Infrared spectrum fault identification method and system for power transformation equipment in high altitude area

Abstract

The embodiment of the invention provides an infrared spectrum fault identification method and system for power transformation equipment in a high-altitude area, and the method comprises the steps: obtaining an original infrared spectrum of the power transformation equipment and environment parameters during collection; performing layered correction on the original infrared spectrum based on the environmental parameters to obtain a corrected infrared spectrum; fusing the corrected characteristics of the infrared spectrum, the high-altitude adaptive parameters and the real-time operation data of the power transformation equipment, and performing intelligent diagnosis through a diagnosis model to obtain a diagnosis result; and based on the diagnosis result, generating a diagnosis report adapted to the high-altitude operation and maintenance demand. The technical scheme provided by the invention can effectively solve the problems of infrared diagnosis distortion, misjudgment and missed judgment, inaccurate trend prediction and the like of the power transformation equipment in the high-altitude area, and ensures the safe operation of a high-altitude power system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power transformation equipment state monitoring and fault diagnosis, and particularly relates to an infrared spectrum fault identification method and system for power transformation equipment in high-altitude areas. BACKGROUND

[0002] In the operation and maintenance scene of power transformation equipment in high-altitude areas, the infrared spectrum is the core basis for judging the heating defects of equipment and evaluating the health status of equipment, and is directly related to the timeliness and accuracy of fault early warning of power transformation equipment. The high-altitude environment has special climatic characteristics such as low air pressure, strong ultraviolet rays, and large temperature difference between day and night. These characteristics cause significant interference to the collection and diagnosis of the infrared spectrum of power transformation equipment, and the traditional infrared spectrum identification and diagnosis method has obvious limitations and is difficult to adapt to the needs of high-altitude scenes.

[0003] The core defects of the traditional method are reflected in three aspects: first, the spectrum correction lacks pertinence, only a general image enhancement algorithm (such as histogram equalization) is used, without considering the infrared radiation attenuation caused by low air pressure in high-altitude areas, the image glare caused by strong ultraviolet rays, and the background temperature difference interference caused by the large temperature difference between day and night, resulting in that the corrected spectrum still has distortion and cannot accurately reflect the real heating state of the equipment; second, the data dimension of the diagnosis model is single, only depends on the infrared features (such as hotspot temperature and temperature difference value) for judgment, without fusing the high-altitude adaptation parameters specific to equipment in high-altitude areas (such as the correction value of insulation withstand voltage changing with altitude, the compensation coefficient of heat dissipation efficiency decreasing due to low air pressure), the historical data of common defects in high-altitude areas (such as leakage heating caused by aging of sealing elements, and contact failure heating caused by low-temperature brittleness of metal parts), and real-time operation data of equipment (load current, oil level, environmental temperature and humidity), which is easy to cause "misjudgment" (such as judging normal heat dissipation difference as a defect) or "omission" (such as not identifying the potential fault of slight heating under low air pressure); third, the heating threshold is not adapted to the low air pressure environment, the device heating threshold standard in plain areas is followed, and the characteristics that the heat dissipation efficiency of equipment decreases under low air pressure and the fault risk is higher under the same heating degree are ignored, resulting in deviation of defect severity evaluation and inability to accurately predict the fault trend.

[0004] In summary, the existing technology has obvious deficiencies in the spectrum correction accuracy, diagnosis model adaptability, and heating threshold rationality in high-altitude scenes, and is difficult to meet the needs of high-precision, high-reliability, and high-adaptation of infrared diagnosis of power transformation equipment in high-altitude areas. Therefore, there is an urgent need for an infrared spectrum fault identification method and system for power transformation equipment in high-altitude areas to solve the problems of infrared diagnosis distortion, misjudgment and omission, and inaccurate trend prediction of power transformation equipment in high-altitude areas, and to ensure the safe operation of high-altitude power systems. SUMMARY

[0005] The high-altitude area power transformation equipment infrared spectrum fault identification method and system provided by the embodiments of the present application can effectively solve the problems of infrared diagnosis distortion, misjudgment and missed judgment, inaccurate trend prediction and the like of the high-altitude area power transformation equipment, and guarantee the safe operation of the high-altitude power system.

[0006] In a first aspect, the embodiments of the present application provide a high-altitude area power transformation equipment infrared spectrum fault identification method, comprising:

[0007] obtaining an original infrared spectrum of the power transformation equipment and environmental parameters at the time of collection;

[0008] performing hierarchical correction on the original infrared spectrum based on the environmental parameters to obtain a corrected infrared spectrum;

[0009] fusing the features of the corrected infrared spectrum, high-altitude adaptation parameters and real-time operation data of the power transformation equipment, performing intelligent diagnosis through a diagnosis model to obtain a diagnosis result;

[0010] generating a diagnosis report adapted to the high-altitude operation and maintenance requirements based on the diagnosis result.

