A power equipment fault detection method and system based on high-precision temperature measurement

Abstract

This invention relates to the field of fault equipment detection technology, and discloses a method and system for power equipment fault detection based on high-precision temperature measurement. The method includes scanning the surface temperature of the power equipment using a high-precision temperature measuring device to obtain a temperature distribution image; processing the temperature distribution image to obtain a temperature distribution matrix; performing gradient analysis and temperature difference identification on the temperature distribution matrix to obtain candidate hotspot regions; performing cluster analysis on the candidate hotspot regions to obtain initial hotspot regions; using a boundary tracking algorithm to generate static boundaries; correcting the static boundaries based on temperature changes at each boundary point to obtain the hotspot regions; and performing a fault risk assessment based on the load data of the power equipment and the temperature data of the hotspot regions. This invention, through temperature data analysis and region boundary correction, can accurately identify hotspot regions, and combined with machine learning algorithms, improves the accuracy of power equipment fault risk assessment, ensuring the safe and stable operation of the power system.

Description

Technical Field

[0001] This invention relates to the field of equipment fault detection technology, and in particular to a method and system for detecting faults in power equipment based on high-precision temperature measurement. Background Technology

[0002] The safe operation of power equipment is essential for energy supply and industrial production. Abnormal surface temperatures on power equipment are often an early warning sign of equipment failure. Current temperature monitoring of power equipment mainly relies on traditional sensing technologies or simple image analysis. The limitations of traditional monitoring methods are twofold: firstly, insufficient monitoring accuracy, making it difficult to capture dynamic changes in minute temperature difference areas. Especially under fluctuating equipment loads and complex environments, the lack of precision in identifying subtle differences in temperature distribution can lead to misjudgments or missed alarms. Secondly, due to differences in the materials used in power equipment, variations in thermal conductivity can cause blurred boundaries when hot spots form due to localized temperature rises. Under the dynamic coupling effect of load changes and material properties, existing methods lack the ability to reliably identify the boundaries of minute temperature difference areas, thus limiting the accuracy of fault warnings. Therefore, how to construct a method that can accurately identify the boundaries of minute temperature difference areas and improve the accuracy of hot spot warnings under the dynamic influence of equipment load changes and differences in surface material properties has become a critical problem that urgently needs to be solved. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a method and system for detecting power equipment faults based on high-precision temperature measurement. By accurately identifying hot spot areas, it improves the accuracy of power equipment fault risk assessment and ensures the safe and stable operation of the power system.

[0004] In a first aspect, the present invention provides a method for fault detection of power equipment based on high-precision temperature measurement, the method comprising:

[0005] The surface temperature of the power equipment is scanned by a high-precision temperature measuring device to obtain a temperature distribution image. The temperature distribution image is then processed to obtain a temperature distribution matrix.

[0006] Gradient analysis and temperature difference identification are performed on the temperature distribution matrix to obtain hot spot candidate regions;

[0007] Cluster analysis is performed on the candidate hot spot regions to obtain initial hot spot regions, and a boundary tracking algorithm is used to generate the static boundary of the initial hot spot regions.

[0008] Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region;

[0009] Based on the load data of the power equipment and the temperature data of the hot spot area, a fault risk assessment is conducted on the power equipment to obtain the fault risk assessment results.

[0010] Further, the step of processing the temperature distribution image to obtain the temperature distribution matrix includes:

[0011] The temperature distribution image is converted into an original temperature matrix, and the original temperature matrix is ​​filtered and denoised to obtain the temperature distribution matrix.

[0012] Furthermore, the step of performing gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions includes:

[0013] Gradient analysis is performed on the temperature distribution matrix to obtain the gradient magnitude of each pixel.

[0014] Based on the comparison between the gradient magnitude and the magnitude threshold, several temperature abrupt change regions are obtained;

[0015] Calculate the maximum temperature difference for each temperature abrupt change region, and select the temperature abrupt change region where the maximum temperature difference is greater than the temperature difference threshold as a hot spot candidate region.

[0016] Further, the step of performing cluster analysis on the candidate hotspot regions to obtain initial hotspot regions, and using a boundary tracking algorithm to generate the static boundary of the initial hotspot regions includes:

[0017] Based on the center point coordinates and the first average temperature value of the hot spot candidate region, a density-based clustering algorithm is used to merge the regions to obtain the initial hot spot region.

[0018] Based on the second average temperature value of the initial hot spot region and the preset temperature difference, the boundary determination temperature value is determined, and based on the boundary determination temperature value, the boundary point set is obtained.

[0019] Based on the set of boundary points, a boundary tracking algorithm is used to generate the static boundary of the initial hot spot region.

