A method for constructing a PT fuse fault diagnosis model in a plateau environment
By constructing a fault diagnosis model for PT fuses in high-altitude environments and combining multi-dimensional feature extraction and fusion, the problem of difficulty in analyzing the causes of PT fuse failures in high-altitude environments is solved, achieving efficient and accurate fault diagnosis and improving the adaptability and efficiency of fault diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HAIXI POWER SUPPLY
- Filing Date
- 2025-06-30
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies make it difficult to accurately analyze the causes of PT fuse failures in high-altitude environments, leading to repeated failures and affecting the formulation of maintenance and operation strategies.
A fault diagnosis model for PT fuse failure in high-altitude environments is constructed. By establishing electrical fault diagnosis sub-models, environmental fault diagnosis sub-models, and condition fault diagnosis sub-models, and combining the structural information and operating environment information of the voltage transformer, a cascade strategy is generated. Electrical, environmental, and physical parameters are collected, and multi-dimensional feature extraction and fusion are performed to determine the cause of the fault.
It improves the accuracy and adaptability of PT fuse fault diagnosis, reduces computation time and resource consumption, can accurately monitor high-stress areas and fault precursors, and significantly improves the efficiency and accuracy of fault diagnosis in high-altitude environments.
Smart Images
Figure CN120804776B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent power distribution system technology, and in particular to a method for constructing a PT fuse fault diagnosis model in high-altitude environments. Background Technology
[0002] A voltage transformer (PT) is similar to a transformer in that it is an instrument used to transform voltage. However, while transformers transform voltage to facilitate the transmission of electrical energy, they have a large capacity, typically measured in kilovolt-amperes (kVA) or megavolt-amperes (MVA). The primary purpose of a voltage transformer, on the other hand, is to power measuring instruments and relay protection devices. It is used to measure line voltage, power, and energy, or to protect valuable equipment, motors, and transformers in the event of a line fault. The basic structure of a voltage transformer is very similar to that of a transformer. It also has two windings: a primary winding and a secondary winding. Both windings are mounted on or wound around an iron core. Insulation exists between the two windings and between the windings and the iron core, providing electrical isolation between them. During operation, the primary winding N1 is connected in parallel to the line, and the secondary winding N2 is connected in parallel to the instrument or relay. Therefore, when measuring voltage on a high-voltage line, although the primary voltage is high, the secondary voltage is low, ensuring the safety of operators and instruments. There are many reasons for PT fuse failure, including system problems, problems with the voltage transformer itself, problems with the fuse itself, operational malfunctions, and environmental factors.
[0003] Currently, the method for analyzing the causes of faulty voltage transformers is to disassemble and observe them. Usually, if burn marks are observed on the primary winding of the voltage transformer, it indicates that there is a problem of excessive transient excitation current. It is believed that during the ground fault recovery period, the presence of low-frequency inrush current causes the magnetic flux of the transformer to continuously saturate under the action of single-phase voltage, which is the cause of the fault. Furthermore, the larger the capacitance to ground, the more severe the fault. However, this method cannot qualitatively determine the cause of fuse blowing, which affects the guidance of maintenance and operation strategy formulation, leading to repeated faults.
[0004] Therefore, there is a need to provide a method for constructing a PT fuse failure diagnosis model in high-altitude environments to achieve accurate analysis of the causes of PT fuse failures. Summary of the Invention
[0005] This invention provides a method for constructing a PT fuse failure diagnosis model in a high-altitude environment, comprising: establishing and training an electrical fault diagnosis sub-model, an environmental fault diagnosis sub-model, and a state fault diagnosis sub-model; acquiring structural information and operating environment information of the voltage transformer; generating a cascading strategy corresponding to the voltage transformer based on the structural information and operating environment information of the voltage transformer; constructing a PT fuse failure diagnosis model based on the cascading strategy, the electrical fault diagnosis sub-model, the environmental fault diagnosis sub-model, and the state fault diagnosis sub-model corresponding to the voltage transformer; collecting electrical parameters of the voltage transformer, wherein the electrical parameters include at least primary voltage, secondary voltage, and secondary current; collecting environmental parameters of the voltage transformer, wherein the environmental parameters include at least temperature, humidity, and air pressure; collecting the physical state of the voltage transformer, wherein the physical state includes at least vibration at multiple locations of the voltage transformer, discharge activity at multiple locations, and temperature at multiple locations of the fuse; and determining the cause of the PT fuse failure based on the electrical parameters, environmental parameters, and physical state of the voltage transformer using the PT fuse failure diagnosis model.
[0006] Furthermore, the electrical fault diagnosis sub-model includes a first input unit, a first feature extraction unit, and an electrical fault diagnosis unit. The first input unit is used to input and preprocess the primary-side voltage sequence, secondary-side voltage sequence, and secondary-side current sequence. The first feature extraction unit includes a first electrical feature extraction layer, a second electrical feature extraction layer, a third electrical feature extraction layer, and an electrical feature fusion layer. The first electrical feature extraction unit extracts the time-domain features, frequency-domain features, and voltage harmonic features of the primary-side voltage sequence. The second electrical feature extraction layer extracts the time-domain features, frequency-domain features, and voltage harmonic features of the secondary-side voltage sequence. The third electrical feature extraction layer extracts the time-domain features, frequency-domain features, and current harmonic features of the secondary-side current sequence. The electrical feature fusion layer fuses the outputs of the first, second, and third electrical feature extraction layers to generate an electrical feature vector. The electrical fault diagnosis unit is used to determine the probability of occurrence of various electrical faults based on the electrical feature vector.
