding to each valid feature are statistically analyzed.

[0164] The correlation characteristics of each fault type are statistically analyzed to obtain the fault diagnosis confidence score of each fault type, and then compared with the fault diagnosis confidence threshold to obtain the fault diagnosis result.

[0165] Each valid feature corresponding to each fault type is statistically analyzed into associated features for each fault type, and a fault index is obtained for each associated feature. Based on the fault index of each associated feature and the number of associated features for each fault type, the coupling process is corrected to obtain the fault diagnosis confidence score for each fault type, specifically including:

[0166] The fault indices for each associated feature are obtained, specifically including:

[0167] The feature difference of each associated feature is obtained by subtracting its feature value from its corresponding feature threshold. Based on the feature difference of each associated feature, the fault index of each associated feature is obtained by mapping and matching with the fault index corresponding to each feature difference in the built-in database of the data processing center. The fault index is used to assess the severity of deviation of each associated feature, that is, the probability of causing a fault. The larger the feature difference of a certain associated feature, the greater the probability that the associated feature will cause a fault.

[0168] The number of associated features for each fault type is counted and mapped and matched with the number of associated features corresponding to the intervals in the database built into the data processing center. The number of associated features for each fault type is obtained. The number of associated features is directly proportional to the number of associated features. The more associated features a fault type has, the greater the confidence in the fault diagnosis of that fault type.

[0169] The fault diagnosis confidence score for each fault type is obtained by correcting the coupling process, specifically including:

[0170] ;

[0171] in, For the first Fault diagnosis confidence score for each fault type For the first The first fault type Fault index of each associated feature, For the quantity factor of associated features, For associated feature numbering, , For the first The number of associated features matched for each fault type. Number the fault type. , This represents the total number of fault types.

[0172] The fault diagnosis result is obtained by comparing it with the fault diagnosis confidence threshold, which specifically includes:

[0173] The fault diagnosis confidence score for each fault type is compared with the fault diagnosis confidence threshold. If the fault diagnosis confidence score for a certain fault type is greater than or equal to the fault diagnosis confidence threshold, then the fault type is determined to be a significant fault.

[0174] If the fault diagnosis confidence score of a certain fault type is less than the fault diagnosis confidence threshold, then the fault type is determined to be a potential fault.

[0175] The fault diagnosis results were obtained, and the specific processing conditions were as follows:

[0176] The fault types that are statistically significant in the fault diagnosis results are denoted as each significant fault type. Based on the magnitude of the fault diagnosis confidence score of each significant fault type, the significant fault types are sorted from largest to smallest and sent to the data management terminal for early warning.

[0177] If the fault diagnosis results for each fault type are all potential faults, then based on the total number of fault types, the dynamic monitoring cycle is obtained by matching it with the preset dynamic monitoring cycle corresponding to the total number of each fault type in the database built into the data processing center. Dynamic monitoring of the DC transmission line is then performed based on this dynamic monitoring cycle, specifically including:

[0178] Based on the dynamic monitoring cycle, real-time monitoring data of DC transmission lines are obtained within the dynamic monitoring cycle, and the tracking and analysis results of DC transmission lines are obtained.

[0179] If, during the dynamic monitoring period, the associated characteristic value of a potential fault exceeds its corresponding characteristic threshold, the tracking analysis result of the DC transmission line will be determined as a tracking fault, and the fault type will be determined as a significant fault, and sent to the data management terminal for early warning.

[0180] If the associated characteristic values ​​of each potential fault do not exceed their corresponding characteristic thresholds within the dynamic monitoring period, the tracking and analysis results of the DC transmission line will be judged as short-term fluctuations, the dynamic monitoring period will be terminated, and the normal monitoring period will be switched.

[0181] As shown in Figure 2, in this embodiment, the present invention provides a DC transmission fault diagnosis system based on multimodal data analysis, comprising:

[0182] The real-time monitoring module is used to monitor DC transmission lines in real time. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated, and the DC transmission line parameters at the abnormal moment are imported into the data processing center for feature extraction.

[0183] The fault type matching module is used to record a key feature as a valid feature and match the fault types corresponding to the valid feature when the feature value of a key feature is greater than its corresponding feature threshold.

[0184] The fault type judgment module is used to statistically analyze the associated features of each fault type, obtain the fault diagnosis confidence score of each fault type, and compare it with the fault diagnosis confidence threshold to obtain the fault diagnosis result.

[0185] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0186] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementation methods. Clearly, many modifications and variations can be made based on the content of this specification. The selection and detailed description of these embodiments in this specification are intended to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. Any modifications or variations that do not deviate from the structure of the invention or exceed the scope defined by the invention should fall within the protection scope of the invention.