[0020] Furthermore, the step of correcting the static boundary based on the temperature changes at each boundary point on the static boundary to obtain the hot spot region includes:

[0021] Obtain the temperature sequence of each boundary point on the static boundary, and determine the temperature change of each boundary point based on the temperature sequence, wherein the temperature change includes temperature rise and temperature fall;

[0022] In response to a temperature rise, the boundary point is extended outward by a predetermined distance along the normal direction;

[0023] In response to a decrease in temperature, the boundary point is contracted inward by a predetermined distance along the normal direction.

[0024] Furthermore, the step of correcting the static boundary based on the temperature changes at each boundary point on the static boundary to obtain the hot spot region includes:

[0025] Obtain the temperature sequence of each boundary point on the static boundary, and determine the temperature change of each boundary point based on the temperature sequence, wherein the temperature change includes temperature rise and temperature fall;

[0026] In response to a temperature rise, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is extended outward by a preset distance along the temperature gradient direction; otherwise, the boundary point is extended outward by a preset distance along the normal direction.

[0027] In response to a temperature decrease, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is shrunk inward by a preset distance along the temperature gradient direction; otherwise, the boundary point is shrunk inward by a preset distance along the normal direction.

[0028] Furthermore, the step of conducting a fault risk assessment of the power equipment based on the load data of the power equipment and the temperature data of the hot spot area, and obtaining the fault risk assessment result, includes:

[0029] Obtain the current sequence of the power equipment within a preset time period, as well as the regional average temperature sequence and duration of sustained high temperature in the hot spot area;

[0030] Calculate the current-temperature correlation coefficient based on the current sequence and the regional average temperature sequence;

[0031] The real-time average temperature value and temperature change amplitude of the hot spot area are obtained, and the real-time average temperature value, the temperature change amplitude, the duration of continuous high temperature and the current-temperature correlation coefficient are used as feature vectors and input into a preset fault risk assessment model to obtain the fault risk assessment result. The fault risk assessment model is constructed based on the support vector machine model.

[0032] Furthermore, the current-temperature correlation coefficient is expressed by the following formula:

[0033]

[0034] In the formula, r represents the current-temperature correlation coefficient, i represents the time point, n represents the total number of time points, and T i This represents the regional average temperature at the i-th time point. I represents the mean of the temperature series. i This represents the current value at the i-th time point. This represents the mean of the current sequence.

[0035] Furthermore, after the step of obtaining the failure risk assessment result, the method further includes:

[0036] Based on the fault risk assessment results, early warning information for power equipment is generated, including hot spot location, current time, current temperature, and fault risk level.

[0037] The real-time environmental parameters and real-time operating parameters of the power equipment are obtained, and the real-time environmental parameters and real-time operating parameters are timestamped and merged with the early warning information to obtain the merged early warning information. The real-time environmental parameters include real-time ambient temperature and real-time ambient humidity, and the real-time operating parameters include real-time current, real-time voltage and real-time equipment load rate.

[0038] The processing priority of power equipment is determined based on the fault risk level and the real-time equipment load rate.

[0039] Secondly, the present invention provides a power equipment fault detection system based on high-precision temperature measurement, the system comprising:

[0040] The data processing module is used to scan the surface temperature of the power equipment using a high-precision temperature measuring device to obtain a temperature distribution image, and to process the temperature distribution image to obtain a temperature distribution matrix.

[0041] The hot spot identification module is used to perform gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions;

[0042] The region correction module is used to perform cluster analysis on the hot spot candidate regions to obtain the initial hot spot regions, and to use a boundary tracking algorithm to generate the static boundary of the initial hot spot regions.

[0043] Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region;

[0044] The risk assessment module is used to assess the fault risk of the power equipment based on the load data of the power equipment and the temperature data of the hot spot area, and obtain the fault risk assessment result.

[0045] This invention provides a method and system for fault detection in power equipment based on high-precision temperature measurement. By denoising and identifying temperature differences in high-precision temperature measurement data, this invention can identify hot spot regions. Through dynamic correction of region boundaries, the accuracy of hot spot region identification can be improved. Furthermore, through machine learning algorithms, the accuracy of fault risk assessment results can be enhanced, providing accurate data support for power equipment operation and maintenance decisions, thereby ensuring the safety and stability of power system operation. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating the power equipment fault detection method based on high-precision temperature measurement in an embodiment of the present invention.