[0011] In an optional implementation, the hierarchical correction on the original infrared spectrum based on the environmental parameters comprises:

[0012] compensating and correcting the radiation attenuation caused by low air pressure based on an atmospheric radiation transmission model;

[0013] identifying and repairing a glare area caused by strong ultraviolet light by using an image processing technology;

[0014] eliminating a pseudo-heat area caused by background temperature difference fluctuation by comparing day and night infrared spectra.

[0015] In an optional implementation, the compensating and correcting the radiation attenuation caused by low air pressure based on the atmospheric radiation transmission model comprises: calculating an actual atmospheric transmittance based on the altitude and the real-time air pressure in the environmental parameters, and using the actual atmospheric transmittance to compensate the radiation attenuation area in the original infrared spectrum in a reverse radiation intensity.

[0016] In an optional implementation, the identifying and repairing the glare area caused by strong ultraviolet light by using the image processing technology comprises: identifying the glare pixels by using an adaptive gray threshold method, and repairing the glare pixels by using an interpolation algorithm based on the radiation values of the surrounding normal pixels.

[0017] In an optional implementation, the eliminating the pseudo-heat area caused by the background temperature difference fluctuation by comparing the day and night infrared spectra comprises: calculating the temperature difference of the same equipment position in a preset daytime period and a nighttime period, and comparing the temperature difference with a pre-stored corresponding altitude threshold to identify and mark the pseudo-heat area for correction.

[0018] In one optional implementation, the diagnostic model is a deep learning model built upon convolutional neural networks and long short-term memory neural networks; intelligent diagnosis is performed through the diagnostic model, including:

[0019] The characteristics of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operational data are standardized to form multi-source fusion data.

[0020] Input multi-source fusion data into the diagnostic model to output defect type, severity, and fault development trend;

[0021] Based on the low air pressure characteristics of the environment where the power equipment is located, the heating threshold is dynamically verified, and the diagnostic result is determined by combining the output of the diagnostic model.

[0022] In one optional implementation, the step of dynamically verifying the heating threshold includes: correcting the basic heating threshold based on the equipment heat dissipation efficiency under low air pressure, and performing a secondary correction by combining the load current in the real-time operating data to obtain the dynamic heating threshold.

[0023] In one alternative implementation, the high-altitude adaptation parameters include an insulation withstand voltage adjustment value and a heat dissipation efficiency compensation coefficient pre-stored based on altitude.

[0024] Secondly, embodiments of this application provide an infrared image fault identification system for power equipment in high-altitude areas, including:

[0025] The data acquisition module is used to acquire the original infrared spectrum of the substation equipment and the environmental parameters at the time of acquisition;

[0026] The calibration module is used to perform layered calibration on the original infrared spectrum based on environmental parameters to obtain the calibrated infrared spectrum.

[0027] The diagnostic module is used to integrate the characteristics of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operating data of the power equipment, and to perform intelligent diagnosis through the diagnostic model to obtain diagnostic results.

[0028] The output module is used to generate diagnostic reports adapted to the operational and maintenance needs of high-altitude areas based on the diagnostic results.

[0029] Thirdly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the method provided in embodiments of this application.

[0030] Fourthly, embodiments of this application provide a computer-readable storage medium, characterized in that it stores a computer program thereon, which, when executed in a computer, causes the computer to perform the method provided in embodiments of this application.

[0031] The technical solution provided in this application has the following beneficial effects:

