Cable online fault diagnosis and positioning method and system based on multi-modal sensing fusion

Abstract

The invention belongs to the technical field of power equipment monitoring, and particularly relates to a cable online fault diagnosis and positioning method and system based on multi-modal sensing fusion. The method comprises the following steps: collecting a vibration signal and a temperature signal in a cable operation process, and extracting various characteristic parameters; before the cable is abnormal, the correlation between different parameters is analyzed, the temperature is predicted in combination with load data, and the temperature prediction deviation is calculated; judging whether the cable is in a normal, overload or fault evolution state or not by comparing the difference quantity with a preset threshold value; early warning is triggered in the fault evolution state, the fault type is further recognized, and the damage position is positioned based on specific parameters. The invention further provides a system for implementing the method. Real-time evaluation and accurate positioning of the state of the cable can be achieved.

Description

Technical Field

[0001] This invention belongs to the field of power equipment monitoring technology, specifically relating to a method and system for online cable fault diagnosis and location based on multimodal sensor fusion. Background Technology

[0002] With the rapid development of complex power supply scenarios such as urban power grids, rail transit, and industrial parks, cables, as the core carrier of power transmission, have a significant impact on the safe operation of the entire power system due to their operational stability. In recent years, frequent cable accidents have been related to problems such as overload, insulation aging, and localized mechanical damage. Therefore, achieving online monitoring and fault identification of cable operating status has become a key aspect of power operation and maintenance.

[0003] Current cable monitoring methods mostly rely on single sensing quantities, such as using temperature sensors to detect thermal instability risks or vibration sensors to identify anomalies like partial discharge and mechanical loosening. However, in actual operation, cable conditions are affected by multiple coupling factors, and a single signal cannot accurately determine whether a cable has entered an abnormal evolution stage, leading to frequent misjudgments and missed detections in existing solutions. Furthermore, traditional technologies mostly rely on fixed thresholds to determine operating status, lacking dynamic adaptation mechanisms and failing to effectively reflect the changing trends of cable conditions under different loads, environments, or aging levels.

[0004] Furthermore, existing fault diagnosis and location methods largely rely on manual judgment or empirical rules, which have limited ability to handle abnormal events in complex scenarios involving multiple types of faults. Therefore, there is an urgent need to provide a cable online monitoring, early warning, and fault diagnosis method based on multi-source sensor information fusion, with the capabilities of correlation identification, trend judgment, and precise location, in order to improve the intelligent operation and maintenance level of power systems and the full life cycle management capabilities of cables. Summary of the Invention

[0005] To address the above problems, the present invention aims to propose a method for online cable fault diagnosis and location based on multimodal sensor fusion, comprising the following steps:

[0006] S1. Cable data acquisition and classification processing: Acquire cable operating status data, which includes vibration signals and temperature signals, and classify them according to data type to generate feature parameters;

[0007] S2. Correlation analysis and prediction deviation calculation before the anomaly: Extract key nodes of characteristic parameters before the cable anomaly occurs as reference information, calculate the correlation between reference information as parameters to be analyzed; obtain cable load data, and use linear regression or time series extrapolation methods to predict cable temperature values; then calculate the temperature prediction deviation based on the difference between the predicted temperature value, the actual temperature value and the parameters to be analyzed.

[0008] S3. Real-time cable status monitoring and fault evolution identification: Compare the real-time temperature data with the temperature prediction deviation and calculate the difference between the two; based on the difference and combined with the preset comparison rules, determine whether the cable is in a normal state, an overload state, or a fault evolution state.

[0009] S4. Operation Status Determination and Risk Warning: When the cable is in a normal or overloaded state, the operation status result is output; when it is determined to be in a fault evolution state, a risk warning is triggered and the vibration signal in the fault evolution state is recorded as the parameter to be measured.

[0010] S5. Fault type identification and damage location: Based on the classification results of the parameters to be measured, the fault type is determined, and then the specific damage location of the cable is determined based on the key node location parameters corresponding to the fault type.