[0047] Figure 2 This is a schematic diagram of the structure of the power equipment fault detection system based on high-precision temperature measurement in an embodiment of the present invention;

[0048] Figure label:

[0049] 10. Data processing module; 20. Hot spot identification module; 30. Area correction module; 40. Risk assessment module. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1 The first embodiment of the present invention proposes a method for fault detection of power equipment based on high-precision temperature measurement, including steps S10 to S50:

[0052] Step S10: Scan the surface temperature of the power equipment using a high-precision temperature measuring device to obtain a temperature distribution image; process the temperature distribution image to obtain a temperature distribution matrix.

[0053] Step S20: Perform gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions;

[0054] Step S30: Perform cluster analysis on the hot spot candidate regions to obtain initial hot spot regions, and use a boundary tracking algorithm to generate the static boundary of the initial hot spot regions;

[0055] Step S40: Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region;

[0056] Step S50: Based on the load data of the power equipment and the temperature data of the hot spot area, a fault risk assessment is performed on the power equipment to obtain the fault risk assessment result.

[0057] This invention enables fault detection and risk assessment of power equipment by accurately identifying hot spot areas on the surface and combining this information with current equipment operating parameters. First, a high-precision temperature measuring device scans the surface temperature of the power equipment to obtain a temperature distribution image. This image is then processed to generate a temperature distribution matrix. Specifically, an infrared thermal imager scans the surface of the power equipment to capture thermal radiation energy, generating a temperature distribution image. The pixels in the image are converted into temperature values ​​to construct the original temperature matrix, which is then denoised using a filtering algorithm.

[0058] This embodiment employs multiple filtering algorithms for joint denoising. First, median filtering is applied to the original temperature matrix. The median of the temperature values ​​within a 3×3 neighborhood of each pixel is calculated, and the temperature value of the center pixel is replaced with the median, resulting in a preliminary denoised temperature matrix. Median filtering effectively eliminates impulse noise, such as abnormally high-temperature noise points caused by transient sensor interference. Next, the temperature difference between adjacent pixels is calculated. This temperature difference is equal to the temperature difference between two adjacent pixels divided by the pixel spacing. If the difference exceeds a preset threshold, bilateral filtering is performed to obtain a secondary denoised temperature matrix. Bilateral filtering removes residual noise while preserving temperature distribution characteristics. Based on the secondary denoised temperature matrix, a set of pixels with temperatures exceeding the upper limit of the device's normal operating temperature is identified, forming a high-temperature anomaly region. Temperature data outside the high-temperature anomaly region remains unchanged, while temperature data within the high-temperature anomaly region is smoothed using Gaussian filtering to obtain a temperature distribution matrix. Gaussian filtering eliminates local temperature abrupt changes, such as abnormal peaks caused by measurement errors, making the temperature distribution more consistent with actual heat conduction patterns.

[0059] This embodiment uses the above-mentioned multi-level noise reduction processing to ensure that the final temperature distribution matrix can accurately reflect the actual temperature status of the equipment, thereby providing reliable data support for subsequent fault diagnosis and early warning, and effectively improving the operational safety of power equipment.

[0060] For the temperature distribution matrix, this embodiment uses gradient analysis and temperature difference identification to initially identify hot spot areas on the surface of power equipment. The specific steps include:

[0061] Gradient analysis is performed on the temperature distribution matrix to obtain the gradient magnitude of each pixel.

[0062] Based on the comparison between the gradient magnitude and the magnitude threshold, several temperature abrupt change regions are obtained;

[0063] Calculate the maximum temperature difference for each temperature abrupt change region, and select the temperature abrupt change region where the maximum temperature difference is greater than the temperature difference threshold as a hot spot candidate region.

[0064] In this embodiment, for the temperature distribution matrix, an edge detection operator, such as the Sobel or Canny operator, is used to calculate the temperature gradient components of each pixel in the horizontal and vertical directions. The horizontal gradient component is equal to the temperature of the pixel to the right minus the temperature of the pixel to the left, and the vertical gradient component is equal to the temperature of the pixel below minus the temperature of the pixel above. The square root of the sum of the squares of the two gradient components is used to calculate the gradient magnitude of each pixel. The gradient magnitude reflects the rate of temperature change between adjacent pixels, and then temperature abrupt change regions are identified based on the gradient magnitude. Specifically, it is determined whether the gradient magnitude of each pixel is greater than a preset magnitude threshold. If the gradient magnitude of a pixel is greater than the threshold, it indicates that there is a significant temperature abrupt change at that point, i.e., there is localized abnormal heating on the device surface. This pixel is identified as an abnormal pixel, and the region formed by adjacent abnormal pixels is identified as a temperature abrupt change region. This is because localized abnormal heating on the device surface exists in the form of regions; therefore, by dividing the region based on abnormal pixels, several temperature abrupt change regions can be obtained. A temperature abrupt change region is essentially a region with a temperature jump relative to its surroundings, meaning the overall temperature is higher than the background temperature. A hot spot, on the other hand, is formed by localized abnormal heating on the equipment surface. Besides the temperature jump, a hot spot also exhibits significant heat accumulation within its surface, meaning there is a large temperature difference between the high-temperature point and its neighbors. A temperature abrupt change region only satisfies the first characteristic; therefore, it's necessary to identify the temperature difference within the region to filter out areas that meet the second characteristic. Specifically, for a temperature abrupt change region, the highest and lowest temperatures in that region are found using a temperature distribution matrix. Subtracting these two values ​​yields the maximum temperature difference. This maximum temperature difference is then compared to a temperature difference threshold. If the threshold is exceeded, it indicates significant heat accumulation within the region, thus satisfying the second characteristic. This temperature abrupt change region is then considered a candidate hot spot region.