Claims

1. A DC transmission fault diagnosis method based on multimodal data analysis, characterized in that, include: Real-time monitoring of DC transmission lines is conducted. When abnormal fluctuations occur in DC transmission lines, temporary line protection measures are activated. At the same time, the DC transmission line parameters at the abnormal moment are collected into the data processing center for feature extraction. When the feature value of a key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature and the corresponding fault types are matched. The correlation characteristics of each fault type are statistically analyzed to obtain the fault diagnosis confidence score of each fault type, and then compared with the fault diagnosis confidence threshold to obtain the fault diagnosis result.

2. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 1, characterized in that: The process involves real-time monitoring of the DC transmission line, and activation of temporary line protection measures when abnormal fluctuations occur in the DC transmission line. The specific procedure is as follows: Real-time acquisition of DC transmission line parameters, including electrical parameters, converter station control parameters, and equipment parameters; The electrical parameters include DC voltage, DC current, active power, and harmonic content. The converter station control parameters include the firing angle and the arc extinguishing angle; The equipment parameters include converter valve temperature, line insulation test score, and grounding current. The parameters of the DC transmission line are transmitted to the data processing center for analysis to obtain the abnormal fluctuation risk value. When the abnormal fluctuation risk value exceeds the abnormal fluctuation risk threshold, it is determined that the DC transmission line has experienced abnormal fluctuation. The time when the DC transmission line parameters are acquired is recorded as the abnormal time. Based on the abnormal fluctuation risk value, a mapping and matching is performed with the temporary line protection measures corresponding to each abnormal fluctuation risk value range pre-stored in the data processing center's built-in database to obtain the temporary line protection measures for that abnormal fluctuation risk value, and the temporary line protection is activated. The specific process includes: When the abnormal fluctuation risk value is greater than or equal to the highest risk threshold, it is determined that the abnormal fluctuation risk value is in the first interval, and the temporary line protection measures set in the first interval are matched and the temporary line protection is activated. When the abnormal fluctuation risk value is less than the highest risk threshold and greater than or equal to the medium risk threshold, the abnormal fluctuation risk value is determined to be in the second range, and is matched with the temporary line protection measures set in the second range and the temporary line protection is activated. When the abnormal fluctuation risk value is less than the medium risk threshold but greater than or equal to the abnormal fluctuation risk threshold, the abnormal fluctuation risk value is determined to be in the third interval, and is matched with the temporary line protection measures set in the third interval and the temporary line protection is activated.

3. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 1, characterized in that: The process of importing DC transmission line parameters at abnormal times into the data processing center for feature extraction is as follows: The parameters of the DC transmission lines at abnormal times are collected into the data processing center and then standardized. Each time-domain feature and each frequency-domain feature is extracted separately. After principal component analysis of each time-domain feature and each frequency-domain feature, the key features of the DC transmission line parameters are obtained. The key features are then imported into the fault diagnosis processor built into the data processing center for fault diagnosis analysis.

4. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 3, characterized in that: When the feature value of a key feature extracted is greater than its corresponding feature threshold, the key feature is recorded as a valid feature and the corresponding fault types are matched, specifically including: After receiving each key feature, the fault diagnosis processor built into the data processing center calculates the feature value of each key feature and extracts the feature threshold of each key feature from the built-in database of the data processing center. It compares the feature value of each key feature with its corresponding feature threshold. When the feature value of a key feature is greater than its corresponding feature threshold, the key feature is determined to be a valid feature, and the fault types corresponding to the valid feature are counted. When the feature value of a key feature is less than or equal to its corresponding feature threshold, the key feature is determined to be an invalid feature, and fault diagnosis is performed based on the determination results of each key feature.

5. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 4, characterized in that: The fault diagnosis based on the determination results of each key feature specifically includes: When all key features are invalid, the abnormal moment is recorded, the DC transmission line is switched to pending state, and a short-term monitoring cycle is set for trial operation. The DC transmission line is monitored during trial operation to obtain electrical test parameters during a short monitoring period. These parameters are then input into the data processing center for feature extraction. Key features are obtained and it is determined whether each key feature is valid, thus yielding the trial operation monitoring results of the DC transmission line. If the test operation monitoring results of the DC transmission line are all invalid, the DC transmission line is determined to be fault-free and switched to normal operation. If the trial operation monitoring results of the DC transmission line show the presence of valid features, then the DC transmission line is determined to have a fault, and the fault types corresponding to each valid feature are statistically analyzed.

6. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 1, characterized in that: The statistical analysis of the correlation features of each fault type yields a fault diagnosis confidence score for each fault type. Specifically, this includes: Each effective feature corresponding to each fault type is statistically analyzed into each associated feature of each fault type, and the fault index of each associated feature is obtained. Based on the fault index of each associated feature and the number of associated features of each fault type, the coupling process is corrected to obtain the fault diagnosis confidence score of each fault type. The failure index obtained for each associated feature specifically includes: The feature difference of each associated feature is obtained by subtracting its feature value from its corresponding feature threshold. Based on the feature difference of each associated feature, the fault index of each effective feature is obtained by mapping and matching it with the fault index corresponding to each feature difference in the built-in database of the data processing center.

7. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 3, characterized in that: The comparison with the fault diagnosis confidence threshold is used to obtain the fault diagnosis result, specifically including: The fault diagnosis confidence score of each fault type is compared with the fault diagnosis confidence threshold. If the fault diagnosis confidence score of a certain fault type is greater than or equal to the fault diagnosis confidence threshold, then the fault type is determined to be a significant fault. If the fault diagnosis confidence score of a certain fault type is less than the fault diagnosis confidence threshold, then the fault type is determined to be a potential fault.

8. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 7, characterized in that: The specific processing conditions for obtaining the fault diagnosis results are as follows: The fault types that are identified as significant faults in the fault diagnosis results are recorded as significant fault types. Based on the fault diagnosis confidence score of each significant fault type, the significant fault types are sorted and sent to the data management terminal for early warning. If the fault diagnosis results for each fault type are all potential faults, then based on the total number of fault types, the dynamic monitoring cycle is obtained by matching it with the preset dynamic monitoring cycle corresponding to the total number of each fault type in the database built into the data processing center. Dynamic monitoring of the DC transmission line is then performed based on this dynamic monitoring cycle, specifically including: Based on the dynamic monitoring cycle, real-time monitoring data of DC transmission lines are obtained within the dynamic monitoring cycle, and trend analysis is performed on the data to obtain the tracking analysis results of DC transmission lines. If, during the dynamic monitoring period, the associated characteristic value of a potential fault continues to rise and exceeds its corresponding characteristic threshold, the tracking analysis result of the DC transmission line will be determined as a tracking fault, and the fault type will be determined as a significant fault, and sent to the data management terminal for early warning. If the associated characteristic values ​​of each potential fault remain stable and do not exceed their corresponding characteristic thresholds during the dynamic monitoring period, the tracking and analysis results of the DC transmission line will be judged as short-term fluctuations, the dynamic monitoring period will be terminated, and the normal monitoring period will be switched.

9. The DC transmission fault diagnosis method based on multimodal data analysis according to claim 2, characterized in that: The real-time acquisition of DC transmission line parameters also includes detecting the sensors and removing the influence of electromagnetic interference inside the sensors. The specific processing conditions are as follows: Monitor sensor power data during abnormal moments, including voltage change rate, current change rate, and electromagnetic interference spectrum; The voltage change rate is used to determine whether it is a high-frequency interference source, the current change rate is used to determine whether the electromagnetic interference has radiation interference characteristics, and the electromagnetic interference spectrum is used to determine the location of the interference source. Based on the sensor power data at abnormal times, the data is imported into the data processing center for feature extraction. After confirming the characteristics of electromagnetic interference, it is mapped and matched with the filtering algorithms corresponding to each feature in the built-in database of the data processing center to obtain the filtering algorithm corresponding to the electromagnetic interference characteristics for electromagnetic interference elimination processing. Record the quality data of DC transmission line parameters before and after filtering, including harmonic content, signal-to-noise ratio, and mean square error; The filtered DC transmission line parameters are re-imported into the data processing center to compare the quality data of the DC transmission line parameters before and after filtering, and to verify whether the electromagnetic interference has been effectively eliminated. If the electromagnetic interference is effectively eliminated, the filtered DC transmission line parameters will be further analyzed and processed. If the electromagnetic interference is not effectively eliminated, an early warning will be sent to the data management terminal.

10. A DC transmission fault diagnosis system based on multimodal data analysis, employing the DC transmission fault diagnosis method based on multimodal data analysis as described in any one of claims 1-9, characterized in that: The real-time monitoring module is used to monitor DC transmission lines in real time. When abnormal fluctuations occur in the DC transmission lines, temporary line protection measures are activated, and the DC transmission line parameters at the abnormal moment are imported into the data processing center for feature extraction. The fault type matching module is used to record the key feature as a valid feature and match the fault types corresponding to the valid feature when the feature value of a certain key feature is greater than its corresponding feature threshold. The fault type judgment module is used to statistically analyze the associated features of each fault type, obtain the fault diagnosis confidence score of each fault type, and compare it with the fault diagnosis confidence threshold to obtain the fault diagnosis result.