Power terminal fault diagnosis system and method based on multi-source data analysis

Abstract

The application belongs to the technical field of power distribution operation and maintenance, and particularly relates to a power terminal fault diagnosis system and method based on multi-source data analysis. The application constructs a standardized fault feature set by integrating multi-source operation data, obtains a weighted comprehensive distortion index through time series data operation, determines the deterioration state of equipment in stages, solves the coupling coefficient in combination with the load, environmental temperature and humidity, and heat dissipation working conditions, accurately locates the fault point and the fault cause relying on the power distribution topology, collects continuous multi-day data to fit the fault deterioration rate, divides the operation and maintenance level according to the rate threshold, and generates maintenance suggestions. The application can identify progressive and implicit faults such as insulation aging, shorten the time consumption of fault troubleshooting, promote the operation and maintenance from after-event repair to preventive maintenance, and effectively reduce the probability of terminal sudden failure shutdown.

Description

Technical Field

[0001] This invention relates to the field of power distribution operation and maintenance technology, specifically to a power terminal fault diagnosis system and method based on multi-source data analysis. Background Technology

[0002] As a key component of the distribution network, the stability of the power terminal directly affects the reliability of regional power supply. With the intelligent upgrading of the distribution network, multiple independent monitoring systems such as partial discharge, fiber optic temperature measurement, intelligent inspection, and video monitoring are gradually deployed on site, and various types of monitoring data are stored in heterogeneous databases with incompatible architectures.

[0003] Current power terminal fault diagnosis generally adopts a technical approach of extracting all raw data, unifying the format, and relying on fixed thresholds to achieve fault alarms. This approach makes it difficult to integrate multi-dimensional operating parameters to quantify the deterioration and development patterns of equipment, and also fails to accurately identify progressive hidden faults and quantify the maintenance levels, in order to meet the development needs of distribution networks to transform from fault repair to preventive maintenance.

[0004] Therefore, developing a power terminal fault diagnosis system based on multi-source data analysis that can quantitatively assess equipment deterioration and accurately locate faults has become an urgent technical problem to be solved in the field of power distribution operation and maintenance. Summary of the Invention

[0005] The purpose of this invention is to provide a power terminal fault diagnosis system and method based on multi-source data analysis to address the technical deficiencies mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a power terminal fault diagnosis system based on multi-source data analysis, including a heterogeneous source fault feature anchoring module, a time-series distortion degree hierarchical deduction module, a working condition coupled fault source tracing mapping module, and a fault deterioration rate quantitative judgment module.

[0007] The heterogeneous source fault feature anchoring module uses equipment asset codes as indexes to complete cross-database data collection and generate a standardized anchoring feature dataset; the time series distortion degree hierarchical inference module relies on benchmark time series data and calibration coefficients to complete multi-level distortion calculations and realize automatic classification of equipment deterioration states.

[0008] The operating condition coupling fault tracing and mapping module integrates multiple types of field operating condition parameters to solve coupling indices, and combines the power distribution topology to complete the location of fault points and fault causes; the fault deterioration rate quantitative analysis module fits the deterioration development rate based on long-cycle coupling index data, generates standardized operation and maintenance diagnosis schemes in a hierarchical manner, and realizes the intelligent transformation of power terminal faults from post-event emergency repair to pre-event prediction based on full-link quantitative analysis.

[0009] Furthermore, the heterogeneous source fault feature anchoring module searches existing databases such as power distribution backend, measurement and control backend, partial discharge, fiber optic temperature measurement, intelligent inspection ledger, video features, and plant benchmark ledger according to equipment asset codes, and extracts the corresponding equipment operation, defects and benchmark parameters in a differentiated manner, standardizing the retrieval path and data source of multi-source heterogeneous data, and providing a regular data source for subsequent calculations.

[0010] Furthermore, the time-series distortion degree hierarchical inference module retrieves the equipment fault-free baseline time-series data, sampling point disturbance correction coefficients, and fault feature weight parameters under the same season and load conditions as the basis for calculation, clarifies the source of distortion-related calculation parameters, and ensures the accuracy and reliability of distortion calculation results by relying on standardized baseline data.