[0032] This invention presents a high-altitude infrared spectrum layering correction mechanism based on atmospheric radiative transfer models and image processing technology. It fundamentally solves multiple environmental interferences such as low-pressure radiation attenuation, strong ultraviolet glare, and pseudo-heating due to diurnal temperature variations. By specifically modeling and compensating for the physical characteristics of different interference sources, it achieves precise purification of the original infrared spectrum, providing a reliable data foundation for subsequent diagnosis and effectively avoiding misjudgments caused by image distortion. Employing a multi-source data fusion and deep learning model diagnostic strategy, it overcomes the limitations of traditional methods that rely solely on image features. This is achieved by standardizing and fusing the corrected infrared features with high-altitude adaptation physical parameters, historical defect records, and real-time operational data, and by utilizing convolutional neural networks and long short-term memory networks for collaborative data mining. By mining spatial and temporal patterns in the data, the ability to identify complex fault modes is significantly enhanced, enabling diagnostic results to comprehensively reflect the type, severity, and development trend of defects. Based on a dynamic threshold verification and operation and maintenance requirement-adaptive output mechanism, the practical value of the technical solution is ensured. By constructing a dynamic adjustment model for equipment heating thresholds, it adapts to the heat dissipation characteristics of low-pressure environments. Combined with diagnostic results, it automatically generates highly instructive operation and maintenance decision-making suggestions, improving the decision-making efficiency and response speed of the entire operation and maintenance chain. This solution, through end-to-end innovation of "precise correction - intelligent diagnosis - practical output," forms a closed-loop solution for special high-altitude environments, achieving multiple technical effects from improving basic data quality to ultimately empowering operation and maintenance decisions. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating a method for identifying infrared spectrum faults in power equipment in high-altitude areas, as provided in an embodiment of this application.

[0034] Figure 2 This is a logical schematic diagram of an infrared spectrum fault identification method for power equipment in high-altitude areas provided in an embodiment of this application;

[0035] Figure 3 This is a schematic diagram of the structure of an infrared image fault identification system for power equipment in high-altitude areas, provided in an embodiment of this application. Detailed Implementation

[0036] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] Infrared thermal imaging diagnostic technology, as a core means of non-contact temperature measurement and defect identification for power equipment, can convert infrared radiation from the equipment surface into a visual thermal image using a thermal imager. This allows for the early detection of thermal faults such as contact overheating and insulation degradation, effectively reducing operation and maintenance costs. Currently, this technology has been explicitly included in the mandatory testing specifications for live equipment by industry standards, becoming a "core line of defense" for ensuring the safe operation of high-altitude power grids.

[0038] Current infrared diagnostic technology mainly covers three core components: first, basic radiation correction, which provides preliminary compensation for radiation attenuation in high-altitude environments by associating with altitude; second, defect identification, which relies on manual analysis of the temperature difference between hotspots and normal areas, or uses traditional machine learning algorithms (such as support vector machines and backpropagation neural networks) for judgment; and third, data fusion, which uses information such as equipment ledgers and real-time load data to assist in diagnostic decision-making.

[0039] However, existing technologies face significant bottlenecks when dealing with the unique interference of high-altitude environments: First, the correction layer only addresses the single interference factor of low air pressure, failing to solve the problems of image glare caused by strong ultraviolet radiation (which can easily lead to misidentification of hotspots) and "false heating" caused by day-night temperature differences (temperature fluctuations between the device and the background can easily lead to misjudgments); Second, the identification layer uses a fixed heating threshold standard, without considering the reduced heat dissipation efficiency of the device under low air pressure conditions, and directly using the threshold standard from plains areas can easily lead to missed fault detection; Third, the data fusion depth is insufficient, failing to incorporate the adaptation parameters unique to high-altitude equipment (such as insulation withstand voltage adjustment values ​​and heat dissipation efficiency compensation coefficients), making it difficult to accurately assess the actual risk of defects.

[0040] In summary, existing infrared diagnostic technologies cannot meet the monitoring needs of high-altitude areas for "high precision, high reliability, and high adaptability." There is an urgent need to develop a layered correction and dynamic diagnostic scheme for high-altitude environments in order to overcome the current technical bottlenecks.

[0041] To address the aforementioned issues, this application provides a method for identifying infrared spectrum faults in power equipment located at high altitudes. Figure 1 This is a flowchart illustrating a method for identifying infrared spectrum faults in power equipment in high-altitude areas, as provided in an embodiment of this application. Figure 2 This is a logical schematic diagram of an infrared spectrum fault identification method for power equipment in high-altitude areas provided in an embodiment of this application. The method can be executed by an infrared spectrum fault identification system for power equipment in high-altitude areas. The system can be implemented by software and / or hardware and can be configured in electronic devices such as computers.

[0042] like Figure 1 and Figure 2 As shown, the technical solution provided in this application includes the following steps:

[0043] S110. Obtain the original infrared spectrum of the substation equipment and the environmental parameters at the time of acquisition.

[0044] Initialize the high-altitude environmental parameter database, pre-storing atmospheric radiative transfer coefficients, ultraviolet intensity correction factors, and diurnal temperature difference compensation thresholds for different altitudes; receive raw infrared images of power equipment collected by an infrared thermal imager, and synchronously record environmental information at the time of acquisition, including altitude. Real-time air pressure (P) and ultraviolet radiation intensity Ambient temperature Collection time This provides parameters for targeted correction.