[0007] Furthermore, the environmental fault diagnosis sub-model includes a second input unit, a second feature extraction unit, and an environmental guidance fault diagnosis unit. The second input unit is used to input temperature, humidity, and air pressure sequences and perform preprocessing. The second feature extraction unit includes an environmental temperature feature extraction layer, a humidity feature extraction layer, an air pressure extraction layer, and an environmental feature fusion layer. The environmental temperature feature extraction layer extracts the time-domain and frequency-domain features of the temperature sequence; the humidity feature extraction layer extracts the time-domain and frequency-domain features of the humidity sequence; the air pressure extraction layer extracts the time-domain and frequency-domain features of the air pressure sequence; and the environmental feature fusion layer fuses the outputs of the environmental temperature feature extraction layer, humidity feature extraction layer, and air pressure extraction layer to generate an environmental feature vector. The environmental guidance fault diagnosis unit is used to determine the probability of occurrence of various environmental guidance faults based on the environmental feature vector.
[0008] Furthermore, the state fault diagnosis sub-model includes a third input unit, a third feature extraction unit, and a state fault diagnosis unit. The third input unit is used to input and preprocess vibration sequences from multiple locations of the voltage transformer, discharge activity sequences from multiple locations of the fuse, and temperature sequences from multiple locations of the fuse. The third feature extraction unit includes a vibration feature extraction layer, a discharge activity extraction layer, a fuse temperature feature extraction layer, and a physical state feature fusion layer. The vibration feature extraction layer extracts and fuses the time-domain and frequency-domain features of the vibration sequence at each location. The discharge activity extraction layer extracts and fuses the time-domain and frequency-domain features of the discharge activity sequence at each location. The fuse temperature feature extraction layer extracts and fuses the time-domain and frequency-domain features of the temperature sequence at each location. The physical state feature fusion layer fuses the outputs of the vibration feature extraction layer, the discharge activity extraction layer, and the fuse temperature feature extraction layer to generate a physical state feature vector. The state fault diagnosis unit is used to determine the probability of occurrence of various physical faults based on the physical state feature vector.
[0009] Furthermore, based on the structural and operating environment information of the voltage transformer, a cascading strategy corresponding to the voltage transformer is generated, including: establishing a sample database, wherein the sample database includes cascading strategies of various sample voltage transformers in various sample operating environments; determining the target sample voltage transformer based on the structural information of the voltage transformer and the sample database; determining the target sample operating environment based on the operating environment information of the voltage transformer and various sample operating environments; and generating the corresponding cascading strategy for the voltage transformer based on the cascading strategy of the target sample voltage transformer in the target sample operating environment.
[0010] Furthermore, determining the cascading strategy for each sample voltage transformer in each sample operating environment includes: acquiring fault information of the sample voltage transformer over multiple historical time periods, wherein the fault information includes the occurrence frequency of electrical faults, environment-induced faults, and physical faults; determining the weights of electrical faults, environment-induced faults, and physical faults based on the fault information of the sample voltage transformer over multiple historical time periods; determining the correlation coefficient between any two of the electrical faults, environment-induced faults, and physical faults based on the fault information of the sample voltage transformer over multiple historical time periods; and determining the cascading strategy for the sample voltage transformer in the sample operating environment based on the weights of the electrical faults, environment-induced faults, and physical faults and the correlation coefficient between any two of the electrical faults, environment-induced faults, and physical faults.
[0011] Furthermore, based on the weights of electrical faults, environmentally guided faults, and physical faults, and the correlation coefficients of any two of these faults, a cascading strategy for the sample voltage transformers in the sample operating environment is determined. This includes: calculating the comprehensive correlation coefficient of electrical faults, environmentally guided faults, and physical faults based on the correlation coefficients of any two of these faults; calculating the priority values for judging electrical faults, environmentally guided faults, and physical faults based on the weights of these faults and the comprehensive correlation coefficients; determining the first-level fault type based on the priority values of these faults; and determining the cascading strategy for the sample voltage transformers in the sample operating environment based on the correlation coefficients of any two of these faults and the first-level fault type.
[0012] Furthermore, a three-dimensional model of the voltage transformer is established. Based on the operating environment information and the three-dimensional model of the voltage transformer, a finite element model of the voltage transformer is established. Finite element analysis is performed based on the finite element model of the voltage transformer to determine the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer. Based on the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer, multiple locations for vibration monitoring are determined. Based on the multiple locations for vibration monitoring, vibration sequences at multiple locations of the voltage transformer are collected.
[0013] Furthermore, discharge activity sequences from multiple locations are collected, including: performing finite element analysis based on the finite element model of the voltage transformer to determine the electric field intensity distribution inside the voltage transformer; determining multiple locations for discharge activity monitoring based on the electric field intensity distribution inside the voltage transformer; and collecting discharge activity sequences from multiple locations based on the multiple locations for discharge activity monitoring.
[0014] Furthermore, temperature sequences at multiple locations of the fuse are collected, including: performing finite element analysis based on the finite element model of the voltage transformer to determine the temperature distribution inside the fuse; determining multiple locations for temperature monitoring based on the temperature distribution inside the fuse; and collecting temperature sequences at multiple locations of the fuse based on the multiple locations for temperature monitoring.
[0015] Compared with existing technologies, the method for constructing a PT fuse fault diagnosis model in a high-altitude environment provided in this specification has at least the following advantages:
[0016] 1. Electrical parameters (voltage, current) reflect the operating status of equipment, environmental parameters (temperature, humidity, air pressure) reveal external influencing factors, and physical conditions (vibration, discharge, temperature) reveal internal damage mechanisms. If only electrical parameters are monitored, the fuse may be mistakenly identified as an overload; by combining environmental parameters (low air pressure leads to a decrease in insulation strength) and physical conditions (enhanced partial discharge), the insulation breakdown caused by the environment can be accurately diagnosed.
[0017] 2. Based on the structural and operating environment information of the voltage transformer, a corresponding cascading strategy is generated, and this strategy is combined with multiple sub-models to construct a PT fuse fault diagnosis model. This cascading strategy allows for flexible adjustment of the model's diagnostic approach according to the characteristics and operating environment of different voltage transformers, improving the model's adaptability and accuracy. Traditional methods require manual inspection of electrical, environmental, and physical conditions, taking several hours; while the model can complete the diagnosis in minutes. In actual fault diagnosis scenarios, running and calculating the sub-models corresponding to all fault types would consume significant time and computational resources. This approach first runs the sub-model corresponding to the first-level fault type. Only when the probability of each fault type output is less than the probability threshold is the sub-model corresponding to the second and third-level fault types run sequentially, thus saving considerable computational time and resources.