[0011] Furthermore, the parameters required for the operation condition coupling fault tracing mapping module are divided into three categories, which are respectively taken from the real-time storage field of the loop load, the switch cabinet environmental temperature and humidity monitoring database, and the start-stop operation log of the heat dissipation equipment in the cabinet, to ensure that the source of the operation condition data is real and traceable and to improve the reliability of the coupling calculation.

[0012] Furthermore, the operating condition coupling fault tracing and mapping module relies on the physical topology ledger of the plant's power distribution to lock the physical location of the equipment. It uses a three-element matching rule composed of comprehensive distortion values, coupling correlation coefficients, and operating condition over-limit items to determine the source of the fault. This limits the logical components of fault location and cause identification, eliminating the shortcomings of manual investigation and quickly and accurately locating the root cause of the fault.

[0013] Furthermore, the fault degradation rate quantification and judgment module sets 24 hours as a single standard diagnostic cycle. After continuously collecting the operating condition coupling correlation coefficients of five complete cycles, it uses the least squares method to fit the daily fault degradation rate, limits the data acquisition cycle and fitting calculation method of the degradation rate, and uses multi-cycle data to reduce the rate fitting error and objectively reflect the fault evolution trend.

[0014] Furthermore, the fault degradation rate quantification and judgment module classifies the degradation rate into three levels of operation and maintenance handling: emergency shutdown, planned maintenance, and routine tracking. It generates a diagnostic report based on the calculated critical failure time of the equipment and sends it to the operation and maintenance management terminal. It limits the operation and maintenance classification rules corresponding to the degradation rate and the report push targets, providing a quantitative basis for operation and maintenance work and optimizing maintenance arrangements.

[0015] This invention also proposes a power terminal fault diagnosis method based on multi-source data analysis, which sequentially performs four processes: feature collection, distortion classification, coupling source tracing, and degradation rate assessment. It relies on modular step-by-step computation to achieve fully automated diagnosis of terminal faults, simplifying the fault diagnosis process and improving the intelligent level of power terminal operation and maintenance.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] 1. In this invention, by abandoning the traditional fixed threshold alarm mode, and relying on the weighted calculation of comprehensive distortion index based on multi-source operating data, it can accurately identify slow-changing latent faults that are difficult to detect by conventional thresholds, such as insulation aging and progressive heating of joints, and effectively improve the accuracy of fault identification.

[0018] 2. In this invention, the fault is automatically traced and located by combining load, ambient temperature and humidity and heat dissipation conditions, which reduces the time spent on manual troubleshooting. Furthermore, by fitting the deterioration rate with multi-cycle data and managing it in a hierarchical manner, the invention achieves a shift from emergency repair to preventive maintenance, thereby reducing the risk of sudden power outages. Attached Figure Description

[0019] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0020] Figure 1 This is a schematic diagram of the overall system structure of the present invention;

[0021] Figure 2 This is a schematic diagram of the operation method of the present invention. Detailed Implementation

[0022] 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.

[0023] Example 1: Refer to Figure 1 As shown, the power terminal fault diagnosis system based on multi-source data analysis proposed in this embodiment breaks through the limitations of traditional crude diagnosis methods such as fixed threshold alarms, manual item-by-item investigation, and post-fault emergency repair. It establishes a full-link intelligent diagnosis architecture based on multi-source data hierarchical analysis, which consists of a heterogeneous source fault feature anchoring module, a time-series distortion degree hierarchical deduction module, an operating condition coupled fault tracing mapping module, and a fault deterioration rate quantitative judgment module.

[0024] After receiving the terminal fault diagnosis trigger command, the heterogeneous source fault feature anchoring module searches multiple types of existing databases according to the asset code of the equipment to be diagnosed. For example, it extracts the original stored fields of circuit voltage and three-phase current from the power distribution background database, retrieves the original fields of tripping signal and zero-sequence voltage record of the same coded equipment from the measurement and control background database, and simultaneously retrieves the partial discharge peak value from the partial discharge database, extracts the cable body temperature from the fiber optic temperature measurement library, extracts the mechanical defect records entered in the intelligent inspection ledger, and extracts the quantitative features of abnormal noise / shell deformation images from the video feature library. The equipment's factory rated parameters, the weight values ​​of each feature item, and the single sampling point fluctuation correction coefficient are retrieved from the plant's benchmark ledger server form, etc.