[0065] Since hot spot candidate regions that are physically close and have similar temperature characteristics may be caused by heat diffusion from the same internal fault point, after obtaining multiple hot spot candidate regions, it is necessary to cluster and merge the hot spot candidate regions to generate an initial hot spot region. Then, a boundary tracking algorithm is used to obtain the static boundary of the initial hot spot region. The specific steps include:

[0066] Based on the center point coordinates and the first average temperature value of the hot spot candidate region, a density-based clustering algorithm is used to merge the regions to obtain the initial hot spot region.

[0067] Based on the second average temperature value of the initial hot spot region and the preset temperature difference, the boundary determination temperature value is determined, and based on the boundary determination temperature value, the boundary point set is obtained.

[0068] Based on the set of boundary points, a boundary tracking algorithm is used to generate the static boundary of the initial hot spot region.

[0069] In this embodiment, a density-based clustering algorithm, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), is used to merge candidate hot spot regions. The input data for DBSCAN is the coordinates of the center point and the average temperature value of each candidate hot spot region. A preset spatial distance threshold is used as the neighborhood radius, a minimum number of cluster points is used as the minimum number of points, and an average temperature difference threshold is used as a constraint. The DBSCAN parameters are initialized, and then the algorithm is executed. This involves traversing the center points of all candidate hot spot regions, calculating the spatial Euclidean distance and average temperature difference between any two points, and determining neighborhoods based on the spatial distance and average temperature difference thresholds. If both are less than the threshold, the two points are considered neighbors. Starting from any point whose neighbor count satisfies the minimum number of points, all density-connected points are recursively merged. Finally, the merged output yields an initial hot spot region.

[0070] For the initial hot spot region, this embodiment employs a boundary tracking algorithm to generate the static boundary of the initial hot spot region. First, the difference between the average temperature value of the initial hot spot region and a preset temperature difference is used as the boundary judgment temperature value. The preset temperature difference value can be determined by analyzing and statistically analyzing the temperature data of historical hot spot regions. Pixels that meet the boundary judgment temperature value are usually distributed around the hot spot, forming an approximately elliptical temperature contour line. To accurately obtain the region boundary of the initial hot spot region, this embodiment uses a boundary tracking algorithm to perform boundary tracking and contour generation on these pixels that meet the boundary judgment temperature value. Specifically, all pixels that meet the boundary judgment temperature value are grouped into a boundary point set. For each boundary point in the set, starting from any point, the search proceeds clockwise, searching its 8-neighborhood to find the next unvisited boundary point. The two points are connected, and the new point is used as the current point to continue the search until the starting point is returned to form a closed contour, resulting in a closed polygon vertex coordinate sequence, which is the static boundary of the initial hot spot region.

[0071] Because heat diffusion is dynamic and time-varying, static boundaries alone cannot reflect the real-time range of thermal impact. To improve the accuracy of hotspot region identification and thus enhance the precision of subsequent fault detection results, this embodiment, after obtaining the static boundary, further corrects the boundary based on the temperature changes of each edge node on the static boundary, thereby obtaining an accurate hotspot region. The specific correction steps include:

[0072] Acquire the temperature time series data of each boundary point on the static boundary, and determine the temperature change of each boundary point based on the temperature time series data, wherein the temperature change includes temperature rise and temperature fall;

[0073] In response to a temperature rise, the boundary point is extended outward by a predetermined distance along the normal direction;

[0074] In response to a decrease in temperature, the boundary point is contracted inward by a predetermined distance along the normal direction.