[0045] The high-altitude environmental parameter database also includes initial values ​​for atmospheric transmittance, which are pre-calculated based on an atmospheric radiative transfer model, as shown in the following formula:

[0046] ;

[0047] in: represents the pre-stored atmospheric transmittance (infrared band 3-5μm); k represents the atmospheric attenuation coefficient (experimental fitted value, k=2.5×10⁻⁶). -5 m -1 H represents altitude (unit: m).

[0048] Due to the influence of water vapor and aerosols in the atmosphere, a measured correction coefficient γ needs to be introduced. γ is a constant less than 1, obtained by fitting a large amount of field measurement data, to make the theoretical model more consistent with the actual environment. For example, it can be set to 0.88. The final pre-stored initial value of atmospheric transmittance... for:

[0049] .

[0050] S120. Perform layered correction on the original infrared spectrum based on environmental parameters to obtain the corrected infrared spectrum.

[0051] In some embodiments, pixel-level analysis is performed on the original infrared spectrum to extract three types of core interference features, and a hierarchical correction algorithm is constructed based on the atmospheric radiative transfer model, including:

[0052] Based on the atmospheric radiative transfer model, compensation and correction are performed for the radiative attenuation caused by low air pressure.

[0053] Specifically, the actual atmospheric transmittance is calculated based on the altitude and real-time air pressure in the environmental parameters, and the actual atmospheric transmittance is used to perform reverse radiation intensity compensation on the radiation attenuation region in the original infrared spectrum.

[0054] Low-pressure interference feature extraction: By analyzing the differences in infrared radiation intensity distribution in the spectrum, the region where radiation attenuation is caused by low pressure is located using the radiation intensity gradient formula, as follows:

[0055] ;

[0056] in: This represents the radiance gradient of pixel (x,y); Represents the radiance of pixel (x,y) in the original spectrum (unit: W / (m²)). 2 ·sr·μm)); , These represent the partial derivatives in the x and y directions, respectively.

[0057] Judgment rule: When ( , When the standard deviation of radiation intensity in the normal area of ​​the equipment is reached, it is determined to be a low-pressure radiation attenuation area.

[0058] By introducing the atmospheric transmittance parameter corresponding to altitude, the radiation intensity in the attenuation region is calculated inversely. The calculation formula is as follows:

[0059] Actual atmospheric transmittance calculation:

[0060] ;

[0061] Calculation of corrected radiation intensity:

[0062] ;

[0063] in, P represents the actual atmospheric transmittance, P represents the real-time air pressure, and P0 represents the standard atmospheric pressure, which is 1013.25 hPa. This represents the corrected radiance intensity of pixel (x,y) in the original spectrum.

[0064] Image processing technology is used to identify and repair glare areas caused by strong ultraviolet radiation;

[0065] Specifically, the method of adaptive grayscale thresholding is used to identify glare pixels, and the glare pixels are repaired by interpolation algorithm based on the radiation value of the surrounding normal pixels.

[0066] Strong UV interference feature extraction: Identify glare areas in the image caused by UV reflection (such as overexposed pixels on the surface of ceramic insulators). An adaptive grayscale threshold formula is used for localization, as follows:

[0067] ;

[0068] in: Indicates the glare recognition threshold; This represents the overall grayscale mean of the graph; This represents the standard deviation of the grayscale value of the image.

[0069] Judgment rule: When the pixel grayscale value At that time, it is marked as a glare pixel.

[0070] An adaptive threshold segmentation algorithm is used to remove glare pixels, and bilinear interpolation is performed using the radiation values ​​of surrounding normal pixels for restoration. The calculation formula is as follows:

[0071] ;

[0072] Where a, b, c, and d are weighting coefficients:

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] Where: (x0, y0) are glare pixels. This represents the radiation value at the glare pixel (x0, y0) after repair. (x1, y1), (x1, y2), (x2, y1), and (x2, y2) are the four normal pixels in the direct neighborhood of the glare pixel, namely the top left, bottom left, top right, and bottom right pixels, and a+b+c+d=1 to ensure smooth interpolation.

[0078] By comparing day and night infrared spectra, false heating areas caused by background temperature fluctuations are eliminated.

[0079] Specifically, the method involves calculating the temperature difference between the same device location during preset daytime and nighttime periods, comparing it with the corresponding pre-stored altitude threshold, identifying and marking false heating areas for correction.

