Wind power plant current collection line drop fuse fault diagnosis method and system

Abstract

The invention discloses a wind power plant current collection line drop fuse fault diagnosis method and system, and the method comprises the steps: deploying a multi-source sensor network on target drop fuse equipment, the method comprises the following steps: acquiring operation and environment data including fusion tube surface temperature, conductive loop contact resistance, fuse action times, environment temperature, humidity, wind speed and pollution degree in real time, performing filtering, denoising and normalization preprocessing on the acquired original data, and constructing a multi-dimensional characteristic parameter set X = {x1, x2,..., xn}; through continuous monitoring and trend analysis of the dynamic health degree score H (t), the system can give out early warning at the early stage of measurable degradation of the equipment performance. Therefore, operation and maintenance personnel can plan a maintenance window in advance, the non-planned downtime is remarkably shortened, the availability rate of the power generation equipment is remarkably improved, and compared with a traditional mode depending on manual inspection or single parameter threshold value alarm, the method can find potential problems of the equipment earlier.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, and in particular to a method and system for diagnosing drop-out fuse faults in wind farm collector lines. Background Technology

[0002] Drop-out fuses (DFTs) are widely used overload and short-circuit protection devices in wind farm power collection lines. Their reliable operation is crucial for ensuring the stability of power transmission within the wind farm. Since wind farms are mostly located in harsh outdoor environments, DFTs are constantly exposed to complex environmental factors such as strong winds, high and low temperatures, humidity, salt spray, dust storms, and lightning. They also frequently withstand load changes and even short-circuit current surges caused by wind fluctuations, resulting in a relatively high failure rate.

[0003] Currently, the operation and maintenance management of drop protection mainly relies on regular manual inspections and post-failure repairs. This model has obvious limitations: the inspection cycle is long, making it difficult to detect potential defects and performance degradation trends in equipment in a timely manner; it depends on the experience and sense of responsibility of maintenance personnel, lacking objective and quantitative assessment standards for equipment status; and it often only takes action after a failure occurs, leading to unplanned downtime, affecting power generation efficiency, and potentially causing greater equipment damage or even safety accidents due to the escalation of the failure.

[0004] Existing equipment condition monitoring technologies typically monitor and trigger threshold alarms for only a single parameter (such as temperature). While this approach can detect some apparent faults, it cannot comprehensively reflect the overall health status of the equipment under the combined effects of multiple factors. It is difficult to provide early warning and accurate location of complex faults, and it cannot quantify and assess the remaining service life of the equipment, thus failing to support the shift from periodic maintenance to predictive maintenance. Summary of the Invention

[0005] The purpose of this invention is to overcome the above-mentioned technical deficiencies and propose a fault diagnosis method and system for drop-out protection of wind farm collector lines, which solves the technical problems of single monitoring dimension, lack of quantitative assessment, difficulty in fault diagnosis, and inability to predict maintenance in the existing technology.

[0006] To achieve the above-mentioned technical objectives, in a first aspect, the present invention provides a method for diagnosing faults in wind farm collector lines due to drop-out fuses, comprising: A multi-source sensor network is deployed on the target drop protection device to collect real-time operational and environmental data, including fuse tube surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. The collected raw data is preprocessed by filtering, denoising, and normalization to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}. Based on historical failure data, Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) are used to identify the set of key failure modes F={F1,F2,...,Fm} that lead to drop fuse failures. The Analytic Hierarchy Process (AHP) is then used to calculate the weight vector W={w1,w2,...,wm} for each failure mode, where ∑wi=1. Based on the multidimensional feature parameter set X and the weight vector W of failure modes, a dynamic health assessment model for the equipment is established, and the comprehensive health score H(t) of the equipment is calculated. Set the health score threshold H_th and the abnormal threshold of each feature parameter. When the real-time calculated H(t) is lower than H_th or any key feature parameter exceeds its abnormal threshold, an early warning is triggered, and the most likely cause of failure is located by matching the deviation mode of the feature parameter with the failure mode library. Generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance recommendations.

[0007] Optionally, the surface temperature of the molten tube is acquired using an infrared temperature sensor or a wireless temperature tag, the contact resistance of the conductive circuit is indirectly calculated by measuring the voltage drop of the circuit, and the environmental pollution level is assessed by leakage current monitoring or an image recognition device.

[0008] Optionally, the set of critical failure modes F includes at least: poor contact, melt aging, flashover of insulating components, jamming of mechanical mechanisms, abnormal installation stress, and mechanical damage caused by external short-circuit current impact.

