A comprehensive monitoring platform for intelligent substations

Abstract

The present invention relates to the field of data processing technology, and in particular to an intelligent substation integrated monitoring platform. The method comprises: collecting thermal imaging images of a monitoring area of ​​electrical equipment of the substation, and performing regional growth according to a self-defined growth criterion and a fusion criterion. The growth criterion is to accurately divide the growth areas to which pixels belong by combining pixel similarity and spatial distance, and the fusion criterion is to further fuse overlapping growth areas based on the similarity between the two during the regional growth process, which not only accelerates the segmentation speed, but also improves the accuracy and completeness of the segmentation results, avoids over-segmentation and under-segmentation, and obtains a final accurate regional growth result. According to the regional growth result, the local high-temperature area of ​​the monitoring area can be accurately identified, and the potential equipment overheating faults and high-temperature hidden dangers in the monitoring area of ​​the substation can be discovered in time, thereby realizing timely, effective and accurate monitoring of the electrical equipment of the substation.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an integrated monitoring platform for an intelligent substation. Background Art

[0002] In the daily operation of smart substations, electrical equipment plays a vital role. Due to the influence of various factors, such as poor contact, excessive load, equipment aging and other potential faults, these electrical equipment often experience abnormal high temperatures during operation. Therefore, the smart substation integrated monitoring platform is used to monitor electrical equipment in real time to eliminate safety hazards in a timely manner and ensure the safety of electrical equipment. However, it is worth noting that some equipment problems are often highly hidden in the early stages and are difficult to detect in time with the naked eye. Therefore, thermal imaging detection technology is usually used to collect thermal images of the monitored objects, and then regional growth is performed on the thermal images. Based on the results of regional growth, it is determined whether there is a high temperature anomaly to determine whether the monitored object has a safety hazard. For example, the patent document with authorization announcement number CN114018982B discloses a method for visual monitoring of air preheater dust accumulation. The method mainly involves installing an infrared detection imaging device, collecting air preheater operating status images, and preprocessing them to obtain characteristic abnormality areas; establishing a two-dimensional coordinate system XOZ plane; and dividing the air preheater operating status image into multiple concentric arcs based on the proportional relationship between the fixed position reference window of the infrared detection imaging device and the fan-shaped area. The coordinates of the characteristic abnormality areas are determined based on the proportional relationship between the infrared image and the actual operation of the air preheater. This patent document analyzes the acquired infrared images through median filtering preprocessing and an improved region growing algorithm to obtain the real-time dust accumulation status of the air preheater, thereby improving the accuracy of air preheater dust accumulation monitoring.

[0003] When the above patent uses the region growing algorithm to analyze infrared images, it first divides the image into regions, calculates the mean pixel value of each region, and uses the central pixel point of the region with the largest mean pixel value as the seed point. The growth criterion is to compare the grayscale difference value with the preset difference threshold to perform region growth. This method of setting seed points and growth criteria is insensitive to changes in image details. For example, when the boundary of the high-temperature area in the thermal image is blurred or the temperature distribution is uneven, the high-temperature area cannot be completely segmented due to improper seed point selection, or some areas with abnormally high temperatures are mistakenly judged as normal due to insufficient sensitivity. This defect reduces the accuracy of the region growing results and cannot accurately identify local high-temperature areas in the image, which in turn affects the accurate judgment of the safety status of electrical equipment and brings potential risks to the stable operation of the power system. Summary of the Invention

[0004] To address the problem of region growing in thermal images of the monitoring area of ​​electrical equipment, traditional region growing algorithms, whose seed point selection and growth criteria are not sensitive enough to changes in image details, affect the accuracy of region growing results, fail to accurately identify local high-temperature areas in thermal images, and hinder accurate judgment of the safety status of electrical equipment, this paper proposes a smart substation integrated monitoring platform. Specifically, the following technical solutions are adopted:

[0005] In one aspect, the present invention provides a smart substation integrated monitoring platform, comprising:

[0006] Collect thermal images of the monitoring area of ​​the electrical equipment in the substation, select seed points in the thermal images, perform region growing based on the seed points according to a user-defined growth criterion and a user-defined fusion criterion, determine whether there is a local high-temperature area in the thermal images based on the results of the region growing, and determine whether an early warning is needed based on the determination result;

[0007] The customized growth criterion includes: for any grown region, if the similarity of pixels within a search range of a certain edge pixel point is greater than a preset similarity threshold, the pixels within the search range are classified as the grown region; otherwise, the pixels within the search range whose distance to the seed point of the grown region is less than a preset distance threshold are classified as the grown region;

[0008] When the similarity of the pixels within the search range of all edge pixels of all grown areas is less than a preset first similarity threshold, the growth is stopped;

[0009] The customized fusion criteria include: if a certain grown area contains pixel points of other grown areas, and the similarity between the grown area and the other grown areas is greater than a preset second similarity threshold, the grown area and the other grown areas are fused into one grown area.