[0075] In this embodiment, temperature time-series data of each boundary point on the static boundary is acquired. The boundary point is the boundary point of the static boundary output by the boundary tracking algorithm. The temperature time-series data includes at least 3 consecutive frames of temperature data. For each boundary point, the temperature difference between two adjacent frames in the temperature time-series data is calculated. It is determined whether the absolute value of the temperature difference is greater than a threshold and whether it is a change in the same direction. If so, the boundary point is marked as a boundary point to be corrected. A change in the same direction refers to a continuous temperature rise or a continuous temperature fall. That is, if the continuous temperature difference is positive and the absolute value of the temperature difference is greater than the threshold, it indicates a continuous temperature rise. If the continuous temperature difference is negative and the absolute value of the temperature difference is greater than the threshold, it indicates a continuous temperature fall.

[0076] Then, based on temperature changes, the correction direction for the boundary points to be corrected is determined. When the temperature rises, the boundary points are expanded outwards by a preset distance along the normal direction (the normal direction of the boundary point is perpendicular to the geometric direction of the boundary contour), for example, by one pixel. When the temperature falls, the boundary points are similarly contracted inwards by a preset distance along the normal direction, for example, by one pixel. This correction method is iteratively applied until an iteration stopping condition is met. The iteration stopping condition includes reaching a preset number of iterations, such as 5, or when more than a preset percentage of boundary points on the static boundary no longer move, for example, when more than 80% of the boundary points no longer need correction. By dynamically correcting the static boundary, the accuracy of boundary positioning can be effectively improved.

[0077] In addition to the dynamic and time-varying nature of heat expansion, different materials also have varying thermal conductivity. Therefore, in a preferred embodiment, the present invention also provides another method for boundary correction, the specific steps of which include:

[0078] Acquire the temperature time series data of each boundary point on the static boundary, and determine the temperature change of each boundary point based on the temperature time series data, wherein the temperature change includes temperature rise and temperature fall;

[0079] In response to a temperature rise, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is extended outward by a preset distance along the temperature gradient direction; otherwise, the boundary point is extended outward by a preset distance along the normal direction.

[0080] In response to a temperature decrease, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is shrunk inward by a preset distance along the temperature gradient direction; otherwise, the boundary point is shrunk inward by a preset distance along the normal direction.

[0081] In this embodiment, the steps for determining the temperature change trend at the boundary points are the same as those in the previous embodiment, and will not be repeated here. The main difference between this embodiment and the previous embodiment lies in the correction direction of the boundary points. In this embodiment, the thermal conduction direction of heat in the device material is determined by analyzing the temperature gradient at the boundary points. The thermal conduction direction is the main direction in which heat spreads from the center of the hot spot to the surrounding area. The propagation direction also differs for different materials. For example, in the copper material region, heat will spread rapidly along the axial direction of the copper conductor, while in the insulating paper material region, heat spreads more slowly and the diffusion path is unclear.

[0082] Based on the differences in thermal conduction of materials, this embodiment determines the heat diffusion path by using the temperature gradient at the boundary points. First, the horizontal and vertical temperature gradients at the boundary points are calculated, following the same steps as those used for gradient analysis of the temperature distribution matrix. Then, based on the comparison between the horizontal and vertical temperature gradients and a preset gradient threshold, it is determined whether a significant temperature gradient exists at the boundary points, i.e., whether the absolute value of either the horizontal or vertical temperature gradient is greater than the gradient threshold. If either gradient is greater than the threshold, it is considered that the material has a path of least resistance for heat diffusion, and this path is consistent with the direction of the temperature gradient. Assuming the horizontal temperature gradient is Gx and the vertical temperature gradient is Gy, the direction of the temperature gradient is θ = arctan(Gx / Gy). If there is no significant temperature gradient at the boundary points, i.e., no gradient is greater than the gradient threshold, it indicates that the material is homogeneous. In this case, the normal direction of the boundary points is taken as the direction of heat diffusion.

[0083] During boundary correction, the first step is to determine if the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is corrected according to the temperature gradient direction. In this case, if the temperature rises, the boundary point expands outward by a preset distance along the temperature gradient direction; if the temperature falls, it contracts inward by a preset distance along the temperature gradient direction. If the temperature gradient is not greater than the gradient threshold, the boundary point is corrected by expanding or contracting inward along the normal direction. The boundary point is then iteratively corrected according to the above steps until a preset iteration stopping condition is met, thus obtaining the accurate hotspot region boundary.

[0084] The boundary correction method provided in this embodiment takes into account the heat diffusion characteristics of different materials. By determining the heat diffusion path and correcting the boundary along the diffusion path, it further improves the accuracy of hot spot region boundary identification under the dynamic coupling effect of load change and material properties.