[0009] Optionally, the comprehensive health score H(t) is defined by the following formula: H(t)=100*exp(-λ*t)*∏_{i=1}^{k}[1-α_i*(S_i(t) / S_{i,max})^β_i] In the formula: H(t) represents the overall health score of the device at time t, ranging from 0 to 100. The higher the score, the healthier the device. exp(-λ*t) is a baseline aging factor based on equipment operating time, and λ is an aging coefficient related to the inherent lifespan of the equipment; k represents the number of key characteristic parameters involved in the evaluation; S_i(t) is the severity value of the i-th key feature parameter deviating from the normal baseline after normalization at time t; S_{i,max} is the maximum allowable deviation severity threshold for the i-th key feature parameter; α_i is the sensitivity coefficient of the i-th feature parameter, 0<α_i≤1, and its value is related to the weight of the failure mode associated with this parameter in the failure mode analysis step. β_i is the morphological coefficient of the i-th feature parameter, β_i≥1, used to adjust the nonlinearity of the parameter's influence on health.

[0010] Optionally, the key characteristic parameters include at least: the rate of temperature rise of the molten tube, the rate of increase of contact resistance, the ratio of the cumulative number of operations to the number of designed life operations, the three-phase current imbalance, and the effective value of leakage current on the insulator surface.

[0011] Optionally, the initial value of the aging coefficient λ is set according to the mean time between failures (MTBF) provided by the equipment manufacturer, and is dynamically corrected during equipment operation using a Bayesian update method based on historical data of similar equipment.

[0012] Optionally, the severity value S_i(t) is calculated as follows: S_i(t)=max(0,(P_i(t)-P_{i,base}) / (P_{i,limit}-P_{i,base})), where P_i(t) is the current value of the feature parameter, P_{i,base} is the normal baseline value, and P_{i,limit} is the safe operation limit value.

[0013] Optionally, the fault diagnosis adopts a hybrid diagnosis mechanism based on case-based reasoning (CBR) and rule-based reasoning (RBR): First, the real-time feature pattern is matched with the historical fault case library for similarity; if the match is successful, the corresponding fault cause is output; if the match fails, predefined expert rules are triggered to make a reasoning judgment.

[0014] Optionally, it also includes: predicting the time point T_m when the health score H(t) drops to the preset maintenance threshold H_maint and the time point T_f when it drops to the failure threshold H_fail based on the historical change curve of the health score H(t) using trend extrapolation, thereby generating a predictive maintenance plan.

[0015] To achieve the above-mentioned technical objectives, in a second aspect, the present invention provides a fault diagnosis system for drop-out fuses in wind farm collector lines, comprising: The data acquisition and processing module is used to deploy a multi-source sensor network on the target drop protection device to collect real-time operational and environmental data, including fuse tube surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. The module also performs filtering, noise reduction, and normalization preprocessing on the collected raw data to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}. The Failure Mode Analysis module is used to identify the set of key failure modes F={F1,F2,...,Fm} that cause drop fuse failures based on historical failure data, using Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) methods, and to calculate the weight vector W={w1,w2,...,wm} for each failure mode using the Analytic Hierarchy Process (AHP), where ∑wi=1; The health assessment module is used to establish a dynamic health assessment model for equipment based on a multidimensional feature parameter set X and a failure mode weight vector W, and to calculate the comprehensive health score H(t) of the equipment. The fault diagnosis module is used to set the health score threshold H_th and the abnormal threshold of each characteristic parameter. When the real-time calculated H(t) is lower than H_th or any key characteristic parameter exceeds its abnormal threshold, an early warning is triggered, and the most likely cause of the fault is located by matching the deviation mode of the characteristic parameter with the failure mode library. The results output module is used to generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance suggestions.