[0010] The above scheme performs region growing in accordance with customized growth and fusion criteria in the thermal imaging map of the monitoring area, and the growth criterion is to accurately divide the growth area to which the pixel points belong by combining pixel similarity and spatial distance, and the fusion criterion is to further fuse the overlapping growth areas based on the similarity of the two during the region growing process. This not only speeds up the segmentation speed, but also improves the accuracy and completeness of the segmentation results, avoids over-segmentation and under-segmentation, and obtains the final accurate region growing results. According to the region growing results, the local high-temperature areas in the monitoring area can be accurately identified, and the temperature characteristics of the thermal imaging map are effectively utilized to ensure the intelligent selection of seed points and the accurate execution of region growing, so as to timely discover the high-temperature anomalies and fault hazards of the electrical equipment of the smart substation, and realize timely, effective and accurate monitoring of the electrical equipment.

[0011] Preferably, the self-defined growth criteria further include:

[0012] If there is no grown area, for any seed point, if the similarity of the pixels within the search range of the seed point is greater than a preset similarity threshold, the pixels within the search range are classified as grown areas; otherwise, the pixels within the search range whose distance to the seed point is less than a preset distance threshold are classified as grown areas.

[0013] The above scheme ensures a stable and reliable starting point for region growth during the region growth process.

[0014] Preferably, the similarity between the grown region and other grown regions satisfies the following relationship:

[0015] , For the Grown areas and The similarity of the grown regions, For the Grown areas and The variance consideration value of all pixels in the union of the grown regions, For the The variance consideration value of all pixels in the grown area, For the The variance consideration value of all pixels in the grown area, For the The mean similarity of pixels within the search range of all seed points and all edge pixels in the grown region, The method of determining is: if The search range of a certain edge pixel point in the grown area contains pixel points in the grown area, mark the edge pixel point, and calculate the mean similarity of the pixels in the search range of all marked edge pixels as .

[0016] The above scheme realizes the precise quantification of the similarity between grown regions. It not only considers the consistency of pixel distribution within the region (through the variance consideration value), but also effectively utilizes the statistical information (intersection) of the overlapping parts between regions, thereby providing a more accurate similarity judgment basis in the fusion criterion. This not only enhances the rationality of regional fusion, reduces unnecessary regional division, but also improves the accuracy and efficiency of high-temperature area identification in the monitoring area of ​​electrical equipment.

[0017] Preferably, the similarity of pixels within the search range of the seed point is calculated according to the following formula:

[0018]

[0019] Where, It is The similarity of pixels within the search range of the seed points, 、 、 is the variance of the red channel pixel value, the green channel pixel value, and the blue channel pixel value of all pixels in the search range. For the The sum of the gradient values ​​of all pixels within the search range of the seed point, All are custom parameters. For the The importance of the seed points, is a natural constant.

[0020] The above scheme provides a more comprehensive and refined similarity measurement method by comprehensively considering the color channel variance (reflecting color consistency) and the sum of gradient values ​​(representing edge information). This method not only improves the sensitivity to image details during the region growing process and can effectively identify changes in image details, but also enhances the ability to accurately track the boundaries of local high-temperature areas in thermal images.

[0021] Preferably, the similarity of the pixels within the search range of the edge pixels satisfies the following relationship:

[0022]

[0023] Where, It is The similarity of pixels within the search range of edge pixels, 、 、 is the variance of the red channel pixel value, the green channel pixel value, and the blue channel pixel value of all pixels in the search range. For the The sum of the gradient values ​​of all pixels within the search range of edge pixels, All are custom parameters. is a natural constant.