[0085] After obtaining the accurate hot spot area, this embodiment analyzes the temperature data of the hot spot area and the load data of the power equipment to perform fault detection and risk assessment on the power equipment. The specific steps include:

[0086] Acquire the current time-series data of the power equipment within a preset time period, as well as the regional average temperature time-series data and duration of continuous high temperature in the hot spot area;

[0087] Calculate the current-temperature correlation coefficient based on the current time series data and the regional average temperature time series data;

[0088] The real-time average temperature value and temperature change amplitude of the hot spot area are obtained, and the real-time average temperature value, the temperature change amplitude, the duration of continuous high temperature and the current-temperature correlation coefficient are used as feature vectors and input into a preset fault risk assessment model to obtain the fault risk assessment result. The fault risk assessment model is constructed based on the support vector machine model.

[0089] In this embodiment, a fault risk assessment model is used to assess the fault risk of power equipment. First, the features of the input fault risk assessment model are extracted. The features of the input model include temperature features and load correlation features. The temperature features include the real-time average temperature value of the hot spot area, the temperature change amplitude, and the duration of continuous high temperature. The load correlation feature is the current-temperature correlation coefficient.

[0090] Specifically, the real-time average temperature value is the arithmetic mean of the temperatures of all pixels within the current hotspot area; the temperature change amplitude is the difference between the highest and lowest temperatures within the current hotspot area; the duration of sustained high temperature is the cumulative time during which the highest temperature exceeds the safety threshold within a preset time period, such as the past 24 hours; and the current-temperature correlation coefficient is calculated based on the current time-series data of the power equipment and the regional average temperature time-series data of the hotspot area within the preset time period. Again, using the past 24 hours as the preset time period and hours as the step size, the regional average temperature time-series data includes the regional average temperature at each time point, and the current time-series data includes the current value of the power equipment at each time point. The current-temperature correlation coefficient is calculated using the Pearson correlation coefficient formula, and its expression is:

[0091]

[0092] In the formula, r represents the current-temperature correlation coefficient, i represents the time point, n represents the total number of time points, and T i This represents the regional average temperature at the i-th time point. I represents the time-series average temperature. i This represents the current value at the i-th time point. This represents the time-series average of the current.

[0093] In this embodiment, the temperature feature is used to quantify the current thermal state, and the load-related feature is used to characterize the dynamic correlation between the load and the temperature, such as whether the temperature rise is caused by a sudden increase in current. This reflects the dynamic coupling relationship between the load and the temperature. Using the temperature feature and the load-related feature together as a feature vector can meet the accuracy and real-time requirements of power equipment fault risk assessment.

[0094] In this embodiment, a support vector machine model is preferably used to construct a fault risk assessment model. The fault risk assessment model maps the input feature vector to a high-dimensional space through a kernel function, calculates the functional distance from the feature point to the decision boundary in the space, and converts the functional distance into a fault probability value. The fault probability value ranges from 0 to 1, thereby obtaining the fault probability value of the hot spot area. Based on the fault probability of the hot spot area, the fault risk assessment result is obtained, that is, the fault risk level of the power equipment, such as high risk level, medium risk level, and low risk level.

[0095] In a preferred embodiment, after obtaining the fault risk assessment results, the present invention correlates the fault risk level of the power equipment with the equipment operating status data to determine the processing priority of the power equipment. Specific steps include:

[0096] Based on the fault risk assessment results, early warning information for power equipment is generated, including hot spot location, current time, current temperature, and fault risk level.

[0097] The real-time environmental parameters and real-time operating parameters of the power equipment are obtained, and the real-time environmental parameters and real-time operating parameters are timestamped and merged with the early warning information to obtain the merged early warning information. The real-time environmental parameters include real-time ambient temperature and real-time ambient humidity, and the real-time operating parameters include real-time current, real-time voltage and real-time equipment load rate.

[0098] The processing priority of power equipment is determined based on the fault risk level and the real-time equipment load rate.

[0099] In this embodiment, firstly, based on the fault risk assessment results, early warning information for the power equipment is generated, including the hot spot location, current time, current temperature, and fault risk level. Then, real-time environmental parameters of the power equipment are acquired, including real-time ambient temperature and real-time ambient humidity, and these real-time environmental parameters are appended to the early warning information to form early warning data containing environmental context. Simultaneously, real-time operating parameters of the power equipment are acquired, including real-time current, real-time voltage, and real-time equipment load rate. Since early warning data and operating status data usually come from different systems and may have different sampling frequencies, in this embodiment, the early warning data, environmental data, and operating status data are aligned according to timestamps. The introduction of environmental data and operating status data makes the early warning information more comprehensive and accurate.