[0016] The beneficial effects of this invention include: 1. Through continuous monitoring and trend analysis of the dynamic health score H(t), the system can issue early warnings at the early stages of measurable degradation in equipment performance. This allows maintenance personnel to plan maintenance windows in advance, significantly reducing unplanned downtime and significantly improving the availability of power generation equipment. Compared with traditional methods that rely on manual inspections or single-parameter threshold alarms, this method can detect potential equipment problems earlier. 2. The system integrates multi-dimensional parameter data such as electrical, thermal, mechanical and environmental parameters, and uses a weighted fusion algorithm to calculate a comprehensive health score. Compared with a single temperature monitoring method, this method can more comprehensively reflect the true status of the equipment, reduce false alarms and missed alarms, and effectively identify complex faults such as poor contact and insulation deterioration. 3. By combining case-based reasoning and rule-based reasoning in a hybrid diagnostic mechanism, the system can automatically analyze fault characteristic patterns and accurately identify common fault types and complex faults. Compared with traditional experience-based diagnostic methods, the diagnostic accuracy is significantly improved and the fault location time is significantly shortened. 4. Based on the differentiated inspection strategy of equipment health score, the inspection frequency distribution can be optimized, the operation and maintenance resources can be rationally allocated, and the predictive maintenance plan can make spare parts inventory management and maintenance manpower arrangement more scientific and reasonable. Overall, the operation and maintenance cost of wind farm collection lines is effectively controlled, the service life of equipment is extended, and the power generation benefits are increased due to reduced downtime. Attached Figure Description

[0017] Figure 1This is a flowchart of a method for diagnosing drop-out fuse faults in wind farm collector lines according to an embodiment of the present invention; Figure 2 This is a structural diagram of a wind farm collector line drop-out fuse fault diagnosis system according to an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Drop-out fuses (DFTs) are widely used overload and short-circuit protection devices in wind farm power collection lines. Their reliable operation is crucial for ensuring the stability of power transmission within the wind farm. Since wind farms are mostly located in harsh outdoor environments, DFTs are constantly exposed to complex environmental factors such as strong winds, high and low temperatures, humidity, salt spray, dust storms, and lightning. They also frequently withstand load changes and even short-circuit current surges caused by wind fluctuations, resulting in a relatively high failure rate.

[0020] Currently, the operation and maintenance management of drop protection mainly relies on regular manual inspections and post-failure repairs. This model has obvious limitations: the inspection cycle is long, making it difficult to detect potential defects and performance degradation trends in equipment in a timely manner; it depends on the experience and sense of responsibility of maintenance personnel, lacking objective and quantitative assessment standards for equipment status; and it often only takes action after a failure occurs, leading to unplanned downtime, affecting power generation efficiency, and potentially causing greater equipment damage or even safety accidents due to the escalation of the failure.

[0021] Existing equipment condition monitoring technologies typically monitor and trigger threshold alarms for only a single parameter (such as temperature). While this approach can detect some apparent faults, it cannot comprehensively reflect the overall health status of the equipment under the combined effects of multiple factors. It is difficult to provide early warning and accurate location of complex faults, and it cannot quantify and assess the remaining service life of the equipment, thus failing to support the shift from periodic maintenance to predictive maintenance.

[0022] To address the aforementioned technical problems, this invention proposes a method for diagnosing drop-out fuse faults in wind farm collector lines.

[0023] Reference Figure 1 , Figure 1 This is a flowchart of a method for diagnosing drop-out fuse faults in wind farm collector lines according to an embodiment of the present invention. The method includes: S101. Deploy a multi-source sensor network on the target drop protection device to collect real-time operational and environmental data, including fuse tube surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. Perform filtering, noise reduction, and normalization preprocessing on the collected raw data to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}.

[0024] By comprehensively deploying multiple types of sensors, including electrical, thermal, mechanical, and environmental sensors, a data acquisition system covering the main operational dimensions of the equipment has been constructed. Data acquisition has transitioned from single-point to multi-point, from static to dynamic, and from offline to online, providing a high spatiotemporal resolution data foundation for subsequent condition assessment and fault diagnosis. Through real-time acquisition, the system can capture transient changes in equipment status, such as the temperature rise process after a short-circuit impact and the gradual trend of contact resistance. Filtering primarily targets high-frequency electromagnetic interference and random noise; denoising algorithms can employ wavelet transform or Kalman filtering. Normalization ensures that parameters of different dimensions and magnitudes can be compared and fused on a unified scale, laying the data foundation for subsequent comprehensive evaluation.

[0025] The surface temperature of the molten tube is collected by an infrared temperature sensor or a wireless temperature tag. The contact resistance of the conductive circuit is indirectly calculated by measuring the voltage drop of the circuit. The environmental pollution level is assessed by leakage current monitoring or an image recognition device.

[0026] S102. Based on historical failure data, Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) are used to identify the set of key failure modes F={F1,F2,...,Fm} that cause drop fuse failures, and the weight vector W={w1,w2,...,wm} of each failure mode is calculated using the Analytic Hierarchy Process (AHP), where ∑wi=1.