[0024] Preferably, the method for selecting seed points in the thermal image includes: randomly selecting a number of pixel points as initial seed points, and calculating the importance of each initial seed point:

[0025]

[0026] Where, For the The importance of the initial seed point, 、 、 Respectively The pixel value of the red channel, the pixel value of the green channel, and the pixel value of the blue channel of the seed point, To obtain the maximum value function, Function is used for normalization;

[0027] The seed points whose importance is less than the preset importance threshold are discarded, and the seed points whose importance is greater than or equal to the preset importance threshold are retained as seed points for subsequent region growing.

[0028] The above scheme effectively evaluates the importance of each candidate seed point by combining the color intensity of the pixel points and their relative differences between channels. Based on the importance, it not only eliminates unrepresentative pixels and retains seed points with a high correlation with potential high-temperature areas, but also improves the targetedness and accuracy of seed point selection through the importance threshold screening mechanism.

[0029] Preferably, the method for judging whether there is a local high temperature area in the thermal imaging image according to the result of region growing is:

[0030] In the thermal image, calculate the average value of the red channel pixel values ​​of all pixels in each grown area;

[0031] If the average value of the red channel pixel values ​​of all pixels in a grown area is greater than a preset pixel value, and the maximum value of the importance of all seed points corresponding to the grown area is greater than a preset importance threshold, the grown area is a local high temperature area existing in the thermal imaging image.

[0032] The above scheme efficiently identifies local high-temperature areas by performing double screening by combining the average pixel value of the red channel in the grown area with the maximum value of the seed point importance. This method not only ensures the authenticity and relevance of the identified high-temperature areas, but also avoids misjudgments caused by local brightness differences, improves the accuracy of overheating monitoring of electrical equipment, and enables timely and effective equipment maintenance and safety management.

[0033] Preferably, the method for obtaining the variance consideration value is: calculating the sum of squares of the variances of pixel values ​​of the three channels of red, green and blue, and taking the arithmetic square root of the sum of squares as the variance consideration value.

[0034] Preferably, the search range of the edge pixel point and the search range of the seed point are both centered on themselves. pixel range, where is the default value.

[0035] Preferably, the method for determining whether an early warning is needed based on the judgment result is:

[0036] Continuous acquisition The thermal image of the minute, calculate the ratio of the pixel point of the local high temperature area of ​​each thermal image to all the pixel points of the thermal image. If the ratio is minutes, an early warning will be issued; if the ratio If the number of increases does not increase sequentially within minutes, no warning will be issued. is the default value.

[0037] The present invention has the following effects:

[0038] The present invention provides a monitoring method for accurately identifying whether there are local high-temperature safety hazards in the electrical equipment of the smart substation platform in the monitoring process. It mainly designs a customized region growing and fusion strategy for the thermal imaging image of the monitoring area, combines the similarity of the pixel points within the search range of the pixel points and the growth criteria and fusion criteria of the spatial information, and obtains accurate region growing results. According to the accurate region growing results, it can accurately identify whether there are high-temperature areas in the electrical equipment of the monitoring area, realize timely and effective monitoring of the electrical equipment, effectively prevent potential faults, and ensure production safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0040] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0042] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0043] Reference Figure 1 , a smart substation integrated monitoring platform, including steps S1 to S3, specifically as follows:

[0044] S1: Collect thermal images of the monitored area.

[0045] The monitoring area of ​​the intelligent substation integrated monitoring platform is mainly used to monitor the electrical equipment in operation. This step uses high-sensitivity thermal imaging technology to obtain thermal images of the monitoring area when the electrical equipment is in operation. The thermal image cleverly uses the color spectrum to map small changes in temperature into visually distinct color levels. The high-temperature area appears in red-orange tones, while the low-temperature area is marked with blue-purple tones. The color gradient delicately depicts the dynamic distribution of heat. The thermal image of the monitoring area can reflect the temperature differences in the monitoring area.

[0046] Furthermore, in view of the gradient changes caused by the uneven distribution of color areas in the thermal image, Sobel edge detection is performed on the thermal image. This algorithm efficiently estimates the gradient direction and gradient value of each pixel in the thermal image by performing differential operations on the surrounding pixels of each pixel in the image, thereby enhancing the visualization effect of the outline of the temperature transition area and the direction of heat flow in the image. This processing not only enhances the resolution of the thermal image, but also provides a more accurate information basis for the rapid identification of overheated areas.

[0047] S2: Selecting seed points in the thermal image, and performing region growing based on the seed points according to a user-defined growth criterion and fusion criterion.