[0080] Diurnal temperature difference interference feature extraction: By comparing the daytime and nighttime data collected by the same device (12:00-13:00 during the day and 22:00-23:00 at night), areas of "pseudo-heating" caused by background temperature differences (such as areas with abnormally fluctuating temperature differences between the device and the surrounding environment) are extracted. The temperature difference formula is used for calculation, as follows:

[0081] ;

[0082] in: Indicates the temperature difference of the same pixel between day and night (unit: °C); Indicates the original pixel temperature during the day; This indicates the temperature of the original pixel at night.

[0083] Judgment rule: When ( , When the maximum diurnal temperature range at the corresponding altitude is stored in advance, it is determined to be a "pseudo-heating" area.

[0084] A background temperature difference elimination model is constructed, using the difference between the real-time ambient temperature and the equipment surface temperature to correct for background influences. The calculation formula is as follows:

[0085] Background temperature difference compensation value:

[0086] ;

[0087] Corrected temperature:

[0088] ;

[0089] in, This represents the background temperature difference compensation value for pixel (x, y). This represents the temperature after pixel (x,y) correction. The original temperature of pixel (x,y) is represented by α, which represents the compensation coefficient obtained through experimental fitting. For example, it can be set to 0.9.

[0090] In some embodiments, the original infrared spectrum is subjected to layered correction using a layered correction algorithm, sequentially completing the layered correction for low air pressure, strong ultraviolet radiation, and diurnal temperature variation, generating a preliminary corrected spectrum; the correction effect is verified using the "feature comparison method," determined by the following three quantitative index formulas:

[0091] Radiation intensity stability (coefficient of variation) CV: ;

[0092] in: To correct the standard deviation of radiation intensity in critical parts of the equipment (such as transformer bushings and circuit breaker contacts), The corresponding mean; requirements: .

[0093] Glare reduction rate R uv : ;

[0094] in: To correct the number of pixels before glare, The number of pixels remaining after correction; requirements .

[0095] Background temperature difference interference elimination rate R bg :

[0096] in: To correct the number of "pseudo-heating" pixels before, To determine the number of remaining "pseudo-heating" pixels after correction; requirements: .

[0097] If the standard is not met, the calibration parameters are readjusted (e.g., adjusting the atmospheric transmittance coefficient and optimizing the interpolation algorithm) until the verification criteria are met. After successful verification, the calibrated infrared spectrum is output, and the core infrared features (including hotspot temperatures) in the calibrated infrared spectrum are extracted. Temperature difference between hotspots and normal areas Hotspot area (etc.), and transmit them to the diagnostic model.

[0098] The S130, combined with the characteristics of the fused and corrected infrared spectrum, high-altitude adaptation parameters, and real-time operating data of the substation equipment, are used to perform intelligent diagnosis through a diagnostic model to obtain diagnostic results.

[0099] In some embodiments, intelligent diagnosis is performed using a diagnostic model, including:

[0100] The characteristics of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operational data are standardized to form multi-source fusion data.

[0101] Among them, the high-altitude adaptation parameters include the insulation withstand voltage adjustment value and heat dissipation efficiency compensation coefficient pre-stored based on altitude.

[0102] Specifically, it includes:

[0103] Multi-source data classification and collection:

[0104] High-altitude adaptation parameters:

[0105] Insulation withstand voltage correction value For every 1000 meters increase in altitude, the insulation withstand voltage decreases by 8%-10%, as shown in the following formula:

[0106] ;

[0107] in, Rated insulation withstand voltage for plains areas, unit: kV;

[0108] Heat dissipation efficiency correction value At low air pressure, heat dissipation efficiency is directly proportional to air pressure, as shown in the following formula:

[0109] ;

[0110] in, For heat dissipation efficiency in plains;

[0111] Historical defect data: Retrieve historical defect databases of power equipment in high-altitude areas (containing 10,000+ records), and filter historical data that matches the current equipment type (such as temperature change patterns of transformer seal aging, leakage, and overheating, and poor contact of disconnecting switches, and defect development cycles). ).

[0112] Real-time operating data: Acquiring equipment load current (Unit: A) Oil level (Unit: %) Runtime (Unit: h) Last maintenance time (Unit: days ago).

[0113] Data standardization processing:

[0114] Min-max normalization transformation is applied to multi-source data to map numerical data to... The interval, the formula is as follows:

[0115] ;

[0116] in: This represents the standardized data; Represents the original data (such as ); , These represent the historical minimum and maximum values ​​of the parameter, respectively.

[0117] After standardization, the data is uniformly converted into a structured format of "device ID + timestamp + parameter name + standardized value" (e.g., "TR-001+1690000000000+HOT_TEMP+0.5"); the classified data is encoded to ensure that the data can be recognized by the deep learning model.