[0011] As a preferred technical solution, in step S1, the characteristic parameters include a first characteristic parameter and a second characteristic parameter; the first characteristic parameter is a spectral characteristic characterizing the local mechanical vibration change of the cable, including amplitude spectrum, envelope energy and peak distribution of a specific frequency band, which is extracted from the vibration signal using fast Fourier transform and wavelet packet decomposition; the second characteristic parameter is a time-series temperature statistical characteristic characterizing the thermal state of the cable, including average temperature, slope, standard deviation and transient rate of change, which is extracted from the temperature signal using sliding window statistics and multi-scale time sampling method.

[0012] As a preferred technical solution, step S2, calculating the correlation between reference information includes:

[0013] By setting several sampling points within a time period, the instantaneous values ​​of the first feature parameter and the second feature parameter are extracted as reference information.

[0014] The correlation between all reference information within this time period is calculated. The correlation is obtained using the Pearson correlation coefficient method and is used as input for subsequent temperature prediction bias.

[0015] As a preferred technical solution, step S2, calculating the temperature prediction deviation, includes the following steps:

[0016] Based on cable load data and historical temperature change trends, a weighted moving average or regression algorithm is used to predict the cable temperature at a specified time point; the actual temperature at that time point is obtained; the predicted temperature is compared with the actual temperature, and the difference is taken as the temperature prediction deviation.

[0017] As a preferred technical solution, in step S2, determining the temperature prediction deviation based on the relationship between the predicted value, the actual temperature value, and the parameter to be analyzed specifically includes:

[0018] Temperature prediction deviation, predicted temperature value, and reference information of the cable are extracted over multiple time periods to construct three sets of data. By setting the functional relationship between the target variable and the input variable, a calculation expression for evaluating the accuracy of temperature prediction is established using the least squares method, interpolation fitting, or polynomial regression method. The expression is used to quantify the changing trend of the prediction error and is used for subsequent difference calculation.

[0019] As a preferred technical solution, in step S3, the operating status is determined by judging the range of the difference and combining it with the following judgment rules:

[0020] Zone 1: Indicates that the cable is in a normal state. The judgment conditions are: the difference is less than the first threshold T1, and the temperature change rate and vibration amplitude are both within their respective set normal operation fluctuation ranges.

[0021] Region 2: This indicates that the cable is in an overload state. The judgment criteria are: the difference is between the first threshold T1 and the second threshold T2, and the rate of temperature change is higher than the upper limit of normal while the vibration characteristic parameters are still within the normal range, indicating that the cable has a trend of increasing heat load.

[0022] Region 3: This indicates that the cable has entered a fault evolution state. The judgment condition is: the difference is greater than the second threshold T2, and at the same time, a sharp rise in temperature or abnormal violent vibration is detected, indicating that the cable is undergoing a deterioration process caused by thermal effects or mechanical disturbances.

[0023] As a preferred technical solution, in step S3, based on the difference and in conjunction with preset comparison rules, during the process of determining whether the cable is in a normal state, an overload state, or a fault evolution state,

[0024] The upper threshold of Region 1, the lower threshold of Region 3, and their difference are fixed. The range of Region 2 is dynamically adjusted according to the actual operating conditions. The dynamic adjustment includes:

[0025] As the data collection time increases, the judgment period is extended to reduce the probability of misjudgment.

[0026] Collect more samples and recalculate the mean and standard deviation of the variance to dynamically update the judgment boundary;

[0027] The threshold range is weighted and adjusted according to the real-time cable load level to reflect the actual operating pressure.

[0028] As a preferred technical solution, in step S4, after recording the parameter to be measured, the difference is compared with the reference information to obtain the difference value, and then compared with the comparison threshold to determine its category, including:

[0029] If the difference value is within the range of the first category, it corresponds to an overload fault.

[0030] If it falls within the second category range, it corresponds to fatigue-related faults;

[0031] If the second category difference range is met again on top of the first category of faults, then the cable can be further confirmed to be in a composite fault state.

[0032] As a preferred technical solution, in step S5: for overload faults and compound faults, the temperature gradient and peak distribution in the temperature parameters are used as the basis for fault location; for fatigue faults, the peak frequency point and vibration duration in the vibration parameters are used as the basis for fault location.