[0018] 3. The electrical fault diagnosis sub-model extracts and fuses the features of primary and secondary voltage and current respectively to avoid information loss. It can diagnose multiple electrical faults at the same time (such as short circuit, overload, resonance, insulation breakdown, etc.). Through multi-dimensional feature extraction, hierarchical fusion and probability output, it significantly improves the diagnostic accuracy of PT fuse electrical faults in high-altitude environments.
[0019] 4. Traditional methods may involve randomly placing sensors, while this method can precisely locate high-stress areas (such as iron cores and winding connections), key locations for discharge activity monitoring, and key locations for temperature monitoring, thus improving monitoring efficiency. Large diurnal temperature variations at high altitudes can cause material expansion and contraction, leading to changes in mechanical stress; precise monitoring can detect such early signs of faults. Finite element analysis of the electric field intensity distribution identifies key locations for discharge activity monitoring. High electric field areas (such as winding ends and weak insulation points) are more prone to partial discharge; precise monitoring can improve the fault detection rate. Under low air pressure, heat dissipation efficiency decreases, and fuses are more prone to overheating; temperature sequences can reflect the current-carrying capacity and contact status of fuses. Attached Figure Description
[0020] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:
[0021] Figure 1 This is a flowchart illustrating a method for constructing a PT fuse fault diagnosis model in a high-altitude environment, as shown in one embodiment of this application;
[0022] Figure 2 This is a structural diagram of an electrical fault diagnosis sub-model shown in one embodiment of this application;
[0023] Figure 3 This is a structural diagram of an environmental fault diagnosis sub-model shown in one embodiment of this application;
[0024] Figure 4 This is a structural diagram of a state fault diagnosis sub-model shown in one embodiment of this application. Detailed Implementation
[0025] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below.
[0026] Figure 1 This is a flowchart illustrating a method for constructing a PT fuse fault diagnosis model in a high-altitude environment, as shown in one embodiment of this application. This method belongs to intelligent power distribution systems, such as... Figure 1 As shown, a method for constructing a PT fuse fault diagnosis model in a high-altitude environment may include the following steps.
[0027] Step 110: Establish and train the electrical fault diagnosis sub-model, the environmental fault diagnosis sub-model, and the condition fault diagnosis sub-model.
[0028] Figure 2 This is a structural diagram of an electrical fault diagnosis sub-model shown in one embodiment of this application, as follows: Figure 2As shown, in some embodiments, the electrical fault diagnosis sub-model includes a first input unit, a first feature extraction unit, and an electrical fault diagnosis unit;
[0029] The first input unit is used to input the primary side voltage sequence, the secondary side voltage sequence, and the secondary side current sequence and perform preprocessing.
[0030] The first feature extraction unit includes a first electrical feature extraction layer, a second electrical feature extraction layer, a third electrical feature extraction layer, and an electrical feature fusion layer. The first electrical feature extraction unit is used to extract the time-domain features, frequency-domain features, and voltage harmonic features of the primary voltage sequence. The second electrical feature extraction layer is used to extract the time-domain features, frequency-domain features, and voltage harmonic features of the secondary voltage sequence. The third electrical feature extraction layer is used to extract the time-domain features, frequency-domain features, and current harmonic features of the secondary current sequence. The electrical feature fusion layer is used to fuse the outputs of the first electrical feature extraction layer, the second electrical feature extraction layer, and the third electrical feature extraction layer to generate an electrical feature vector.
[0031] The electrical fault diagnosis unit is used to determine the probability of occurrence of various electrical faults (e.g., single-phase grounding fault, secondary side short circuit, switching overvoltage, etc.) based on electrical feature vectors.
[0032] Specifically, the first input unit serves as the data entry point for the entire electrical fault diagnosis sub-model. It is responsible for receiving and processing the electrical parameter sequences related to the voltage transformer, providing a suitable data format for subsequent feature extraction and fault diagnosis. The first input unit receives the primary side voltage sequence, the secondary side voltage sequence, and the secondary side current sequence. These sequences are electrical parameter values arranged in chronological order, reflecting the operating status of the voltage transformer at different points in time. Since the actual collected electrical parameter sequences may be affected by various noises, such as electromagnetic interference and measurement errors, filtering algorithms (such as mean filtering and median filtering) are needed during preprocessing to remove noise and improve data quality. The dimensions and numerical ranges of different electrical parameters vary significantly. To eliminate the impact of these differences on subsequent model training, the sequence data needs to be normalized. For example, the value of each sequence is mapped to the interval [0,1] or [-1,1] to give the data the same scale. The primary voltage sequence, secondary voltage sequence, and secondary current sequence are aligned to ensure that they are synchronized in time, meaning that the data at corresponding time points are collected simultaneously, so that the relationship between them can be accurately analyzed later.
[0033] Meaningful features are extracted from the preprocessed electrical parameter sequence. These features can characterize the operating status and potential fault information of the voltage transformer. By extracting features from the primary voltage, secondary voltage, and secondary current sequences separately and fusing these features, a comprehensive electrical feature vector is generated, providing richer information for fault diagnosis.
[0034] The time-domain characteristics of the primary voltage sequence can include the voltage mean, variance, and peak value. The first electrical feature extraction layer calculates the average value of the primary voltage sequence over a period of time, reflecting the overall voltage level. An abnormal mean value may indicate voltage shifts or other problems on the primary side. Variance measures the dispersion of the primary voltage sequence; a larger variance indicates more severe voltage fluctuations. For example, a sudden increase in the primary voltage variance may indicate interference or a fault on the primary side. The peak value is the maximum value of the primary voltage sequence, and changes in the peak value can reflect the transient characteristics of the voltage. An abnormal increase in the primary voltage peak value may indicate an overvoltage situation on the primary side.