[0100] In addition to enriching the content of the early warning information, this embodiment further determines the processing priority of power equipment based on the fault risk level and the real-time equipment load rate. For example, when the fault risk level is high and the real-time equipment load rate is greater than the load rate threshold, the processing priority of the power equipment is marked as high priority. When the fault risk level is high and the real-time equipment load rate is greater than the load rate threshold, the processing priority of the power equipment is marked as medium priority. Finally, equipment that does not meet either condition is marked as low priority. Of course, this is only a preferred method, and the specific priority setting can be flexibly set according to the actual situation. No further restrictions are imposed here.

[0101] When there are multiple power devices with the same priority, the processing order of the power devices can be sorted according to real-time environmental parameters. In fact, the risk level of the same equipment temperature will be different under different environmental parameters. For example, high ambient temperature means poor heat dissipation of the equipment, and high ambient humidity can accelerate the aging of equipment materials and affect the speed of fault development. Therefore, by further adjusting the processing order of power devices through environmental parameters, it can be ensured that emergency faults are responded to in a timely manner.

[0102] This invention provides a power equipment fault detection method based on high-precision temperature measurement. By performing multi-level filtering and noise processing on temperature data, the method effectively eliminates the influence of environmental noise. Through gradient analysis and dynamic correction of regional boundaries, it can adapt to equipment load fluctuations and differences in equipment material properties, accurately identify hot spot areas, and quantify fault risks through machine learning algorithms. This achieves intelligent early warning of power equipment fault risks and ensures the safe and stable operation of the power system.

[0103] Please see Figure 2 Based on the same inventive concept, the second embodiment of this invention proposes a power equipment fault detection system based on high-precision temperature measurement, comprising:

[0104] Data processing module 10 is used to scan the surface temperature of power equipment using a high-precision temperature measuring device to obtain a temperature distribution image, and to process the temperature distribution image to obtain a temperature distribution matrix.

[0105] Hot spot identification module 20 is used to perform gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions;

[0106] The region correction module 30 is used to perform cluster analysis on the hot spot candidate regions to obtain the initial hot spot regions, and to use a boundary tracking algorithm to generate the static boundary of the initial hot spot regions.

[0107] Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region;

[0108] The risk assessment module 40 is used to assess the fault risk of the power equipment based on the load data of the power equipment and the temperature data of the hot spot area, and obtain the fault risk assessment result.

[0109] The technical features and effects of the power equipment fault detection system based on high-precision temperature measurement proposed in this invention are the same as those of the method proposed in this invention, and will not be repeated here. Each module in the above-mentioned power equipment fault detection system based on high-precision temperature measurement can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or it can be stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0110] In summary, the present invention proposes a method and system for fault detection of power equipment based on high-precision temperature measurement. The method scans the surface temperature of the power equipment using a high-precision temperature measuring device to obtain a temperature distribution image. This image is then processed to obtain a temperature distribution matrix. Gradient analysis and temperature difference identification are performed on the temperature distribution matrix to obtain candidate hotspot regions. Cluster analysis is then performed on these candidate hotspot regions to obtain initial hotspot regions. A boundary tracking algorithm is used to generate static boundaries for these initial hotspot regions. Based on the temperature changes at each boundary point on the static boundaries, the static boundaries are corrected to obtain the hotspot regions. Finally, based on the load data of the power equipment and the temperature data of the hotspot regions, a fault risk assessment is performed on the power equipment to obtain the fault risk assessment result. This invention, through denoising and temperature difference identification of the high-precision temperature measurement data, can identify hotspot regions. Dynamic correction of the region boundaries improves the accuracy of hotspot region identification. Machine learning algorithms enhance the accuracy of the fault risk assessment results, providing accurate data support for power equipment operation and maintenance decisions, thereby ensuring the safety and stability of the power system.

[0111] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0112] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for fault detection of power equipment based on high-precision temperature measurement, characterized in that, include: The surface temperature of the power equipment is scanned by a high-precision temperature measuring device to obtain a temperature distribution image. The temperature distribution image is then processed to obtain a temperature distribution matrix. Gradient analysis and temperature difference identification are performed on the temperature distribution matrix to obtain hot spot candidate regions; Cluster analysis is performed on the candidate hot spot regions to obtain initial hot spot regions, and a boundary tracking algorithm is used to generate the static boundary of the initial hot spot regions. Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region; Based on the load data of the power equipment and the temperature data of the hot spot area, a fault risk assessment is conducted on the power equipment to obtain the fault risk assessment results. The step of correcting the static boundary based on the temperature changes at each boundary point on the static boundary to obtain the hot spot region includes: Acquire the temperature time series data of each boundary point on the static boundary, and determine the temperature change of each boundary point based on the temperature time series data, wherein the temperature change includes temperature rise and temperature fall; In response to a temperature rise, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is extended outward by a preset distance along the temperature gradient direction; otherwise, the boundary point is extended outward by a preset distance along the normal direction. In response to a temperature decrease, it is determined whether the temperature gradient at the boundary point is greater than a gradient threshold. If so, the boundary point is shrunk inward by a preset distance along the temperature gradient direction; otherwise, the boundary point is shrunk inward by a preset distance along the normal direction.