[0033] This invention also provides a cable online fault diagnosis and location system based on multimodal sensor fusion, used to implement the method, including:

[0034] The data processing module is used to classify cable operating status data and extract temperature and vibration characteristic parameters;

[0035] The analysis and evaluation module is used to extract reference information and calculate its correlation, obtain the relationship between cable load and temperature, perform temperature prediction and prediction deviation calculation;

[0036] The status determination module is used to determine the cable operating status type based on the difference and a preset threshold.

[0037] The diagnostic location module is used to determine the fault type based on the classification of the parameter to be measured and select the corresponding vibration or temperature location parameters to locate the cable fault location.

[0038] Beneficial effects:

[0039] This invention utilizes multimodal sensing fusion to collect temperature and vibration signals during cable operation. It then classifies and extracts features from different data types, constructing multiple sets of key parameters, including spectral and statistical features. By analyzing these parameters at critical points before cable anomalies occur, correlation indices between the two types of signals can be obtained, providing rich foundational information for judging the cable's state evolution trend.

[0040] This invention predicts the trend of cable temperature change based on the relationship between load data and reference features, combined with sampling points within a time period, and compares this prediction with the actual temperature value to form a temperature prediction deviation. Subsequently, by calculating the difference and dividing the range, the cable operating state is divided into three situations: normal, overload, and fault evolution. An adaptive threshold adjustment mechanism is introduced, which can dynamically correct the judgment boundary based on the real-time operating environment, thereby improving the accuracy and stability of state identification.

[0041] During the fault diagnosis phase, this invention determines whether a fault is overload-related, fatigue-related, or a combination of both based on real-time difference characteristics and pre-defined classification intervals. For different types, temperature or vibration parameters are selected for fault location. The location method, based on parameters such as characteristic amplitude, time difference distribution, or frequency response, effectively identifies damaged areas of the cable, improving the timeliness and spatial resolution of fault identification. It is suitable for cable condition monitoring and maintenance deployment needs in complex operating environments. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0043] Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0044] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0045] Example 1

[0046] like Figure 1 As shown, this embodiment provides a method for online fault diagnosis and location of cables based on multimodal sensor fusion, applicable to the monitoring of operational status and fault identification of medium-voltage and higher voltage power cables in scenarios such as urban underground tunnels, industrial parks, rail transit, and data centers. The method includes the following steps:

[0047] S1. Cable Data Acquisition and Classification Processing:

[0048] First, multimodal sensing units are deployed at several key nodes along the cable. Each sensing unit includes both a vibration sensor (such as a MEMS accelerometer) and a temperature sensor (such as a PT100 platinum resistance thermometer) to acquire mechanical and thermal status information during cable operation. The sampling frequencies are set as follows: the vibration signal sampling frequency is 10 kHz to capture high-frequency partial discharge and mechanical shock; the temperature signal sampling frequency is 1 Hz to acquire steady-state and transient temperature rise changes.

[0049] All raw signals are uploaded to edge computing nodes for preliminary feature extraction and classification. Vibration signals are processed using Fast Fourier Transform (FFT) and wavelet packet decomposition to extract first feature parameters, including but not limited to envelope energy, center frequency, and peak distribution, to reflect local mechanical disturbances in the cable. Temperature signals are processed using sliding window statistics and multi-scale time decomposition to extract second feature parameters, including average temperature, temperature rise slope, local fluctuation amplitude, and transient rate of change, to characterize the cable's thermal load response. These processed feature parameters serve as the basis for subsequent analysis.

[0050] S2. Correlation analysis and prediction bias calculation before anomalies:

[0051] To identify potential latent anomalies during cable operation, key nodes in historical stable states during the cable's operating cycle are selected, and their corresponding first and second feature parameters are extracted as reference information. These data are then aligned and sampled within a specified time window (e.g., the past 30 minutes, divided into 60 sampling segments) to obtain the instantaneous values ​​of each feature, thus forming a reference information sequence.

[0052] Subsequently, the correlation between the first and second characteristic parameters was calculated using the Pearson correlation coefficient method. These parameters were used as the parameters to be analyzed, reflecting the changes in the coupling behavior between the mechanical and thermal signals.