[0048] In this step, the seed point of the region growing algorithm is determined by the color characteristics of the thermal image during the region growing process of the thermal image of the monitored area. The customized growth criteria and fusion criteria are further set in combination with the image details to perform region growing to obtain accurate region growing results.

[0049] The method for selecting seed points in the thermal image includes:

[0050] S21: randomly selecting K pixels in the thermal image as initial seed points. In this embodiment, K=3000.

[0051] S22: Analyze the pixel values ​​of each color channel of the seed point:

[0052] If the red channel pixel value of a seed point is higher, it means that the seed point is more likely to be in a high temperature area. If the red channel pixel value of a seed point is lower, it means that the seed point is more likely to be in a low temperature area. However, the overall temperature of electrical equipment is relatively high during operation, and the high temperature area needs to be further divided. The red channel pixel values ​​of the pixels in the high temperature area of ​​the thermal image are more prominent than the red channel pixel values ​​of the pixels in other areas.

[0053] Therefore, the importance of the seed point is reflected by the red channel pixel value of the seed point and the prominence of the red channel pixel value. Specifically, the importance of the seed point is calculated by the following formula:

[0054]

[0055] Where, For the The importance of the seed points, 、 、 Respectively The pixel value of the red channel, the pixel value of the green channel, and the pixel value of the blue channel of the seed point, To obtain the maximum value function, Function is used for normalization;

[0056] In this formula:

[0057]

[0058] Indicates the The prominence of the red channel pixel value of each seed point is used to further divide the possible areas where the seed point is located. The higher the prominence of the red channel pixel value of a seed point, the more likely the seed point is to be located in the red area of ​​the high temperature area, and the higher the importance of the seed point. The lower the prominence of the red channel pixel value of a seed point, the more likely the seed point is to be located in other areas outside the red area, and the lower the importance of the seed point.

[0059] S23: Screening seed points according to importance;

[0060] The preset importance threshold is 0.8, seed points with importance less than the preset importance threshold are discarded, and seed points with importance greater than or equal to the preset importance threshold are retained as seed points for subsequent region growing.

[0061] Among them, regional growth is performed based on the seed points according to custom growth and fusion criteria. Specifically, since the seed points are randomly selected, although they have been screened, the screened seed points may be distributed in different regions. For high-temperature regions, the seed points may be distributed at the edge of the high-temperature region or near the center or other positions. Therefore, for seed points at different positions, dynamic regional growth criteria need to be adopted to make the regional growth results more accurate.

[0062] For customized growth criteria, the specific contents include:

[0063] First, for any grown area, each edge pixel point of the grown area is analyzed, and the search range of each edge pixel point is set, that is, the edge pixel point itself is the center. Pixel range, N is a preset value. In order to analyze the image features in a refined manner, this embodiment selects ;

[0064] Then, calculate the similarity of the pixels in the search range corresponding to each edge pixel:

[0065]

[0066] In the formula, It is The similarity of pixels within the search range of edge pixels, Represents the variance of the red channel pixel values ​​of all pixels within the search range of the edge pixel point. Represents the variance of the green channel pixel values ​​of all pixels within the search range of the edge pixel point. Indicates the variance of the blue channel pixel values ​​of all pixels within the search range of the edge pixel point. The smaller the variance of the red, green, and blue channels of all pixels within the search range of a certain edge pixel point, the more likely the pixels within the search range are to belong to the same region during the region growing process. The larger the variance, the more likely the pixels within the search range are to belong to different regions during the region growing process.

[0067] In the formula, For the The sum of the gradient values ​​of all pixels within the search range of the edge pixel points. If the gradient value of the pixel points in the search range is smaller, it indicates that the pixel points in the search range are more likely to be in a certain area or inside certain areas. If the gradient value of the pixel points in the search range is larger, it indicates that the pixel points in the range are more likely to be at the edge of a certain area or at the edge of certain areas.

[0068] In the formula, It is a custom parameter used to adjust the proportion of the variance of the pixel values ​​of each channel. In this example, The reason for this setting is that in the thermal imaging map of the monitored area, the red channel (or analogous temperature range) is more sensitive to identifying key features (such as overheated areas), followed by green, and blue (or low temperature) is relatively less important. Such weight distribution can emphasize the capture of key information, and implementers can adjust it according to actual conditions.