[0025] Finally, the aggregated data forms an anchored feature dataset, which is then pushed to the time-series distortion degree hierarchical inference module in real time. This generates an anchored feature dataset with unified feature coding, providing a standardized feature index benchmark for downstream analysis and computation.

[0026] The time-series distortion degree hierarchical inference module receives the anchored feature dataset, first calculates the time-series distortion coefficient of each single feature, and then weights and synthesizes the comprehensive distortion index of the equipment. Based on the fixed numerical threshold calibrated by the historical fault data of the plant, it automatically classifies the equipment into three levels of deterioration. It replaces the industry-standard fixed alarm threshold with a dynamic quantitative distortion index, realizing the hierarchical identification of slowly changing latent faults. This is conducive to accurately capturing latent faults that are difficult to detect with fixed thresholds, such as slow insulation aging and gradual heating of joints.

[0027] Specifically, after receiving the anchored feature dataset, the time-series distortion degree hierarchical inference module retrieves the full-time-series measured data of the target equipment based on feature encoding. Simultaneously, it retrieves the fault-free operating benchmark time-series data stored on the benchmark ledger server under the same season and load conditions. Then, it substitutes the instantaneous measured value of each sampling point for each fault feature and the corresponding benchmark single-point value into the single-feature distortion degree calculation formula to complete the calculation. The single-feature distortion degree calculation formula is as follows: ;

[0028] Wherein, Dti: the single-feature time-series distortion coefficient of the i-th fault feature, dimensionless; T: the total number of time-series sampling points; Xtik: the instantaneous original measured value of the k-th sampling point of the i-th fault feature, taken from the real-time time-series single-point storage data of each operation and maintenance database; Xoik: the instantaneous benchmark value of the k-th sampling point of the i-th fault feature under the same working condition health benchmark, taken from the factory equipment factory benchmark ledger server; αik: the instantaneous disturbance correction coefficient of the k-th sampling point of the i-th fault feature, preferably ranging from 0.7 to 1.3, stored in the ledger sampling fluctuation correction form, calibrated one by one according to the equipment factory impact test data, used to eliminate single-point data anomalies caused by lightning strikes and instantaneous load impacts;

[0029] After all sampling points for a single feature are calculated, the feature item weight wi parameter stored independently in the ledger is retrieved and substituted into the comprehensive distortion formula to solve for the overall comprehensive distortion coefficient Davg of the equipment. The comprehensive distortion formula is as follows: ;

[0030] Wherein, Davg: full feature weighted comprehensive distortion coefficient; n: total number of fault features involved in the calculation, retrieved from the factory equipment outgoing baseline ledger server according to equipment type; wi: global fault contribution weight of the i-th fault feature, preferably, the value range is 0 to 1, stored in the ledger feature importance form, assigned by the fault big data statistics of the past three years, characterizing the proportion of the impact of different parameters such as partial discharge, temperature, and current on equipment failure;

[0031] Subsequently, based on thousands of fault samples from the factory's fault handling archive over the past three years, the threshold for classifying the three-level state was statistically determined. The degradation level was automatically classified according to the calculated Davg values. Corresponding degradation level labels and Davg value fields were added to the original feature dataset, generating a distorted dataset with hierarchical annotations, which was then transmitted in real-time to the downstream condition-coupled fault tracing mapping module. An example of the three-level temporal distortion state quantification classification rules is as follows:

[0032] If Davg < D1, it is judged as a slight deterioration; if D1 ≤ Davg ≤ D2, it is judged as a moderate anomaly; if Davg > D2, it is judged as an emergency failure. The classification thresholds D1 and D2 are all obtained by statistical fitting of the failure archives over the years and are fixed in the system parameters, and D2 > D1 > 0.