[0053] In terms of temperature prediction, based on the cable load data and corresponding temperature change trends of the previous 10 minutes, a weighted moving average method is used to predict the cable temperature value at the next moment. The formula is as follows:

[0054]

[0055] in, Historical temperature values Weighting factors are set according to load changes.

[0056] Get the current actual temperature , and predicted temperature The comparison yielded the temperature prediction deviation. :

[0057]

[0058] In addition, the temperature prediction deviation, predicted temperature value and correlation parameters of the cable were extracted in multiple time periods to construct three sets of variable relationships. The prediction error trend was evaluated by constructing an expression using the least squares method, which provides a reference for the next step of calculating the difference.

[0059] S3. Real-time cable status monitoring and fault evolution identification:

[0060] The difference between the predicted temperature deviation and the current temperature value at each moment is analyzed and calculated as follows:

[0061]

[0062] This difference Used to determine the region to which the current running status belongs, and classified according to the following rules:

[0063] Region 1 (Normal): Difference Meanwhile, both the temperature change rate and the vibration amplitude are within the set range;

[0064] Zone 2 (Overload): Furthermore, the rate of temperature rise increased, but the vibration parameters showed no obvious abnormalities.

[0065] Area 3 (Fault Evolution): And there are obvious sudden changes in vibration peak, frequency concentration, or abnormal temperature jumps.

[0066] T1 is set as the first judgment threshold, and T2 is set as the second judgment threshold. The specific values ​​can be dynamically adjusted according to the on-site operating environment. To enhance adaptability, if a stable increase in load or an increase in false alarms is detected during operation, the system automatically extends the judgment period and recalibrates T1 and T2 based on the sample mean and variance to ensure the effectiveness of state division.

[0067] S4. Operational Status Determination and Risk Warning:

[0068] When the judgment result is "normal" or "overloaded," the system continues to output operating parameters in real time and record the changing trends. When the judgment is "fault evolution," the system immediately triggers an early warning, records the complete vibration data segment during the fault evolution period as the parameter to be measured, and uploads it to the backend server for further diagnostic analysis. The system can interface with SCADA systems or automated operation and maintenance platforms via industrial bus to achieve alarm linkage.

[0069] S5. Fault type identification and damage location:

[0070] The recorded vibration parameters to be measured are compared with the reference information. If the difference falls within the following ranges, the following judgments are made:

[0071] First category: Overload-related faults;

[0072] Second category: Fatigue-related faults;

[0073] Those that meet both types of difference indicators are: composite faults.

[0074] For overload and combined faults, the gradient distribution and local peak values ​​in the temperature characteristic parameters are selected as the location basis; for fatigue faults, the main peak frequency and envelope duration in the vibration spectrum are selected for location. Location is achieved by synchronously recording signal arrival times at multiple nodes, calculating the propagation time difference of the reflected wave, and combining this with the cable node number. For example:

[0075]

[0076] Where v is the signal propagation speed in the cable, This represents the time difference between the arrival of the vibration wave at the two sensors.

[0077] Finally, the system outputs the fault type and specific location segment number, and generates a diagnostic report for maintenance personnel to handle.

[0078] This embodiment can be adapted to cable monitoring tasks of various voltage levels, and is particularly suitable for operation status analysis and maintenance management in high temperature, humid, and long-distance laying scenarios. It has the advantages of flexible deployment, accurate identification, and fast response.

[0079] Example 2

[0080] This embodiment provides a cable online fault diagnosis and location system based on multimodal sensor fusion, used to implement comprehensive perception, intelligent analysis, and accurate diagnosis of the operating status of power cables. By fusing temperature and vibration signals, the system achieves real-time assessment of cable operating status and identification of fault evolution trends, making it particularly suitable for complex operation and maintenance needs in urban underground cables, tunnel-laid cables, high-temperature power supply lines, and industrial scenarios.

[0081] The system mainly consists of four core modules: a data processing module, an analysis and evaluation module, a status judgment module, and a diagnostic and location module. Each module can be deployed on edge computing nodes for local processing, or it can be centrally integrated into the main control platform for unified computation and control.