[0069] In this formula, Indicates the color consistency of the pixels within the search range. The smaller the value, the closer the colors of the pixels within the search range are. The larger the value, the greater the color difference of the pixels within the search range. Indicates the The gradient size of the pixel points in the search range of the edge pixel points is as follows: the larger the gradient of the pixel points in the search range, the smaller the value, indicating that the search range is more likely to be at the edge of a high temperature area, and the more inconsistent the pixels in the search range are; conversely, the smaller the gradient of the pixel points in the search range, the larger the value, indicating that the search range is more likely to be inside a high temperature area, and the more consistent the pixels corresponding to the search range are. In short, the formula The item dynamically adjusts the similarity of pixels through gradient information, making pixels at the edge of the region (high gradient) more difficult to merge, while pixels inside the region (low gradient) are more likely to be regarded as similar and merged. This dynamic adjustment helps avoid over-segmentation and under-segmentation, and improves the accuracy and adaptability of region division.

[0070] Finally, the threshold comparison determines the pixel's belonging area:

[0071] In this embodiment, the first similarity threshold is set to 0.85 (empirical value). If the similarity between the pixels within the search range corresponding to a certain edge pixel of the grown region is greater than 0.85, indicating that the colors of the pixels within the search range are highly consistent, all pixels within the search range corresponding to the edge pixel are assigned to the grown region, effectively expanding the grown region. This operation assigns pixels with similar colors and positions to the same grown region.

[0072] If the similarity of the pixels within the search range corresponding to the edge pixel is less than or equal to 0.85, the distance between all pixels within the search range and the seed point of the grown region is calculated, and pixels with distances less than the preset distance threshold are assigned to the grown region. This operation provides an additional basis for division and attribution. Even in the case of slightly different colors, closely adjacent pixels in space have the opportunity to be correctly merged into the same region, enhancing the coherence and integrity of the region. This is particularly important for areas in thermal images where there may be local temperature gradients with gentle transitions. It can more accurately capture these transition areas, making the growth criteria more reasonable and avoiding segmentation errors.

[0073] The distance here refers to the color space distance, which is calculated similarly to the Euclidean distance. Take the first pixel in the search range as an example:

[0074]

[0075] In the formula, The first The distance between the pixel point and the seed point, the pixel values ​​of the three channels of the first pixel point are ， ， , the pixel values ​​of the three channels of the seed point are ， ， .

[0076] In addition, the termination condition of the customized growth criterion is: when the similarity of the pixels in the search range of all edge pixels in all grown areas is less than the preset first similarity threshold, the growth is stopped. The first similarity is used to represent the similarity of the pixels in the search range.

[0077] In addition, when there is no growth area, that is, for each seed point, if it has not yet generated a grown area, the seed point has just been determined and is about to be used for regional growth. The search range of the seed point is also set first, and then the similarity of the pixels within the search range of the seed point is calculated. If the similarity of the pixels within the search range is greater than the first similarity threshold, all the pixels within the search range are attributed to the grown area of ​​the seed point. If the similarity of the pixels within the search range is less than or equal to the first similarity threshold, the distance between all the pixels within the search range and the seed point is calculated, and the pixels whose distance is less than the preset distance threshold are attributed to the grown area. This process is basically the same as the analysis process of each edge pixel point in the grown area.

[0078] Here, the similarity of pixels within the search range of the seed point is calculated according to the following formula:

[0079]

[0080] Where, It is The similarity of pixels within the search range of the seed points, 、 、 is the variance of the red channel pixel value, the green channel pixel value, and the blue channel pixel value of all pixels in the search range. No. The sum of the gradient values ​​of all pixels within the search range of the seed point, All are custom parameters. For the The importance of the seed points, is a natural constant.

[0081] For the thermal image seed points, and there is an indicator to measure their importance .if The larger the The larger the value is, the The seed points are more important and are likely to be in the red area. In order to better segment the complete red area, at this time, the formula right It plays the role of upward correction. In this way, the similarity of pixels within the search range of this seed point will become greater, and the region growing algorithm will be easier to segment the red high-temperature area. The smaller it is, the The seed point is not important. The smaller it is, the smaller it is. In this case, The seed points are more likely to be in areas other than the red high-temperature area. Will It plays a downward correction role, making the similarity of pixels within the search range of this seed point smaller. This operation makes it less likely that pixels in the red high-temperature area will be mistakenly segmented into other areas.