[0033] The operating condition coupled fault tracing mapping module immediately starts calculations after receiving each distortion dataset with Davg and degradation level sent from the upstream. It constructs a coupled calculation model based on three types of predetermined operating condition parameters: circuit load conditions, switch cabinet environmental temperature and humidity conditions, and internal heat dissipation equipment operating conditions. Relying on the three-element matching rule of "comprehensive distortion value + coupling coefficient + operating condition over-limit item", it locks the physical location of the fault and the root cause of the fault by combining the physical topology ledger of the power distribution in the plant area. It realizes accurate fault tracing across systems and equipment, replacing the traditional crude tracing method of manually reviewing videos and inspection records. It is conducive to accurately locking the physical location and root cause of the fault, and significantly reducing the on-site fault investigation time.

[0034] Specifically, the operating condition coupling fault tracing and mapping module triggers a full-process real-time calculation every time it receives a set of distorted datasets. Based on the equipment asset code, it retrieves three types of real-time operating condition parameters: the first type is the circuit load operating condition parameters, which are taken from the real-time load storage field in the background; the second type is the switch cabinet environmental temperature and humidity operating condition parameters, which are taken from the plant environmental monitoring database and fiber optic temperature measurement archive; and the third type is the operating condition parameters of the heat dissipation equipment in the cabinet, which are taken from the fan start-stop operation log in the power distribution room.

[0035] Substitute the real-time data of the above three types of operating conditions and the issued Davg into the operating condition coupling calculation formula to solve for the operating condition coupling correlation coefficient R: R=β×Davg+γ×L+δ×S; where R: operating condition coupling correlation coefficient, dimensionless, the higher the value, the stronger the correlation between the current operating condition and equipment fault distortion; β: comprehensive distortion matching coefficient, which is fixed in the fault handling archive of previous years according to cabinet type and can be directly retrieved by the system;

[0036] L: Loop load deviation coefficient, L=|Lnow-Lstd| / Lstd, where Linow is the real-time load and Lstd is the rated load, taken from the factory equipment factory benchmark ledger server; γ: Load coupling correction coefficient, generated from historical overload fault statistics and stored in the system parameter library; S: Environmental and heat dissipation combined disturbance, S=ΔT×H+F, where ΔT is the difference between the measured temperature and the benchmark operating temperature, H is the environmental humidity (taken from the environmental monitoring database), and F is the quantitative value of the heat dissipation equipment shutdown fault (preferred, 1 for shutdown and 0 for normal); δ: Environmental disturbance correction coefficient, fixed in the system's built-in parameter file;

[0037] It also has a built-in fault cause mapping map generated by the topology ledger of the plant’s high-voltage main substation and multiple 10kV distribution rooms. The map pre-stores the binding relationship between the topology point code, fault cause, and parameter over-limit item. When locking the physical location of the fault, it prioritizes matching the asset code with the topology ledger to obtain accurate physical location information such as the distribution room number, switch cabinet bay, and outgoing circuit number of the equipment.

[0038] To pinpoint the root cause of the fault, a ternary matching logic is employed: the degradation level of Davg, the R value range, and the specific out-of-limit items in the three operating conditions are simultaneously retrieved, and the fault cause is matched against preset rules in the graph. An example of the matching strategy is as follows:

[0039] For example, if Davg is in a moderate abnormality, R>R1 and the circuit load parameters exceed the limit, the matching cause is long-term overload heating; if Davg is in an emergency fault, R>R2 and the temperature and humidity conditions exceed the standard, the matching cause is insulation damage caused by a humid environment; where R1 and R2 are preset thresholds, and R1<R2.

[0040] Finally, after the source tracing is completed, a source tracing data packet is generated, which carries the precise location number, the source of the fault, and the operating condition coupling correlation coefficient R, and is sent to the fault deterioration rate quantification and judgment module in real time.

[0041] The fault degradation rate quantification and judgment module uses 24 hours as a standard diagnostic cycle and continuously collects the upstream operating condition coupling correlation coefficients for 5 complete standard diagnostic cycles (a total of 5 days). At the end of each diagnostic cycle, the least squares method is used to uniformly fit the fault degradation rate. Based on historical failure records, the rate threshold is calibrated to classify three levels of operation and maintenance. The module quantifies and predicts the critical failure time of equipment and outputs standardized operation and maintenance suggestions. It upgrades from passive emergency repair after a fault occurs to quantitative predictive preventive maintenance. The classification standards are all based on the quantitative calibration of historical fault data on site, avoiding the reliance on experience in operation and maintenance decisions.