[0027] The FMEA method systematically analyzes the consequences and severity of each potential failure mode, helping to comprehensively identify risk points. The FTA method traces the root cause of failures from top to bottom, clarifying the logical relationships of failures. The combination of these two methods ensures the systematic and in-depth nature of failure mode identification. The Analytic Hierarchy Process (AHP) quantifies expert experience, scientifically determining the relative weights of each failure mode through pairwise comparison judgment matrices. This reflects the differences in the impact of different failure modes on the overall reliability of the equipment. This weight vector W provides a direct basis for setting the sensitivity coefficients α_i of each characteristic parameter in the subsequent health assessment model.

[0028] The set of critical failure modes F includes at least: poor contact, melt aging, flashover of insulating components, jamming of mechanical mechanisms, abnormal installation stress, and mechanical damage caused by external short-circuit current impact.

[0029] S103. Based on the multidimensional feature parameter set X and the weight vector W of the failure mode, establish a dynamic health assessment model for the equipment and calculate the comprehensive health score H(t) of the equipment.

[0030] This model integrates multidimensional feature parameters reflecting the real-time status of equipment with time-dependent factors that reflect long-term aging patterns, and quantifies them through a mathematical model containing multiple adjustable coefficients. The model not only assesses the current health status but also reflects the cumulative impact of the "journey" through aging factors and can predict the "future" degradation trend. Its dynamism is reflected in the fact that the aging coefficient λ can be updated using Bayesian methods, and the model parameters can self-optimize as the equipment group's operating data accumulates. This model realizes the condensation of complex, multidimensional equipment status information into an intuitive health score between 0 and 100, greatly facilitating the status perception and decision-making of maintenance personnel.

[0031] The comprehensive health score H(t) is defined by the following formula: H(t)=100*exp(-λ*t)*∏_{i=1}^{k}[1-α_i*(S_i(t) / S_{i,max})^β_i] In the formula: H(t) represents the overall health score of the device at time t, ranging from 0 to 100. The higher the score, the healthier the device. exp(-λ*t) is a baseline aging factor based on equipment operating time, and λ is an aging coefficient related to the inherent lifespan of the equipment; k represents the number of key characteristic parameters involved in the evaluation; S_i(t) is the severity value of the i-th key feature parameter deviating from the normal baseline after normalization at time t; S_{i,max} is the maximum allowable deviation severity threshold for the i-th key feature parameter; α_i is the sensitivity coefficient of the i-th feature parameter, 0<α_i≤1, and its value is related to the weight of the failure mode associated with this parameter in the failure mode analysis step. β_i is the morphological coefficient of the i-th feature parameter, β_i≥1, used to adjust the nonlinearity of the parameter's influence on health.

[0032] The key characteristic parameters include at least: the rate of temperature rise of the fused tube, the rate of increase of contact resistance, the ratio of the cumulative number of operations to the number of designed life operations, the three-phase current imbalance, and the effective value of leakage current on the insulator surface.

[0033] The initial value of the aging coefficient λ is set according to the mean time between failures (MTBF) provided by the equipment manufacturer, and is dynamically corrected during equipment operation using a Bayesian update method based on historical data of similar equipment.

[0034] The severity value S_i(t) is calculated as follows: S_i(t)=max(0,(P_i(t)-P_{i,base}) / (P_{i,limit}-P_{i,base})), where P_i(t) is the current value of the feature parameter, P_{i,base} is the normal baseline value, and P_{i,limit} is the safe operating limit value.

[0035] S104. Set the health score threshold H_th and the abnormal threshold of each characteristic parameter. When the real-time calculated H(t) is lower than H_th or any key characteristic parameter exceeds its abnormal threshold, trigger an alarm and match the characteristic parameter deviation mode with the failure mode library to locate the most likely cause of failure.

[0036] Fault diagnosis adopts a hybrid diagnostic mechanism based on case-based reasoning (CBR) and rule-based reasoning (RBR): First, the real-time feature patterns are matched with the historical fault case database for similarity; if the match is successful, the corresponding fault cause is output; if the match fails, predefined expert rules are triggered to make inference judgments.

[0037] It achieves a closed loop from condition assessment to fault diagnosis. The dual threshold triggering mechanism (comprehensive scoring threshold and single parameter threshold) ensures the sensitivity and comprehensiveness of early warning. The hybrid diagnostic mechanism based on case-based reasoning (CBR) and rule-based reasoning (RBR) combines the advantages of data-driven and knowledge-driven approaches. CBR can provide fast and accurate diagnoses for common faults by retrieving similar historical cases, while RBR can handle new, complex, or CBR-unmatched fault scenarios based on logical rules developed by experts. This hybrid mechanism significantly improves the coverage, accuracy, and efficiency of fault diagnosis.