[0082] In summary, the customized growth criteria in this step are an improvement on the existing region growing algorithm. The existing region growing algorithm determines whether to include a pixel point in a certain region based on the grayscale value difference between the pixels. Its ability to segment pixels with similar grayscale values ​​is very limited, and the segmentation efficiency is not high. This step determines whether the pixels in the search range can be included in the grown region by the similarity of the seed point and the edge pixels of the grown region. The efficiency of segmentation is improved by taking the search range as a unit. At the same time, the similarity of the pixels in the search range is calculated by the pixel values ​​and gradient values ​​of the three channels of the pixels in the search range, which helps to identify the boundaries of different color regions and improves the efficiency and quality of segmentation.

[0083] The customized fusion criteria include:

[0084] First, during the region growing process, if the search range of the edge pixels of a certain grown region contains pixels of other grown regions, it means that the grown region and the other grown regions are too close to each other and have overlapping parts. In this case, the similarity between the grown region and the other grown regions is calculated.

[0085]

[0086] Where, For the Grown areas and The similarity of the grown regions, For the Grown areas and The variance consideration value of all pixels in the union of the grown regions, For the The variance consideration value of all pixels in the grown area, For the The variance consideration value of all pixels in the grown area is obtained by calculating the sum of squares of the variances of the pixel values ​​of the three channels red, green, and blue, and taking the arithmetic square root of the sum of squares as the variance consideration value, for example , For the Grown areas and The variance of the red channel pixel values ​​of all pixels in the union of the grown regions is For the Grown areas and The variance of the green channel pixel values ​​of all pixels in the union of the grown regions is For the Grown areas and The variance of the blue channel pixel values ​​of all pixels in the union of the grown regions is and The calculation method of is similar.

[0087] In this formula, is the hyperbolic tangent function, The bigger the Grown areas and The greater the color difference between the two grown areas, the less likely they are to merge. Conversely, the smaller the value, the greater the possibility of fusion. Grown areas and The smaller the color difference between the two grown areas, the higher the possibility that the two can be merged.

[0088] In this formula, For the The mean similarity of pixels in the search range of all seed points and all edge pixels in the grown area, let Grown areas include seed points and edge pixels, first calculate the similarity of pixels within the search range of each seed point, and then calculate the similarity of pixels within the search range of each edge pixel. Seed points correspond to search scopes, and Similarity, Edge pixels correspond to search scopes, and Similarity, finally, calculate this The mean of similarities is .

[0089] In this formula, The method of determining is: if The search range of a certain edge pixel point in the grown area contains pixel points in the grown area, mark the edge pixel point, and calculate the mean similarity of the pixels in the search range of all marked edge pixels as .

[0090] During the region growing process, there may be overlaps in the grown regions. In this case, it is necessary to consider whether to merge these regions. The value of plays a key role in determining whether the region is fused. The mean similarity of all seed points and edge pixels in the growing area within the search range Between these two grown areas ( Grown areas and The mean value of the similarity of the edge pixels in the search range of the intersection of the grown regions) The larger the gap, the greater the The pixel points in the grown area and ( Grown areas and The more similar the features of the pixels at the intersection of the grown areas are, the Grown areas and On the contrary, if the gap is smaller, it means that the first grown area is more suitable for fusion. The pixel points in the grown area and ( Grown areas and The more dissimilar the features of the pixels at the intersection of the grown regions, the Grown areas and The more grown areas there are, the less likely they are to be fused.

[0091] Secondly, in this embodiment, the second similarity threshold is set to 0.9. If the similarity between the grown region and other grown regions is greater than the preset second similarity threshold, the grown region and other grown regions are merged into one grown region.

[0092] In addition, if the similarity between the grown region and other grown regions is less than or equal to a second similarity threshold, no fusion is performed.

[0093] S3: extracting a local high-temperature area in the thermal image according to the result of region growing to monitor the electrical equipment.

[0094] According to the operation of S2, a refined and accurate region growing result is obtained, wherein the region growing result is to divide the thermal image of the monitoring area into a plurality of grown regions.

[0095] Next, the default pixel value is 240 (an empirical value, adjustable). The red channel pixel value is generally used to represent temperature. A higher red channel pixel value generally indicates a higher temperature or a more significant feature. 240 is selected as the default pixel value to distinguish areas with higher temperatures or significant features. The importance threshold is also preset to 0.85 (an empirical value, adjustable). Therefore, the average red channel pixel value of all pixels in each grown region is calculated. When the average value is greater than 240, it indicates that the pixels in that region have a higher temperature or a significant feature.