[0035] The first electrical feature extraction layer converts the primary voltage time-domain sequence into a frequency-domain sequence, obtaining the voltage spectrum. By analyzing the spectrum, the amplitude and phase information of different frequency components can be observed, the main frequency components and existing harmonics in the voltage can be identified, and the content of each harmonic in the primary voltage can be calculated. Excessive harmonic content may lead to voltage waveform distortion.
[0036] The second and third electrical feature extraction layers have similar structures to the first electrical feature extraction layer, and will not be described in detail here.
[0037] The outputs of the first, second, and third electrical feature extraction layers are concatenated to generate a comprehensive electrical feature vector. Through feature fusion, the interrelationships between different electrical parameters can be comprehensively considered, improving the accuracy and reliability of fault diagnosis. For example, the harmonic characteristics of the primary voltage and the harmonic characteristics of the secondary current may be correlated. The fused feature vector can better reflect this correlation, helping to more accurately determine the fault type.
[0038] An electrical fault diagnosis unit can include a multilayer perceptron neural network, trained using a large amount of historical electrical parameter sequences and corresponding fault label data. During training, the parameters of the electrical fault diagnosis unit (such as the weights and biases of the neural network) are adjusted to enable it to accurately learn the relationship between electrical features and fault types. Cross-validation and regularization are used to optimize the electrical fault diagnosis unit, preventing overfitting and improving its generalization ability. A well-trained electrical fault diagnosis unit can output the probability of various electrical faults based on the input electrical feature vector. For example, it might output "the probability of a primary short-circuit fault is 70%, the probability of a secondary short-circuit fault is 20%, and the probability of a ferroresonant fault is 10%."
[0039] The training process for the electrical fault diagnosis sub-model may include:
[0040] Initialize model parameters: Set initial parameters for the electrical fault diagnosis sub-model, such as the weights and biases of the neural network.
[0041] Forward propagation: Input the training data into the electrical fault diagnosis sub-model and output the probability of each fault type.
[0042] Loss calculation: Based on the fault probability output by the electrical fault diagnosis sub-model and the actual fault label, calculate the value of the loss function, which includes the cross-entropy loss function.
[0043] Backpropagation and parameter update: Based on the gradient information of the loss function, the parameters of the electrical fault diagnosis sub-model are updated, so that the performance of the electrical fault diagnosis sub-model on the training data is continuously optimized.
[0044] Iterative training: Repeat the forward propagation, loss calculation and parameter update process until the electrical fault diagnosis sub-model converges or reaches the preset number of training rounds.
[0045] Figure 3 This is a structural diagram of an environmental fault diagnosis sub-model shown in one embodiment of this application, as follows: Figure 3 As shown, in some embodiments, the environmental fault diagnosis sub-model includes a second input unit, a second feature extraction unit, and an environmental-guided fault diagnosis unit.
[0046] The second input unit is used to input temperature, humidity and air pressure sequences and perform preprocessing.
[0047] The second feature extraction unit includes an ambient temperature feature extraction layer, a humidity feature extraction layer, an air pressure extraction layer, and an environmental feature fusion layer. The ambient temperature feature extraction layer is used to extract the time-domain and frequency-domain features of the temperature sequence, the humidity feature extraction layer is used to extract the time-domain and frequency-domain features of the humidity sequence, the air pressure extraction layer is used to extract the time-domain and frequency-domain features of the air pressure sequence, and the environmental feature fusion layer is used to fuse the outputs of the ambient temperature feature extraction layer, the humidity feature extraction layer, and the air pressure extraction layer to generate an environmental feature vector.
[0048] The environment-guided fault diagnosis unit is used to determine the probability of occurrence of various environment-guided faults (e.g., humid environment leading to decreased PT insulation performance, causing surface discharge or flashover, environment-induced ferroresonance, etc.) based on environmental feature vectors.
[0049] The structure and principle of the environmental fault diagnosis sub-model are similar to those of the electrical fault diagnosis sub-model, and will not be elaborated here.
[0050] Figure 4 This is a structural diagram of a state fault diagnosis sub-model shown in one embodiment of this application, as follows: Figure 4 As shown, in some embodiments, the state fault diagnosis sub-model includes a third input unit, a third feature extraction unit, and a state fault diagnosis unit.
[0051] The third input unit is used to input and preprocess the vibration sequence of multiple positions of the voltage transformer, the discharge activity sequence of multiple positions of the fuse, and the temperature sequence of multiple positions of the fuse.
[0052] The third feature extraction unit includes a vibration feature extraction layer, a discharge activity extraction layer, a fuse temperature feature extraction layer, and a physical state feature fusion layer. The vibration feature extraction layer is used to extract and fuse the time-domain and frequency-domain features of the vibration sequence at each location. The discharge activity extraction layer is used to extract and fuse the time-domain and frequency-domain features of the discharge activity sequence at each location. The fuse temperature feature extraction layer is used to extract and fuse the time-domain and frequency-domain features of the temperature sequence at each location. The physical state feature fusion layer is used to fuse the outputs of the vibration feature extraction layer, the discharge activity extraction layer, and the fuse temperature feature extraction layer to generate a physical state feature vector.
[0053] The condition fault diagnosis unit is used to determine the probability of occurrence of various physical faults (e.g., winding deformation, insulation damage, contact / terminal overheating, etc.) based on physical state feature vectors.
[0054] The structure and principle of the condition-based fault diagnosis sub-model are similar to those of the electrical fault diagnosis sub-model, and will not be elaborated here.
[0055] Step 120: Obtain the structural information and operating environment information of the voltage transformer.
[0056] Specifically, structural information about voltage transformers can be obtained from design drawings, technical specifications, installation manuals, and other documents. These documents detail the overall structure of the voltage transformer, the dimensions, materials, connection methods, and electrical parameters of each component. For example, design drawings will indicate key information such as the number of turns in the winding, the shape and material of the core, and the insulation structure.
[0057] Environmental monitoring equipment can be installed near the voltage transformer installation location to monitor environmental parameters such as temperature, humidity, and air pressure in real time. For example, temperature and humidity sensors can be used to monitor changes in ambient temperature and humidity, and barometers can be used to measure air pressure.