[0118] Input multi-source fusion data into the diagnostic model to output defect type, severity, and fault development trend;

[0119] In some embodiments, the diagnostic model is a deep learning model built on convolutional neural networks and long short-term memory neural networks (CNN-LSTM).

[0120] The model structure specifically includes:

[0121] Input layer: Input standardized multi-source data (dimension 1) (Contains 3 infrared features, 2 plateau features, 2 historical features, and 5 operational features), reshape as Adapted to CNN input format.

[0122] CNN feature extraction layer (extracts spatially correlated features):

[0123] Convolutional layer 1: (32 convolution kernels, size) , For convolution operations, Let W1 represent the output of convolutional layer 1, W1 represent the weight tensor of convolutional layer 1, and b1 represent the bias term of convolutional layer 1. (where x is the activation function), x is the input data, and z is the linear output after the convolution operation.

[0124] Pooling layer 1 (max pooling): (Pooling kernel size) (downsampling dimensionality reduction) This represents the output of pooling layer 1; j is the starting index of the pooling window.

[0125] Convolutional layer 2: (64 convolution kernels, size) ), b1 represents the output of convolutional layer 2, W2 represents the weight tensor of convolutional layer 2, and b2 represents the bias term of convolutional layer 2.

[0126] The output feature map has a dimension of .

[0127] LSTM time series analysis layer (mining time series correlation features):

[0128] Using a 2-layer LSTM, the core gating formula is as follows (taking a 1-layer LSTM as an example):

[0129] Forgotten Gate: ; This indicates the output of the forget gate, which decides to discard historical information. For the sigmoid function, x t This represents the input at the current moment;

[0130] Input Gate: ; This represents the input gate output, used to update the cell state. Indicates the state of candidate cells;

[0131] Memory cells: ; This indicates the current state of memory cells. Element-wise multiplication;

[0132] Output gate: ; This indicates that the output gate is outputting, showing the current hidden state. Indicates the current hidden state;

[0133] W f Wi W c W o These are the weight matrices for the forget gate, input gate, memory cell, and output gate, respectively. f b i b c b o These are the bias terms for the forget gate, input gate, memory cell, and output gate, respectively.

[0134] The output dimension of the second-layer LSTM is .

[0135] Output layer (3 branches):

[0136] Defect type branches: ;

[0137] The output results for the defect type branch include three categories: "Seal Aging / Poor Contact / Insulation Damage". The weight matrix represents the defect type branch. The bias term representing the defect type branch. This represents the hidden state output by the LSTM layer;

[0138] Severity branches: ;

[0139] The output results for the severity branch include three levels: "Mild / Moderate / Severe". The weight matrix representing the severity branch. Bias terms indicating severity branches;

[0140] Fault trend branch: ;

[0141] The output of the fault trend branch includes three categories: "No deterioration within 1 week / Deterioration within 1-2 weeks / Deterioration is inevitable within 2 weeks". The weight matrix represents the fault trend branch. The bias term represents the fault trend branch.

[0142] Furthermore, training was conducted using historical diagnostic data of power equipment in high-altitude areas (10,000+ sets of "multi-source data - actual defect results" samples):

[0143] Dataset partitioning: 70% training set, 20% validation set, 10% test set;

[0144] Loss function: The cross-entropy loss function Loss is used, and the formula is as follows:

[0145] ;

[0146] Where: N is the total number of samples, i represents the i-th sample, K is the total number of categories, and k represents the k-th category. For real labels, To predict probabilities;

[0147] Optimizer: Adam optimizer, learning rate 0.001. , ;

[0148] Training termination criteria: 50 iterations, batch size 32, validation set loss value decreases by less than 0.001 for 5 consecutive iterations, and test set accuracy is greater than 92%;

[0149] After training is complete, save the model weights (.h5 format) to the model library.

[0150] Based on the low air pressure characteristics of the environment where the power equipment is located, the heating threshold is dynamically verified, and the diagnostic result is determined by combining the output of the diagnostic model.

[0151] The steps for dynamically verifying the heating threshold include: correcting the basic heating threshold based on the equipment's heat dissipation efficiency under low air pressure, and performing a secondary correction by combining the load current in the real-time operating data to obtain the dynamic heating threshold.

[0152] The calculation process is as follows:

[0153] The basic high-altitude threshold is calculated using the following formula, which incorporates a correction value for heat dissipation efficiency under low air pressure:

[0154] ;

[0155] in: The heating threshold for equipment in plain areas;

[0156] Secondary load current correction: Adjust the threshold based on the real-time load current of the equipment, using the following formula:

[0157] ;

[0158] in: Indicates the final dynamic heating threshold (unit: °C); This represents the load impact factor; for every 10% overload exceeding the rated load, the threshold is lowered by 1%. This indicates the rated current of the equipment (unit: A).