[0042] Specifically, the fault deterioration rate quantification and judgment module continuously receives the source tracing data packets sent one after another in real time and classifies them into the corresponding diagnostic cycle data pool according to date. The data pool continuously caches all R values ​​within 5 complete 24-hour diagnostic cycles. Only at the end of a single 24-hour standard diagnostic cycle is a batch operation triggered to extract the equipment coupling coefficient R data of the 5 consecutive diagnostic cycles that have been collected, and the deterioration rate V (unit: daily average R change amplitude) is obtained by fitting using the least squares method.

[0043] Then, based on the handling grading threshold and the degradation rate V, the handling level is determined. After determining the handling level, the remaining safe operating time of the equipment is predicted by combining the fault location, the root cause, and the degradation rate. Finally, all information is integrated to generate a standardized fault diagnosis report and push it to the plant operation and maintenance management terminal. Further, the specific grading rules can be referred to as follows:

[0044] If the rate of deterioration V > 0.08 / day, the emergency shutdown level is determined; if 0.03 ≤ rate of deterioration V ≤ 0.08 / day, the planned maintenance level is determined; if the rate of deterioration V < 0.03 / day, the routine monitoring level is determined.

[0045] Example 2: Refer to Figure 2 As shown, the difference between this embodiment and Embodiment 1 lies in the fact that the power terminal fault diagnosis method based on multi-source data analysis proposed in this embodiment relies on the serial linkage of four-level modules and the quantitative calculation based on multi-source existing data. This method overcomes the shortcomings of fixed threshold alarms and manual on-site inspections, and can identify progressive hidden faults in equipment, accurately locate the fault location and causes, quantify the rate of deterioration, and automatically push maintenance plans in a graded manner, thereby reducing maintenance errors and inspection costs. An example of the operation steps is as follows:

[0046] Step 1, Feature Collection: Upon receiving the fault diagnosis instruction, the heterogeneous source fault feature anchoring module retrieves operational, temperature measurement, inspection, and video data from various existing databases based on the equipment asset code, unifies the feature codes, removes invalid alarm data, and generates an anchoring feature dataset which is then distributed in real time.

[0047] Step 2, Distortion Classification: Retrieve instantaneous measured data and baseline data from the equipment, calculate the single-feature temporal distortion coefficient Dti and the comprehensive distortion coefficient Davg, classify them into three levels of deterioration according to the corresponding preset thresholds, and pass the distortion dataset down.

[0048] Step 3, Coupling Source Tracing: Retrieve load, temperature and humidity, and heat dissipation data for three types of operating conditions, calculate the operating condition coupling correlation coefficient R, and combine the comprehensive distortion coefficient Davg, the operating condition coupling correlation coefficient R, and the operating condition over-limit matching map to pinpoint the fault location and cause, and output the source tracing data package.

[0049] Step 4: Degradation Rate Assessment: Accumulate the R value for 5 consecutive days using 24-hour standard diagnostic cycle, calculate the degradation rate V using the least squares method, classify the operation and maintenance plan based on the threshold, and generate a fault diagnosis report.

[0050] The working principle of this invention is as follows: When in use, a feature set is constructed by collecting multi-source power operation data. Based on time-series distortion calculation, a comprehensive distortion index is obtained and the equipment degradation level is classified. In addition, the coupling coefficient is calculated by combining load, ambient temperature and humidity, and heat dissipation conditions to accurately locate the fault location and cause. Furthermore, based on the degradation rate fitted by multi-cycle data, a graded operation and maintenance strategy is formulated. This overcomes the shortcomings of traditional fixed thresholds, can identify progressive hidden faults, shorten the fault investigation time, and realize preventive maintenance based on degradation prediction. This promotes the transformation of operation and maintenance from fault repair to preventive maintenance and reduces the risk of sudden power outages.