[0058] Step 130: Based on the structural information and operating environment information of the voltage transformer, generate the cascading strategy corresponding to the voltage transformer.
[0059] Specifically, it includes:
[0060] Establish a sample database, which includes cascading strategies for various sample voltage transformers in various sample operating environments;
[0061] Based on the structural information of voltage transformers and the sample database, the target sample voltage transformer is determined.
[0062] Based on the operating environment information of voltage transformers and the operating environments of various samples, the operating environment of the target sample is determined.
[0063] Based on the cascading strategy of the target sample voltage transformer in the target sample operating environment, the corresponding cascading strategy of the voltage transformer is generated.
[0064] Specifically, the sample voltage transformers can be from different manufacturers and of different models. Various typical operating environments are considered, including different temperature ranges, humidity conditions, altitudes, and electromagnetic interference intensities, as the sample operating environments.
[0065] The cascading strategy of the sample operating environment can reflect the execution logic of fault determination based on the electrical fault diagnosis sub-model, the environmental fault diagnosis sub-model, and the state fault diagnosis sub-model.
[0066] In some embodiments, determining a cascading strategy for each sample voltage transformer in each sample operating environment includes:
[0067] The fault information of the sample voltage transformer is obtained in multiple historical time periods. The fault information includes the number of occurrences of electrical faults, environmentally guided faults and physical faults. Specifically, for each historical time period, the number of occurrences of each type of electrical fault can be counted, and the sum of the number of occurrences of each type of electrical fault is taken as the number of occurrences of electrical faults. The method for determining the number of occurrences of environmentally guided faults and physical faults is similar and will not be described in detail here.
[0068] Based on the fault information of sample voltage transformers in multiple historical time periods, the weights of electrical faults, environmentally guided faults, and physical faults are determined.
[0069] Based on the fault information of sample voltage transformers in multiple historical time periods, the correlation coefficient between any two of electrical faults, environmentally guided faults, and physical faults is determined. Specifically, the correlation coefficient can be determined based on the number of occurrences of electrical faults, environmentally guided faults, and physical faults of sample voltage transformers in multiple historical time periods, according to the correlation coefficient (e.g., Spearman's rank correlation coefficient) calculation formula. The correlation coefficient ranges from [-1, 1].
[0070] Based on the weights of electrical faults, environmentally guided faults, and physical faults, and the correlation coefficients of any two of these faults, the cascading strategy of the sample voltage transformers in the sample operating environment is determined.
[0071] Specifically, the weights of electrical faults, environmentally driven faults, and physical faults can be determined using the following formula:
[0072]
[0073] Among them, w i N represents the weight of the i-th type of fault. (i,t) Let n be the number of times the i-th type of fault occurs in the t-th historical time period. (j,t) Let T be the number of times the j-th type of fault occurs in the t-th historical time period, and T be the total number of historical time periods.
[0074] Understandably, dividing the total number of occurrences of type i faults by the total number of occurrences of all three types of faults yields the weight w of type i faults. i This weight reflects the relative importance of the i-th type of fault in the entire fault system. The larger the weight, the higher the frequency of this type of fault occurring throughout the historical period, and the greater the possibility of causing the PT to trip.
[0075] In some embodiments, based on the weights of electrical faults, environment-guided faults, and physical faults, and the correlation coefficient of any two of the electrical faults, environment-guided faults, and physical faults, a cascading strategy for the sample voltage transformers in the sample operating environment is determined, including:
[0076] Based on the correlation coefficients of any two of the electrical faults, environmentally guided faults, and physical faults, calculate the comprehensive correlation coefficient of the electrical faults, environmentally guided faults, and physical faults. For example, for electrical faults, the average of the absolute values of the correlation coefficients between electrical faults and environmentally guided faults and between electrical faults and physical faults can be used as the comprehensive correlation coefficient of the electrical faults. The calculation method for the comprehensive correlation coefficients of environmentally guided faults and physical faults is similar and will not be repeated here.
[0077] Based on the weights of electrical faults, environmentally guided faults, and physical faults, and the comprehensive correlation coefficients of electrical faults, environmentally guided faults, and physical faults, the priority values for judging electrical faults, environmentally guided faults, and physical faults are calculated.
[0078] Based on the priority values for electrical faults, environmentally guided faults, and physical faults, the first-level fault type is determined;
[0079] Based on the correlation coefficients of any two of electrical faults, environmentally guided faults, and physical faults, and the first-level fault type, the cascading strategy of the sample voltage transformers in the sample operating environment is determined.
[0080] Specifically, the priority value can be calculated using the following formula:
[0081] P i =w i ×r i
[0082] Where, p i Let r be the priority value for determining the i-th type of fault. i Let be the comprehensive correlation coefficient for the i-th type of fault.
[0083] Understandable, w in the formula i This represents the weight of the i-th type of fault, calculated based on fault information from sample voltage transformers over multiple historical time periods. It reflects the relative importance of this type of fault within the overall fault system. A larger weight indicates a higher frequency of occurrence of this type of fault in historical operation, potentially a greater impact on voltage transformer operation, and thus requiring more attention in fault diagnosis. iThis is the comprehensive correlation coefficient for the i-th type of fault. It is obtained by calculating the average absolute value of the correlation coefficients between this type of fault and the other two types of faults, reflecting the degree of association between this type of fault and other faults. The larger the comprehensive correlation coefficient, the closer the connection between this type of fault and other faults, indicating that they may influence each other or occur together when the faults occur. Multiplying the weight and the comprehensive correlation coefficient yields the judgment priority value, which comprehensively considers the importance and correlation of the fault. The judgment priority value calculated in this way can more comprehensively reflect the priority of each type of fault in fault diagnosis, avoiding diagnostic bias caused by considering only a single factor.
[0084] The fault with the highest priority value can be designated as the first-level fault type.