[0159] The process of dynamically verifying the heating threshold based on the low-pressure characteristics of the environment where the power equipment is located, and determining the diagnostic result by combining the output of the diagnostic model, includes:

[0160] Threshold comparison: Corrected hotspot temperatures With dynamic heating threshold In comparison, if If it is determined to be "defect-free", the result is output directly; if Enter model reasoning.

[0161] Model inference: The standardized multi-source data is input into the trained CNN-LSTM model, which outputs preliminary diagnostic results.

[0162] Rule validation: The preliminary diagnosis results are validated based on the high-altitude defect rule base. If the validation passes, the final diagnosis result is determined and transmitted to the output module; if the validation fails, the data is retrieved again to check for anomalies and the diagnosis is re-executed.

[0163] S140. Based on the diagnostic results, generate a diagnostic report adapted to the operation and maintenance needs of high-altitude areas.

[0164] In some embodiments, the output module first receives and parses the diagnostic results, then performs result adaptation processing according to the high-altitude operation and maintenance requirements, and finally generates multiple forms of output content and provides feedback, providing operation and maintenance personnel with intuitive and usable diagnostic basis.

[0165] 1) Receiving and interpreting diagnostic results

[0166] Receive the final diagnostic results transmitted by the multi-source fusion intelligent diagnostic module, and analyze the core information in the results:

[0167] Basic equipment information: Equipment ID, Equipment type, Installation location altitude ;

[0168] Defect information: defect type, severity, hotspot location coordinates, current hotspot temperature Dynamic heating threshold ;

[0169] Trend information: Fault development trend and recommended maintenance time window;

[0170] The above information is organized into a structured results dataset.

[0171] 2) Adaptation and handling of high-altitude operation and maintenance requirements

[0172] Due to the unique characteristics of power operation and maintenance at high altitudes, the diagnostic results were adapted accordingly:

[0173] Defect severity is correlated with operational priority: Priority is calculated using an operational priority determination formula, as follows:

[0174] ;

[0175] in:

[0176] Severity level (1 = mild, 2 = moderate, 3 = severe);

[0177] Fault trend (1 = no deterioration within 1 week, 2 = deterioration within 1-2 weeks, 3 = deterioration within 2 weeks).

[0178] Severity weighting , ;

[0179] : Round up function (ensure priority is 1=low, 2=medium, 3=high);

[0180] It also notes the special requirements for high-altitude operation and maintenance.

[0181] Matching Fault Trends with Maintenance Cycles: The maintenance window is calculated using a suggested maintenance time formula, as follows:

[0182] ;

[0183] in:

[0184] The critical time for fault deterioration is as follows: Trend 1 corresponds to 7 days, Trend 2 corresponds to 10 days, and Trend 3 corresponds to 14 days.

[0185] The diagnosis was made after the time had passed. ;

[0186] Explanation of the correlation between defect causes and high altitude factors: Add "High Altitude Impact Analysis" to the results to help maintenance personnel understand the root cause of the defect.

[0187] 3) Generation and output feedback of multiple results

[0188] Based on the operational and maintenance scenario requirements, generate output results. After the output is completed, receive feedback information from operations and maintenance personnel and establish a feedback loop:

[0189] If the feedback is "the result is accurate", the process ends and the diagnostic results are archived to the historical database.

[0190] If the feedback is "required for review", the high-altitude infrared spectrum correction module and the multi-source fusion intelligent diagnosis module will be re-executed to generate a review report.

[0191] If a "data anomaly" is reported, investigate the data collection process, fix it, and then re-execute the entire process.

[0192] Finally, the confirmation results and feedback records are updated in the device profile and added to the model training dataset.

[0193] Figure 3This is a schematic diagram of the structure of an infrared image fault identification system for power equipment in high-altitude areas, provided in an embodiment of this application. Figure 3 As shown, the system includes:

[0194] The data acquisition module is used to acquire the original infrared spectrum of the substation equipment and the environmental parameters at the time of acquisition;

[0195] The calibration module is used to perform layered calibration on the original infrared spectrum based on environmental parameters to obtain the calibrated infrared spectrum.

[0196] The diagnostic module is used to integrate the characteristics of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operating data of the power equipment, and to perform intelligent diagnosis through the diagnostic model to obtain diagnostic results.

[0197] The output module is used to generate diagnostic reports adapted to the operational and maintenance needs of high-altitude areas based on the diagnostic results.