2. The power equipment fault detection method based on high-precision temperature measurement according to claim 1, characterized in that, The step of processing the temperature distribution image to obtain the temperature distribution matrix includes: The temperature distribution image is converted into an original temperature matrix, and the original temperature matrix is ​​filtered and denoised to obtain the temperature distribution matrix.

3. The power equipment fault detection method based on high-precision temperature measurement according to claim 1, characterized in that, The step of performing gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions includes: Gradient analysis is performed on the temperature distribution matrix to obtain the gradient magnitude of each pixel. Based on the comparison between the gradient magnitude and the magnitude threshold, several temperature abrupt change regions are obtained; Calculate the maximum temperature difference for each temperature abrupt change region, and select the temperature abrupt change region where the maximum temperature difference is greater than the temperature difference threshold as a hot spot candidate region.

4. The power equipment fault detection method based on high-precision temperature measurement according to claim 1, characterized in that, The steps of performing cluster analysis on the candidate hot spot regions to obtain initial hot spot regions, and using a boundary tracking algorithm to generate the static boundary of the initial hot spot regions include: Based on the center point coordinates and the first average temperature value of the hot spot candidate region, a density-based clustering algorithm is used to merge the regions to obtain the initial hot spot region. Based on the second average temperature value of the initial hot spot region and the preset temperature difference, the boundary determination temperature value is determined, and based on the boundary determination temperature value, the boundary point set is obtained. Based on the set of boundary points, a boundary tracking algorithm is used to generate the static boundary of the initial hot spot region.

5. The power equipment fault detection method based on high-precision temperature measurement according to claim 1, characterized in that, The step of conducting a fault risk assessment of the power equipment based on the load data of the power equipment and the temperature data of the hot spot area, and obtaining the fault risk assessment result, includes: Acquire the current time-series data of the power equipment within a preset time period, as well as the regional average temperature time-series data and duration of continuous high temperature in the hot spot area; Calculate the current-temperature correlation coefficient based on the current time series data and the regional average temperature time series data; The real-time average temperature value and temperature change amplitude of the hot spot area are obtained, and the real-time average temperature value, the temperature change amplitude, the duration of continuous high temperature and the current-temperature correlation coefficient are used as feature vectors and input into a preset fault risk assessment model to obtain the fault risk assessment result. The fault risk assessment model is constructed based on the support vector machine model.

6. The power equipment fault detection method based on high-precision temperature measurement according to claim 5, characterized in that, The current-temperature correlation coefficient is expressed by the following formula: In the formula, r represents the current-temperature correlation coefficient, i represents the time point, n represents the total number of time points, and T i This represents the regional average temperature at the i-th time point. I represents the time-series average temperature. i This represents the current value at the i-th time point. This represents the time-series average of the current.

7. The power equipment fault detection method based on high-precision temperature measurement according to claim 1, characterized in that, Following the step of obtaining the failure risk assessment results, the following is also included: Based on the fault risk assessment results, early warning information for power equipment is generated, including hot spot location, current time, current temperature, and fault risk level. The real-time environmental parameters and real-time operating parameters of the power equipment are obtained, and the real-time environmental parameters and real-time operating parameters are timestamped and merged with the early warning information to obtain the merged early warning information. The real-time environmental parameters include real-time ambient temperature and real-time ambient humidity, and the real-time operating parameters include real-time current, real-time voltage and real-time equipment load rate. The processing priority of power equipment is determined based on the fault risk level and the real-time equipment load rate.

8. A power equipment fault detection system based on high-precision temperature measurement, wherein the system is applied to the method described in any one of claims 1 to 7, characterized in that, include: The data processing module is used to scan the surface temperature of the power equipment using a high-precision temperature measuring device to obtain a temperature distribution image, and to process the temperature distribution image to obtain a temperature distribution matrix. The hot spot identification module is used to perform gradient analysis and temperature difference identification on the temperature distribution matrix to obtain hot spot candidate regions; The region correction module is used to perform cluster analysis on the hot spot candidate regions to obtain the initial hot spot regions, and to use a boundary tracking algorithm to generate the static boundary of the initial hot spot regions. Based on the temperature changes at each boundary point on the static boundary, the static boundary is corrected to obtain the hot spot region; The risk assessment module is used to assess the fault risk of the power equipment based on the load data of the power equipment and the temperature data of the hot spot area, and obtain the fault risk assessment result.

Images