[0051] 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 any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, enabling those skilled in the art to better understand and utilize it. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A power terminal fault diagnosis system based on multi-source data analysis, characterized in that, It includes a heterogeneous source fault feature anchoring module, a time-series distortion degree hierarchical deduction module, a working condition coupled fault source tracing mapping module, and a fault deterioration rate quantitative analysis module; The heterogeneous source fault feature anchoring module uses equipment asset codes as indexes to complete cross-database data collection and generate a standardized anchoring feature dataset; The time-series distortion degree hierarchical inference module relies on the benchmark time-series data and calibration coefficients to complete multi-level distortion calculations and realize automatic classification of equipment degradation status; The working condition coupling fault tracing and mapping module integrates multiple types of field working condition parameters to solve coupling indices, and combines power distribution topology to complete the location of fault points and fault causes. The fault degradation rate quantitative analysis module fits the degradation development rate based on long-term coupled index data and generates standardized operation and maintenance diagnosis solutions in a hierarchical manner.

2. The power terminal fault diagnosis system based on multi-source data analysis according to claim 1, characterized in that, The time-series distortion degree hierarchical inference module retrieves the equipment fault-free baseline time-series data, sampling point disturbance correction coefficients, and fault feature weight parameters under the same season and load conditions as the basis for calculation.

3. The power terminal fault diagnosis system based on multi-source data analysis according to claim 1, characterized in that, The parameters required for the operation condition coupling fault tracing mapping module are taken from the real-time storage field of the loop load, the switch cabinet environmental temperature and humidity monitoring database, and the start-up and shutdown log of the heat dissipation equipment in the cabinet.

4. The power terminal fault diagnosis system based on multi-source data analysis according to claim 3, characterized in that, The operating condition coupling fault tracing and mapping module relies on the physical topology ledger of the plant's power distribution to lock the physical location of the equipment, and uses a three-element matching rule composed of comprehensive distortion value, coupling correlation coefficient and operating condition over-limit item to determine the source of the fault.

5. The power terminal fault diagnosis system based on multi-source data analysis according to claim 1, characterized in that, The fault degradation rate quantification and judgment module sets a standard diagnostic cycle, and after continuously collecting the operating condition coupling correlation coefficients of five complete standard diagnostic cycles, it uses the least squares method to fit the daily fault degradation rate.

6. The power terminal fault diagnosis system based on multi-source data analysis according to claim 5, characterized in that, The fault degradation rate quantification and judgment module classifies the degradation rate into three levels of operation and maintenance handling: emergency shutdown, planned maintenance, and routine tracking. It generates a diagnostic report based on the calculated critical failure time of the equipment and sends it to the operation and maintenance management terminal.

7. A power terminal fault diagnosis method based on multi-source data analysis, applied to the power terminal fault diagnosis system based on multi-source data analysis as described in any one of claims 1-6, characterized in that, Includes the following steps: Step 1, Feature Collection: After receiving the fault diagnosis instruction, the system retrieves the operation, temperature measurement, inspection, and video data from various existing databases based on the equipment asset code, unifies the feature codes, removes invalid alarm data, and generates an anchored feature dataset which is then distributed in real time. Step 2, Distortion Classification: Retrieve instantaneous measured data and baseline data from the equipment, calculate the single-feature temporal distortion coefficient Dti and the comprehensive distortion coefficient Davg, classify them into three levels of deterioration according to the corresponding preset thresholds, and form a distortion dataset to be passed down. Step 3, Coupling Source Tracing: Retrieve load, temperature and humidity, and heat dissipation data for three types of operating conditions, calculate the operating condition coupling correlation coefficient R, and combine the comprehensive distortion coefficient Davg, the operating condition coupling correlation coefficient R, and the operating condition over-limit item matching map to locate the fault location and cause, and output the source tracing data package. Step 4: Degradation Rate Assessment: Accumulate the R value for 5 consecutive days using 24 hours as the standard diagnostic cycle, calculate the degradation rate V using the least squares method, classify the operation and maintenance plan based on the threshold, and generate a fault diagnosis report.

Images