[0082] 1. Data Processing Module:

[0083] The data processing module is primarily used to acquire and initially process raw sensing data from cable operation. This module collects vibration and temperature signals from the cable through distributed multimodal sensing units. The vibration sensors, employing high-sensitivity accelerometers or piezoelectric elements, are installed on key cable nodes or supports to sense the mechanical responses generated when the cable is subjected to stress, partial discharge, friction, or breakdown. The temperature sensors, using high-precision resistive or semiconductor thermistors, are adhered to or embedded in the cable's outer sheath to monitor real-time changes in the thermal state of the cable conductor or insulation.

[0084] The system performs standardized preprocessing on the acquired signals, including noise reduction, filtering, sampling alignment, and anomaly removal. Then, feature extraction is performed on the temperature and vibration signals respectively. For the temperature signal, thermal statistical features such as average temperature, rate of rise, standard deviation, and fluctuation amplitude are calculated over the time series using a sliding window method. For the vibration signal, mechanical features such as envelope energy, spectral distribution, and amplitude abrupt changes are extracted using frequency domain analysis and time-frequency analysis methods. These processed feature data are categorized and stored, and used as input for subsequent analysis.

[0085] In addition, the data processing module is equipped with a time synchronization mechanism to ensure that all data from different types of sensors can be uniformly calibrated on the time axis, so as to ensure accurate alignment and correlation analysis of different modal data in subsequent analysis.

[0086] 2. Analysis and Evaluation Module:

[0087] The analysis and evaluation module, as the logical hub of the system, is responsible for tasks such as extracting reference information from characteristic parameters, analyzing correlations, predicting temperature change trends, and calculating temperature prediction deviations.

[0088] During normal system operation, this module periodically selects key moments under stable conditions as reference nodes, extracting temperature and vibration characteristic parameters at those moments as multimodal reference information. This information is used to observe the behavioral relationships between different modes and determine whether the thermal response is affected by mechanical disturbances. The module analyzes the correlation between the characteristic parameters by examining their changing trends and coupling degrees, thus providing a basis for subsequent anomaly detection.

[0089] In terms of temperature prediction, this module reads cable load information and temperature change trends over the current and past periods, and uses predefined calculation rules to estimate the expected temperature value for the next moment. Then, it calculates the difference between the actual temperature and the predicted value at that moment, generating a prediction deviation. The prediction deviation can reflect whether there are any abnormalities in the current system's thermal response, and can also be used to analyze whether sudden changes in heat load pose a potential fault risk.

[0090] This module also cross-compares predicted values, actual values, and correlation parameters across multiple time periods to assess the trend of prediction deviation over time, thereby identifying abnormal development trends before the cable reaches thermal breakdown conditions. All analysis results, including prediction deviation, modal correlation, and trend data, will be transmitted to the status judgment module for further processing.

[0091] 3. Status Judgment Module:

[0092] The status judgment module is used to classify and identify the current cable status based on the difference and threshold rules, and clearly label it as one of the three status types: "normal", "overload" or "fault evolution".

[0093] This module first receives the temperature prediction deviation, correlation indicators, and difference characteristics generated by the analysis and evaluation module, and compares them with preset difference ranges for judgment. If the current difference is below a first set threshold, and both vibration and temperature are within a stable range, it is judged as a normal state. If the difference is between two thresholds, and the temperature change rate increases significantly but no severe vibration occurs, it is judged as an overload state. If the difference exceeds the highest threshold, and is accompanied by temperature jumps or sudden abnormal vibration signals, the cable is considered to have entered a fault evolution state.

[0094] To adapt to the status judgment needs under different load scenarios, this module has the ability to dynamically adjust thresholds. During operation, the system statistically updates the large amount of differential data collected, and calculates indicators such as the mean and standard deviation of the differences in real time to adjust the boundaries of status division, so that the judgment criteria can be automatically adjusted according to the operating conditions, avoiding false alarms or missed alarms.

[0095] While outputting the operating status, the status judgment module can also generate corresponding risk level labels, such as "low risk", "medium risk" or "high risk", and send them to the control center or operation and maintenance system through the communication interface to realize the real-time early warning function.