[0198] The execution process of the system part of this application embodiment is the same as that of the method part of the embodiment described above, and will not be repeated here.

[0199] This application also provides an electronic device, including: at least one processor; a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute any of the new energy-based power grid multi-resource coordinated control methods.

[0200] This application also proposes a computer storage medium storing a computer program, which, when executed by a processor, implements any one of the new energy-based power grid multi-resource coordinated control methods.

[0201] Computer storage media may be simply referred to as media. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Dual Data SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), Rambus Direct RAM (RDRAM), Direct Memory Bus Dynamic RAM (DRDRAM), and Memory Bus Dynamic RAM (RDRAM). The various embodiments described in this specification are presented in a progressive manner, with reference allowed to each other for similar or identical parts. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for apparatuses, devices, and non-volatile computer storage media are described simply because they are substantially similar to the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0202] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this application.

Claims

1. A method for infrared spectrum fault identification of power equipment in high-altitude areas, characterized in that, include: Acquire the original infrared spectrum of the substation equipment and the environmental parameters at the time of acquisition; Based on the environmental parameters, the original infrared spectrum is subjected to layered correction to obtain the corrected infrared spectrum; By integrating the features of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operating data of the substation, intelligent diagnosis is performed through a diagnostic model to obtain diagnostic results. Based on the diagnostic results, a diagnostic report adapted to the operation and maintenance needs at high altitudes is generated.

2. The method according to claim 1, characterized in that, Based on the environmental parameters, the original infrared spectrum is subjected to layered correction, including: Based on the atmospheric radiative transfer model, compensation and correction are performed for the radiative attenuation caused by low air pressure. Image processing technology is used to identify and repair glare areas caused by strong ultraviolet radiation; By comparing day and night infrared spectra, false heating areas caused by background temperature fluctuations are eliminated.

3. The method according to claim 2, characterized in that, Based on the atmospheric radiative transfer model, the radiation attenuation caused by low air pressure is compensated and corrected, including: calculating the actual atmospheric transmittance based on the altitude and real-time air pressure in the environmental parameters, and using the actual atmospheric transmittance to perform reverse radiation intensity compensation on the radiation attenuation region in the original infrared spectrum.

4. The method according to claim 2, characterized in that, Image processing technology is used to identify and repair glare areas caused by strong ultraviolet radiation, including: identifying glare pixels using an adaptive grayscale thresholding method, and repairing the glare pixels based on the radiation values ​​of surrounding normal pixels using an interpolation algorithm.

5. The method according to claim 2, characterized in that, By comparing day and night infrared images, false heating areas caused by background temperature fluctuations are eliminated. This includes: calculating the temperature difference between the same device location during preset daytime and nighttime periods, comparing it with the corresponding pre-stored altitude threshold, identifying and marking false heating areas for correction.

6. The method according to claim 1, characterized in that, The diagnostic model is a deep learning model built on convolutional neural networks and long short-term memory neural networks; the intelligent diagnosis through the diagnostic model includes: The features of the corrected infrared spectrum, the high-altitude adaptation parameters, and the real-time operating data are standardized to form multi-source fusion data. The multi-source fusion data is input into the diagnostic model, which outputs the defect type, severity, and fault development trend. Based on the low air pressure characteristics of the environment where the power equipment is located, the heating threshold is dynamically verified, and the diagnostic result is determined by combining the output of the diagnostic model.

7. The method according to claim 6, characterized in that, The steps for dynamically verifying the heating threshold include: correcting the basic heating threshold based on the equipment's heat dissipation efficiency under low air pressure, and performing a secondary correction by combining the load current in the real-time operating data to obtain the dynamic heating threshold.

8. The method according to claim 1, characterized in that, The high-altitude adaptation parameters include the insulation withstand voltage adjustment value and heat dissipation efficiency compensation coefficient pre-stored based on altitude.

9. A fault identification system for power equipment in high-altitude areas using infrared imagery, characterized in that, include: The data acquisition module is used to acquire the original infrared spectrum of the substation equipment and the environmental parameters at the time of acquisition; The correction module is used to perform layered correction on the original infrared spectrum based on the environmental parameters to obtain the corrected infrared spectrum. The diagnostic module is used to integrate the features of the corrected infrared spectrum, high-altitude adaptation parameters, and real-time operating data of the substation equipment, and to perform intelligent diagnosis through a diagnostic model to obtain diagnostic results. The output module is used to generate a diagnostic report adapted to the operation and maintenance needs at high altitudes based on the diagnostic results.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed in a computer, causes the computer to perform the method of any one of claims 1-8.

Images