[0038] S105. Generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance recommendations.

[0039] The analysis results from all the aforementioned steps are integrated into a structured and standardized diagnostic report. The report not only identifies the suspected fault causes and warning levels, but also provides maintenance recommendations, such as "It is recommended to schedule a power outage within 30 days, focusing on inspecting and cleaning the contacts." This makes maintenance decisions based on evidence and actions precise, realizing the digitalization, refinement, and intelligence of maintenance work.

[0040] Furthermore, the method also includes predicting the time point T_m when the health score H(t) drops to the preset maintenance threshold H_maint and the time point T_f when it drops to the failure threshold H_fail based on the historical change curve of the health score H(t) using trend extrapolation, thereby generating a predictive maintenance plan.

[0041] Extending health management from current state perception to future trend prediction, by performing regression analysis (such as linear regression, exponential regression, etc.) on the H(t) time series, the system can estimate the time point when the equipment health deteriorates to the point where planned maintenance H_maint is required or when functional failure H_fail may occur. Based on this prediction result, the system can automatically generate predictive maintenance work orders with time windows, guiding the maintenance team to prepare personnel, spare parts, and planned power outage windows in advance. This achieves a fundamental shift from "post-fault maintenance" or "periodic inspection" to "condition-based predictive maintenance," minimizing unplanned downtime, optimizing maintenance resources, and improving the economic benefits of the equipment throughout its entire life cycle.

[0042] Reference Figure 2 , Figure 2 This is a structural diagram of a wind farm collector line drop-out fuse fault diagnosis system according to an embodiment of the present invention. The system includes: The data acquisition and processing module 100 is used to deploy a multi-source sensor network on the target drop protection device to collect real-time operational and environmental data, including fuse surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. The module also performs filtering, noise reduction, and normalization preprocessing on the collected raw data to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}. The Failure Mode Analysis module 200 is used to identify the set of key failure modes F={F1,F2,...,Fm} that cause drop fuse failure based on historical failure data, using Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) methods, and to calculate the weight vector W={w1,w2,...,wm} of each failure mode using the Analytic Hierarchy Process (AHP), where ∑wi=1; The health assessment module 300 is used to establish a dynamic health assessment model for the equipment based on the multidimensional feature parameter set X and the weight vector W of the failure mode, and to calculate the comprehensive health score H(t) of the equipment. The fault diagnosis module 400 is used to set the health score threshold H_th and the abnormal threshold of each characteristic parameter. When the real-time calculated H(t) is lower than H_th or any key characteristic parameter exceeds its abnormal threshold, an early warning is triggered, and the most likely cause of the fault is located by matching the deviation mode of the characteristic parameter with the failure mode library. The output module 500 is used to generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance suggestions. The implementation process of the functions and roles of each module in the above system is detailed in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0043] Based on the above embodiments, this invention also provides an electronic device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the device may also include various necessary network interfaces, a power supply, and other components.

[0044] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by an execution terminal or processor, can implement the methods provided in this invention; the storage medium may include: a USB flash drive, a portable hard drive, or a read-only memory (ROM). Various media that can store program code, such as ROM, RAM, magnetic disks, or optical disks.

[0045] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying 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 a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for diagnosing drop-out fuse faults in wind farm collector lines, characterized in that, include: A multi-source sensor network is deployed on the target drop protection device to collect real-time operational and environmental data, including fuse tube surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. The collected raw data is preprocessed by filtering, denoising, and normalization to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}. Based on historical failure data, Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) are used to identify the set of key failure modes F={F1,F2,...,Fm} that lead to drop fuse failures. The Analytic Hierarchy Process (AHP) is then used to calculate the weight vector W={w1,w2,...,wm} for each failure mode, where ∑wi=1. Based on the multidimensional feature parameter set X and the weight vector W of failure modes, a dynamic health assessment model for the equipment is established, and the comprehensive health score H(t) of the equipment is calculated. Set the health score threshold H_th and the abnormal threshold of each feature parameter. When the real-time calculated H(t) is lower than H_th or any key feature parameter exceeds its abnormal threshold, an early warning is triggered, and the most likely cause of failure is located by matching the deviation mode of the feature parameter with the failure mode library. Generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance recommendations.

2. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 1, characterized in that, The surface temperature of the molten tube is collected by an infrared temperature sensor or a wireless temperature tag. The contact resistance of the conductive circuit is indirectly calculated by measuring the voltage drop of the circuit. The environmental pollution level is assessed by leakage current monitoring or an image recognition device.

3. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 1, characterized in that, The set of critical failure modes F includes at least: poor contact, melt aging, flashover of insulating components, jamming of mechanical mechanisms, abnormal installation stress, and mechanical damage caused by external short-circuit current impact.

4. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 1, characterized in that, The comprehensive health score H(t) is defined by the following formula: H(t)=100*exp(-λ*t)*∏_{i=1}^{k}[1-α_i*(S_i(t) / S_{i,max})^β_i] In the formula: H(t) represents the overall health score of the device at time t, ranging from 0 to 100. The higher the score, the healthier the device. exp(-λ*t) is a baseline aging factor based on equipment operating time, and λ is an aging coefficient related to the inherent lifespan of the equipment; k represents the number of key characteristic parameters involved in the evaluation; S_i(t) is the severity value of the i-th key feature parameter deviating from the normal baseline after normalization at time t; S_{i,max} is the maximum allowable deviation severity threshold for the i-th key feature parameter; α_i is the sensitivity coefficient of the i-th feature parameter, 0<α_i≤1, and its value is related to the weight of the failure mode associated with this parameter in the failure mode analysis step. β_i is the morphological coefficient of the i-th feature parameter, β_i≥1, used to adjust the nonlinearity of the parameter's influence on health.

5. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 4, wherein the key characteristic parameters include at least: The rate of temperature rise of the fused tube, the rate of increase of contact resistance, the ratio of the cumulative number of operations to the design life number of operations, the three-phase current imbalance, and the effective value of leakage current on the surface of the insulator.

6. The method for diagnosing drop-out protection faults in wind farm collector lines according to claim 4, wherein the initial value of the aging coefficient λ is set according to the mean time between failures (MTBF) provided by the equipment manufacturer, and is dynamically corrected during equipment operation using a Bayesian update method based on the group historical data of similar equipment.

7. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 4, characterized in that, The severity value S_i(t) is calculated as follows: S_i(t)=max(0,(P_i(t)-P_{i,base}) / (P_{i,limit}-P_{i,base})), where P_i(t) is the current value of the feature parameter, P_{i,base} is the normal baseline value, and P_{i,limit} is the safe operating limit value.

8. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 1, characterized in that, The fault diagnosis adopts a hybrid diagnostic mechanism based on case-based reasoning (CBR) and rule-based reasoning (RBR): First, the real-time feature patterns are matched with the historical fault case database for similarity; if the match is successful, the corresponding fault cause is output; if the match fails, predefined expert rules are triggered to make inference judgments.

9. The method for diagnosing drop-out fuse faults in wind farm collector lines according to claim 1, characterized in that, Also includes: Based on the historical change curve of the health score H(t), the time point T_m when it drops to the preset maintenance threshold H_maint and the time point T_f when it drops to the failure threshold H_fail are predicted by trend extrapolation, thereby generating a predictive maintenance plan.

10. A fault diagnosis system for drop-out fuses in wind farm collector lines, characterized in that, include: The data acquisition and processing module is used to deploy a multi-source sensor network on the target drop protection device to collect real-time operational and environmental data, including fuse tube surface temperature, conductive circuit contact resistance, fuse activation frequency, ambient temperature, humidity, wind speed, and pollution level. The module also performs filtering, noise reduction, and normalization preprocessing on the collected raw data to construct a multi-dimensional feature parameter set X={x1,x2,...,xn}. The Failure Mode Analysis module is used to identify the set of key failure modes F={F1,F2,...,Fm} that cause drop fuse failures based on historical failure data, using Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) methods, and to calculate the weight vector W={w1,w2,...,wm} for each failure mode using the Analytic Hierarchy Process (AHP), where ∑wi=1; The health assessment module is used to establish a dynamic health assessment model for equipment based on a multidimensional feature parameter set X and a failure mode weight vector W, and to calculate the comprehensive health score H(t) of the equipment. The fault diagnosis module is used to set the health score threshold H_th and the abnormal threshold of each characteristic parameter. When the real-time calculated H(t) is lower than H_th or any key characteristic parameter exceeds its abnormal threshold, an early warning is triggered, and the most likely cause of the fault is located by matching the deviation mode of the characteristic parameter with the failure mode library. The results output module is used to generate a diagnostic report that includes the device number, health score, fault warning level, suspected fault cause, and maintenance recommendations.

Images