[0085] The second-level fault type can be determined based on the correlation coefficient between the remaining fault types and the first-level fault type. For example, the fault type with a higher correlation coefficient with the first-level fault type can be designated as the second-level fault type, and the remaining fault types can be designated as the third-level fault types. The cascading strategy can be as follows: during fault diagnosis, first run the sub-model corresponding to the first-level fault type. When the probability of each fault output by the sub-model corresponding to the first-level fault type is less than the probability threshold, then run the sub-model corresponding to the second-level fault type. When the probability of each fault output by the sub-model corresponding to the second-level fault type is less than the probability threshold, then run the sub-model corresponding to the third-level fault type. The probability threshold can be set based on experience or experimental data; for example, a probability threshold of 30% can be used.
[0086] In real-world fault diagnosis scenarios, running and calculating sub-models for all fault types would consume significant time and computational resources. This solution first runs the sub-models for the first-level fault types. Only when the probability of each fault type is less than a probability threshold is the sub-model for the second and third-level fault types run sequentially, thus saving considerable computational time and resources.
[0087] The sample database can also include structural information of various sample voltage transformers, such as winding type (e.g., single-winding, double-winding, or multi-winding), insulation medium (e.g., oil-immersed, dry-type, SF6 gas-insulated), rated voltage and current values of the primary and secondary sides, and structural form (e.g., electromagnetic, capacitive, photoelectric). The structural information of the sample voltage transformers is represented as a structural feature vector, facilitating subsequent similarity calculations. For example, numerical codes can be used to represent winding type (e.g., single-winding = 1, double-winding = 2, multi-winding = 3), insulation medium type (e.g., oil-immersed = 1, dry-type = 2, SF6 gas-insulated = 3), etc. These coded values are concatenated to generate the structural feature vector of the sample voltage transformer. Similarly, the structural information of the voltage transformer is represented as a structural feature vector. The Euclidean distance between the structural feature vector of the current voltage transformer and the structural feature vector of the sample voltage transformers is calculated to obtain the structural feature vector similarity. The sample voltage transformer with the highest structural feature vector similarity is selected as the target sample voltage transformer.
[0088] It can generate environmental feature vectors of the sample operating environment, which can include specific values of temperature, humidity, and air pressure. Similarly, it can generate environmental feature vectors of the current voltage transformer's operating environment information. It can calculate the Euclidean distance between the environmental feature vectors of the sample operating environment and the current voltage transformer's environmental feature vectors to obtain the environmental feature vector similarity. The sample operating environment with the highest environmental feature vector similarity is taken as the target sample operating environment.
[0089] The cascading strategy of the target sample voltage transformer in the target sample operating environment can be used as the corresponding cascading strategy for the voltage transformer.
[0090] Step 140: Based on the cascade strategy, electrical fault diagnosis sub-model, environmental fault diagnosis sub-model and state fault diagnosis sub-model corresponding to the voltage transformer, construct the PT fuse fault diagnosis model.
[0091] Step 150: Collect the electrical parameters of the voltage transformer.
[0092] The electrical parameters include at least the primary side voltage, the secondary side voltage, and the secondary side current.
[0093] Step 160: Collect environmental parameters of the voltage transformer.
[0094] Among them, environmental parameters include at least temperature, humidity and air pressure.
[0095] Step 170: Collect the physical state of the voltage transformer.
[0096] The physical state includes at least the vibration at multiple locations of the voltage transformer, the discharge activity at multiple locations, and the temperature at multiple locations of the fuse.
[0097] In some embodiments, the vibration sequence of a voltage transformer at multiple locations is collected, including:
[0098] Establish a three-dimensional model of the voltage transformer;
[0099] Based on the operating environment information and three-dimensional model of the voltage transformer, a finite element model of the voltage transformer is established. Finite element analysis is performed based on the finite element model of the voltage transformer to determine the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer. Based on the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer, multiple locations for vibration monitoring are determined.
[0100] Vibration sequences were collected from multiple locations of the voltage transformer based on the vibration monitoring.
[0101] Specifically, using CAD software, a detailed 3D model of the voltage transformer is created based on its actual dimensions and structure. This model should include all key components, such as windings, core, insulation structure, casing, and connectors. Correct material properties are assigned to each component in the model, including density, elastic modulus, Poisson's ratio, permeability, dielectric constant, conductivity, breakdown field strength, and thermal conductivity, to ensure the model's physical behavior matches reality. The 3D model is divided into a finite number of small elements (mesh), with the mesh density and type optimized according to the analysis requirements. Based on the actual operating current and voltage of the voltage transformer, corresponding electromagnetic excitation is applied. Electromagnetic field analysis is performed to calculate the magnetic field distribution inside the voltage transformer, and then the electromagnetic force is calculated. Thermal stress, pressure loads, and boundary conditions are applied based on the voltage transformer's operating environment information. Finite element software (such as ANSYS Maxwell, COMSOL Multiphysics, etc.) is used to perform electromagnetic field analysis and calculate the electromagnetic force distribution inside the voltage transformer. The electromagnetic force is then applied as a load to the structural model for structural mechanics analysis, calculating the mechanical stress distribution and deformation. Electromagnetic force is one of the main excitation sources causing vibration in voltage transformers. Electromagnetic field analysis identifies areas of concentrated electromagnetic force, serving as multiple locations for vibration monitoring. For example, at the winding ends: the leakage magnetic field at the ends is relatively large, resulting in a significant electromagnetic force effect. At the core joints: uneven magnetic flux density may lead to increased local electromagnetic force. At the gap between the winding and the core: electromagnetic force may cause relative movement between the winding and the core. Areas with high mechanical stress are prone to deformation or fatigue, thus inducing vibration. Structural mechanics analysis identifies areas of concentrated mechanical stress, serving as multiple locations for vibration monitoring. For example, the winding support structure: support parts may bear significant mechanical stress. Core fixing points: stress concentration may exist at the connection between the core and the outer shell. Insulation structure: insulation materials may undergo slight deformation under mechanical stress, affecting vibration characteristics.
[0102] Sensors can be set at each vibration monitoring location to collect vibration sequences at that location.
[0103] In some embodiments, collecting discharge activity sequences at multiple locations includes:
[0104] Finite element analysis is performed based on the finite element model of the voltage transformer to determine the electric field intensity distribution inside the voltage transformer. Based on the electric field intensity distribution inside the voltage transformer, multiple locations for monitoring discharge activities are determined.