[0096] 4. Diagnostic localization module:

[0097] The diagnostic location module is used to identify the fault type and further determine the specific location of the cable damage when the cable condition is determined to be "fault evolution".

[0098] Upon receiving the "fault evolution state" determination signal, the system will retrieve complete temperature and vibration characteristic parameters for that time period for comprehensive analysis. First, the system determines the fault category based on the difference between the vibration signal and the reference value. If the difference value is within the overload threshold range and the main abnormality is in temperature characteristics, it is classified as an overload fault; if the main abnormalities are concentrated in vibration amplitude, frequency, energy distribution, and other characteristics, it is classified as a fatigue fault; if both types of characteristics are significantly abnormal, it is identified as a composite fault.

[0099] After diagnosis, the system invokes the corresponding location method based on the fault type. For overload or complex faults, the system uses temperature parameters, such as temperature rise gradient changes and local peak distribution, combined with multi-point temperature acquisition data, to infer the possible damage location. For fatigue faults, the system focuses on analyzing spectral anomalies in the vibration signal, such as sustained enhancement of vibration in a specific frequency band, dominant frequency shift, and enhanced low-frequency impact, to determine the location of potential structural damage.

[0100] Furthermore, with multi-point sensor deployment, the system can also perform spatial back-calculation based on the response time difference of multi-channel signals. Combined with a cable path coding table, it maps specific fault locations to corresponding cable segment numbers, providing visualized location results. The diagnostic location module outputs results including fault type, cable segment location, confidence level, and historical trends, used to generate maintenance instructions or operation and maintenance reports.

[0101] The various modules of this system work collaboratively to achieve closed-loop control throughout the entire process, from acquiring cable operating status data, feature extraction, status judgment to precise fault location. The system boasts advantages such as clear structure, flexible deployment, implementable algorithms, and strong adaptability, making it particularly suitable for long-term intelligent monitoring and maintenance management of medium- and high-voltage cables operating in complex environments with high reliability requirements. The system supports integration with third-party power monitoring platforms and SCADA systems, and can be used as an important functional sub-module of platforms such as smart power distribution and intelligent inspection.

[0102] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for online cable fault diagnosis and location based on multimodal sensor fusion, characterized in that, Includes the following steps: S1. Cable data acquisition and classification processing: Acquire cable operating status data, which includes vibration signals and temperature signals, and classify them according to data type to generate feature parameters; S2. Correlation analysis and prediction deviation calculation before the anomaly: Extract key nodes of characteristic parameters before the cable anomaly occurs as reference information, calculate the correlation between reference information as parameters to be analyzed; obtain cable load data, and use linear regression or time series extrapolation methods to predict cable temperature values; then calculate the temperature prediction deviation based on the difference between the predicted temperature value, the actual temperature value and the parameters to be analyzed. S3. Real-time cable status monitoring and fault evolution identification: Compare the real-time temperature data with the temperature prediction deviation and calculate the difference between the two; based on the difference and combined with the preset comparison rules, determine whether the cable is in a normal state, an overload state, or a fault evolution state. S4. Operation Status Determination and Risk Warning: When the cable is in a normal or overloaded state, the operation status result is output; when it is determined to be in a fault evolution state, a risk warning is triggered and the vibration signal in the fault evolution state is recorded as the parameter to be measured. S5. Fault type identification and damage location: Based on the classification results of the parameters to be measured, the fault type is determined, and then the specific damage location of the cable is determined based on the key node location parameters corresponding to the fault type.

2. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 1, characterized in that: In step S1, the feature parameters include a first feature parameter and a second feature parameter; The first characteristic parameter is the spectral characteristics that characterize the local mechanical vibration changes of the cable, including the amplitude spectrum, envelope energy and peak distribution of a specific frequency band, which are extracted from the vibration signal using fast Fourier transform and wavelet packet decomposition. The second characteristic parameter is a time-series temperature statistical feature characterizing the thermal state of the cable, including average temperature, slope, standard deviation and transient rate of change, which is extracted from the temperature signal using sliding window statistics and multi-scale time sampling methods.

3. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 1, characterized in that: In step S2, calculating the correlation between reference information includes: By setting several sampling points within a time period, the instantaneous values ​​of the first feature parameter and the second feature parameter are extracted as reference information. The correlation between all reference information within this time period is calculated. The correlation is obtained using the Pearson correlation coefficient method and is used as input for subsequent temperature prediction bias.

4. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 3, characterized in that: Step S2, calculating the temperature prediction deviation includes the following steps: Based on cable load data and historical temperature change trends, weighted moving average or regression algorithms are used to predict the cable temperature at a specified time point. Obtain the actual temperature value at that time point; The predicted temperature is compared with the actual temperature, and the difference is taken as the temperature prediction deviation.

5. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 4, characterized in that: In step S2, determining the temperature prediction deviation based on the relationship between the predicted value, the actual temperature value, and the parameter to be analyzed specifically includes: The temperature prediction deviation, predicted temperature value, and reference information of the cable were extracted over multiple time periods to construct three sets of data. By defining the functional relationship between the target variable and the input variable, a calculation expression for evaluating the accuracy of temperature prediction is established using the least squares method, interpolation fitting, or polynomial regression method. The expression is used to quantify the changing trend of the prediction error and is used for subsequent calculation of the difference.

6. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 1, characterized in that: In step S3, the operating status is determined by judging the range of the difference and combining the following judgment rules: Zone 1: Indicates that the cable is in a normal state. The judgment conditions are: the difference is less than the first threshold T1, and the temperature change rate and vibration amplitude are both within their respective set normal operation fluctuation ranges. Region 2: This indicates that the cable is in an overload state. The judgment criteria are: the difference is between the first threshold T1 and the second threshold T2, and the rate of temperature change is higher than the upper limit of normal while the vibration characteristic parameters are still within the normal range, indicating that the cable has a trend of increasing heat load. Region 3: This indicates that the cable has entered a fault evolution state. The judgment condition is: the difference is greater than the second threshold T2, and at the same time, a sharp rise in temperature or abnormal violent vibration is detected, indicating that the cable is undergoing a deterioration process caused by thermal effects or mechanical disturbances.

7. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 6, characterized in that: In step S3, based on this difference and in conjunction with preset comparison rules, the process of determining whether the cable is in a normal state, an overload state, or a fault evolution state is carried out. The upper threshold of Region 1, the lower threshold of Region 3, and their difference are fixed. The range of Region 2 is dynamically adjusted according to the actual operating conditions. The dynamic adjustment includes: As the data collection time increases, the judgment period is extended to reduce the probability of misjudgment. Collect more samples and recalculate the mean and standard deviation of the variance to dynamically update the judgment boundary; The threshold range is weighted and adjusted according to the real-time cable load level to reflect the actual operating pressure.

8. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 1, characterized in that: In step S4, after recording the parameter to be tested, the difference is compared with the reference information to obtain the difference value, and then compared with the comparison threshold to determine its category. Specifically, this includes: If the difference value is within the range of the first category, it corresponds to an overload fault. If it falls within the second category range, it corresponds to fatigue-related faults; If the second category difference range is met again on top of the first category of faults, then the cable can be further confirmed to be in a composite fault state.

9. The method for online cable fault diagnosis and location based on multimodal sensor fusion according to claim 1, characterized in that: In step S5: Both overload faults and complex faults use temperature gradient and peak distribution from temperature parameters as the basis for fault location. For fatigue-related faults, the peak frequency and vibration duration from vibration parameters are used as the basis for fault location.

10. A cable online fault diagnosis and location system based on multimodal sensor fusion, used to implement the method according to any one of claims 1 to 9, characterized in that, include: The data processing module is used to classify cable operating status data and extract temperature and vibration characteristic parameters; The analysis and evaluation module is used to extract reference information and calculate its correlation, obtain the relationship between cable load and temperature, perform temperature prediction and prediction deviation calculation; The status determination module is used to determine the cable operating status type based on the difference and a preset threshold. The diagnostic location module is used to determine the fault type based on the classification of the parameter to be measured and select the corresponding vibration or temperature location parameters to locate the cable fault location.

Images