[0105] Discharge activity sequences were collected from multiple locations where discharge activity was monitored.
[0106] Specifically, an electrostatic field solver (such as the Electrostatic module in ANSYS Maxwell) is used to calculate the steady-state electric field distribution, or a transient electric field solver is used to analyze the electric field changes under transient overvoltages, outputting a distribution cloud map of the electric field intensity. Partial discharge is prone to occur in areas where the electric field intensity exceeds the material's tolerance threshold; for example, at the edges of the insulating cardboard at the winding ends; or in the air gap between the core and the winding. Abrupt geometric changes (such as sharp corners or notches) or material inhomogeneities can lead to electric field concentration, for example, in the form of tiny gaps in the inter-turn insulation of the winding; or on the surface of the insulator, due to contamination or cracks. Multiple locations for monitoring discharge activity should be placed in areas where the electric field intensity exceeds 80% of the material's tolerance threshold, and multiple locations for monitoring discharge activity should also be placed along paths where discharge may propagate along the insulation surface or gas gaps, for example: surface discharge paths from the winding ends to the outer casing; discharge channels in the SF6 air gap of a gas-insulated switchgear (GIS).
[0107] Sensors can be set up at each location where discharge activity is monitored to collect the discharge activity sequence at that location.
[0108] In some embodiments, collecting temperature sequences at multiple locations of the fuse includes:
[0109] Finite element analysis based on the finite element model of the voltage transformer was performed to determine the temperature distribution inside the fuse, and based on the temperature distribution inside the fuse, multiple locations for temperature monitoring were determined.
[0110] Temperature sequences of the fuse are collected from multiple locations where temperature monitoring is performed.
[0111] Specifically, steady-state or transient thermal analysis solvers (such as ANSYS Mechanical and COMSOL Multiphysics) are used to calculate the temperature distribution and output a temperature distribution contour map. Multiple locations for temperature monitoring can be set in areas where the temperature exceeds a temperature threshold or the temperature gradient exceeds a temperature gradient threshold. The temperature threshold and temperature gradient threshold can be set based on human experience. For example, areas with the highest melt resistance loss (such as abrupt changes in cross-section or connection points) are prone to becoming high-temperature points, potentially leading to premature fuse melting or thermal aging. Areas with large temperature gradients in the arc-extinguishing medium (such as around the melt) may affect arc-extinguishing performance, causing arc reignition. Localized overheating caused by contact resistance may lead to connection failure or uneven melt melting.
[0112] Sensors can be set up at each location where temperature monitoring is performed to collect temperature sequences for that location.
[0113] Step 180: Determine the cause of PT fuse failure based on the electrical parameters, environmental parameters and physical state of the voltage transformer using the PT fuse failure diagnosis model.
[0114] For example, suppose the cascading strategy determined in step 130 is:
[0115] Level 1 Fault Type: Electrical Fault
[0116] Level 2 Fault Type: Environmental Fault
[0117] Level 3 fault type: Status fault.
[0118] When applying:
[0119] Input data: Real-time electrical parameters, environmental parameters, and status parameters of the voltage transformer.
[0120] Level 1 diagnosis: Run the electrical fault diagnosis sub-model and output the probability of electrical faults.
[0121] If the probability is ≥30%, it is diagnosed as an electrical fault.
[0122] If the probability is less than 30%, proceed to the second-level diagnosis.
[0123] Second-level diagnosis: Operating environment fault diagnosis sub-model, outputting the probability of environmental faults.
[0124] If the probability is ≥30%, it is diagnosed as an environmental fault.
[0125] If the probability is less than 30%, proceed to the third-level diagnosis.
[0126] The third level of diagnosis: the operational status fault diagnosis sub-model, which outputs the probability of status faults and performs the final diagnosis.
[0127] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and are considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.
Claims
1. A method for constructing a PT fuse fault diagnosis model in a high-altitude environment, characterized in that, include: Establish and train electrical fault diagnosis sub-models, environmental fault diagnosis sub-models, and condition fault diagnosis sub-models; Obtain structural and operating environment information of the voltage transformer; Based on the structural and operating environment information of the voltage transformer, a cascading strategy corresponding to the voltage transformer is generated. Based on the cascade strategy, electrical fault diagnosis sub-model, environmental fault diagnosis sub-model and state fault diagnosis sub-model corresponding to voltage transformers, a PT fuse fault diagnosis model is constructed. The electrical parameters of the voltage transformer are collected, wherein the electrical parameters include at least the primary side voltage, the secondary side voltage, and the secondary side current; Collect environmental parameters of the voltage transformer, wherein the environmental parameters include at least temperature, humidity and air pressure; The physical state of the voltage transformer is collected, wherein the physical state includes at least the vibration at multiple locations of the voltage transformer, the discharge activity at multiple locations, and the temperature at multiple locations of the fuse. The cause of PT fuse failure is determined by using a PT fuse failure diagnosis model based on the electrical parameters, environmental parameters, and physical state of the voltage transformer. Among them, based on the structural information and operating environment information of the voltage transformer, a cascading strategy corresponding to the voltage transformer is generated, including: Establish a sample database, which includes cascading strategies for various sample voltage transformers in various sample operating environments; Based on the structural information of voltage transformers and the sample database, the target sample voltage transformer is determined. Based on the operating environment information of voltage transformers and the operating environments of various samples, the operating environment of the target sample is determined. Based on the cascading strategy of the target sample voltage transformer in the target sample operating environment, generate the corresponding cascading strategy for the voltage transformer. Determine the cascading strategy for each sample voltage transformer in each sample operating environment, including: Fault information of sample voltage transformers over multiple historical time periods is obtained, wherein the fault information includes the number of occurrences of electrical faults, environmentally triggered faults, and physical faults; Based on the fault information of sample voltage transformers in multiple historical time periods, the weights of electrical faults, environmentally guided faults, and physical faults are determined. Based on the fault information of sample voltage transformers in multiple historical time periods, determine the correlation coefficient between any two of electrical faults, environmentally guided faults, and physical faults. Calculate the comprehensive correlation coefficient of electrical faults, environmentally guided faults, and physical faults based on the correlation coefficients of any two of them. Based on the weights of electrical faults, environmentally guided faults, and physical faults, and the comprehensive correlation coefficients of electrical faults, environmentally guided faults, and physical faults, the priority values for judging electrical faults, environmentally guided faults, and physical faults are calculated. Based on the priority values of electrical faults, environmentally guided faults, and physical faults, the first-level fault type is determined. Specifically, the fault with the highest priority value is taken as the first-level fault type. Based on the correlation coefficients of any two of the electrical faults, environmentally guided faults, and physical faults, and the first-level fault type, a cascading strategy for the sample voltage transformer in the sample operating environment is determined. Specifically, the fault type with a larger correlation coefficient with the first-level fault type is designated as the second-level fault type, and the remaining fault types are designated as the third-level fault types. The cascading strategy is as follows: during fault diagnosis, the sub-model corresponding to the first-level fault type is run first. When the probability of each fault output by the sub-model corresponding to the first-level fault type is less than the probability threshold, the sub-model corresponding to the second-level fault type is run. When the probability of each fault output by the sub-model corresponding to the second-level fault type is less than the probability threshold, the sub-model corresponding to the third-level fault type is run.
2. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 1, characterized in that, The electrical fault diagnosis sub-model includes a first input unit, a first feature extraction unit, and an electrical fault diagnosis unit; The first input unit is used to input the primary side voltage sequence, the secondary side voltage sequence, and the secondary side current sequence and perform preprocessing. The first feature extraction unit includes a first electrical feature extraction layer, a second electrical feature extraction layer, a third electrical feature extraction layer, and an electrical feature fusion layer. The first electrical feature extraction unit is used to extract the time-domain features, frequency-domain features, and voltage harmonic features of the primary-side voltage sequence. The second electrical feature extraction layer is used to extract the time-domain features, frequency-domain features, and voltage harmonic features of the secondary-side voltage sequence. The third electrical feature extraction layer is used to extract the time-domain features, frequency-domain features, and current harmonic features of the secondary-side current sequence. The electrical feature fusion layer is used to fuse the outputs of the first electrical feature extraction layer, the second electrical feature extraction layer, and the third electrical feature extraction layer to generate an electrical feature vector. The electrical fault diagnosis unit is used to determine the probability of occurrence of various electrical faults based on electrical feature vectors.
3. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 1, characterized in that, The environmental fault diagnosis sub-model includes a second input unit, a second feature extraction unit, and an environmental-guided fault diagnosis unit. The second input unit is used to input temperature sequence, humidity sequence and air pressure sequence and perform preprocessing; The second feature extraction unit includes an ambient temperature feature extraction layer, a humidity feature extraction layer, an air pressure extraction layer, and an environmental feature fusion layer. The ambient temperature feature extraction layer is used to extract the time-domain and frequency-domain features of the temperature sequence, the humidity feature extraction layer is used to extract the time-domain and frequency-domain features of the humidity sequence, the air pressure extraction layer is used to extract the time-domain and frequency-domain features of the air pressure sequence, and the environmental feature fusion layer is used to fuse the outputs of the ambient temperature feature extraction layer, the humidity feature extraction layer, and the air pressure extraction layer to generate an environmental feature vector. The environment-guided fault diagnosis unit is used to determine the probability of occurrence of various environment-guided faults based on environmental feature vectors.
4. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 1, characterized in that, The state fault diagnosis sub-model includes a third input unit, a third feature extraction unit, and a state fault diagnosis unit. The third input unit is used to input and preprocess the vibration sequence of multiple positions of the voltage transformer, the discharge activity sequence of multiple positions of the fuse, and the temperature sequence of multiple positions of the fuse. The third feature extraction unit includes a vibration feature extraction layer, a discharge activity extraction layer, a fuse temperature feature extraction layer, and a physical state feature fusion layer. The vibration feature extraction layer is used to extract and fuse the time-domain and frequency-domain features of the vibration sequence at each location. The discharge activity extraction layer is used to extract and fuse the time-domain and frequency-domain features of the discharge activity sequence at each location. The fuse temperature feature extraction layer is used to extract and fuse the time-domain and frequency-domain features of the temperature sequence at each location. The physical state feature fusion layer is used to fuse the outputs of the vibration feature extraction layer, the discharge activity extraction layer, and the fuse temperature feature extraction layer to generate a physical state feature vector. The state fault diagnosis unit is used to determine the probability of occurrence of various physical faults based on physical state feature vectors.
5. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 4, characterized in that, Vibration sequences were collected from multiple locations of the voltage transformer, including: Establish a three-dimensional model of the voltage transformer; Based on the operating environment information and three-dimensional model of the voltage transformer, a finite element model of the voltage transformer is established. Finite element analysis is performed based on the finite element model of the voltage transformer to determine the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer. Based on the electromagnetic force distribution and mechanical stress distribution inside the voltage transformer, multiple locations for vibration monitoring are determined. Vibration sequences were collected from multiple locations of the voltage transformer based on the vibration monitoring.
6. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 5, characterized in that, Discharge activity sequences were collected from multiple locations, including: Finite element analysis is performed based on the finite element model of the voltage transformer to determine the electric field intensity distribution inside the voltage transformer. Based on the electric field intensity distribution inside the voltage transformer, multiple locations for monitoring discharge activities are determined. Discharge activity sequences were collected from multiple locations where discharge activity was monitored.
7. The method for constructing a PT fuse fault diagnosis model in a high-altitude environment according to claim 5, characterized in that, Collect temperature sequences at multiple locations of the fuse, including: Finite element analysis based on the finite element model of the voltage transformer was performed to determine the temperature distribution inside the fuse, and based on the temperature distribution inside the fuse, multiple locations for temperature monitoring were determined. Temperature sequences of the fuse are collected from multiple locations where temperature monitoring is performed.