[0096] Set the judgment condition for determining whether a grown area is a local high temperature area: if the average value of the red channel pixel values ​​of all pixels in a certain grown area is greater than 240, and the maximum importance value of all seed points contained in the grown area is greater than 0.85, the grown area is considered to be a local high temperature area in the thermal imaging image.

[0097] In the thermal image, if there is no grown area that meets the above judgment conditions, it means that there is no local high temperature area in the thermal image.

[0098] Finally, continuous acquisition For each of the multiple thermal images of the minute, the local high temperature area of ​​the thermal image is obtained by the method from step S1 to this step, and the ratio of the pixel point of the local high temperature area of ​​each thermal image to all the pixel points of the thermal image is calculated (if there is no local high temperature area in a thermal image, the ratio is 0). If the ratio is If the ratio increases within 1 minute, it means that there is an abnormally high temperature area that is gradually expanding in the thermal image, indicating that there is a high temperature hazard in the electrical equipment in the monitoring area, and an early warning is issued. If the temperature does not increase sequentially within a few minutes, such as increasing or decreasing occasionally, it means that the temperature fluctuations are caused by the normal operation of the electrical equipment in the monitoring area, and no warning is issued. It is the default value, which is set to 1 here.

[0099] In the description of this specification, "multiple" and "several" mean at least two, such as two, three or more, unless otherwise clearly defined.

[0100] While several embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, variations, and alternatives will occur to those skilled in the art without departing from the concept and spirit of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention.

Claims

1. A comprehensive monitoring platform for smart substations, characterized by: include: Collect thermal images of the monitoring area of ​​the electrical equipment in the substation, select seed points in the thermal images, perform region growing based on the seed points according to a user-defined growth criterion and a user-defined fusion criterion, determine whether there is a local high-temperature area in the thermal images based on the results of the region growing, and determine whether an early warning is needed based on the determination result; The customized growth criterion includes: for any grown region, if the similarity of pixels within a search range of a certain edge pixel point is greater than a preset similarity threshold, the pixels within the search range are classified as the grown region; otherwise, the pixels within the search range whose distance to the seed point of the grown region is less than a preset distance threshold are classified as the grown region; When the similarity of the pixels within the search range of all edge pixels of all grown areas is less than a preset first similarity threshold, the growth is stopped; The customized fusion criterion includes: if a certain grown area contains pixels of other grown areas, and the similarity between the grown area and the other grown areas is greater than a preset second similarity threshold, the grown area and the other grown areas are fused into one grown area; the similarity between the grown area and the other grown areas satisfies the following relationship: , For the Grown areas and The similarity of the grown regions, For the Grown areas and The variance consideration value of all pixels in the union of the grown regions, For the The variance consideration value of all pixels in the grown area, For the The variance consideration value of all pixels in the grown area, For the The mean similarity of pixels within the search range of all seed points and all edge pixels in the grown region, The method of determining is: if The search range of a certain edge pixel point in the grown area contains pixel points in the grown area, mark the edge pixel point, and calculate the mean similarity of the pixels in the search range of all marked edge pixels as .

2. The intelligent substation integrated monitoring platform according to claim 1, characterized in that: The customized growth criteria also include: If there is no grown area, for any seed point, if the similarity of the pixels within the search range of the seed point is greater than a preset similarity threshold, the pixels within the search range are classified as grown areas; otherwise, the pixels within the search range whose distance to the seed point is less than a preset distance threshold are classified as grown areas.

3. The intelligent substation integrated monitoring platform according to claim 1, characterized in that: The similarity of pixels within the search range of the seed point is calculated according to the following formula: ； Where, It is The similarity of pixels within the search range of the seed points, 、 、 is the variance of the red channel pixel value, the green channel pixel value, and the blue channel pixel value of all pixels in the search range. For the The sum of the gradient values ​​of all pixels within the search range of the seed point, All are custom parameters. For the The importance of the seed points, is a natural constant.

4. The intelligent substation integrated monitoring platform according to claim 1, characterized in that: The similarity of the pixels within the search range of the edge pixels satisfies the following relationship: ； Where, It is The similarity of pixels within the search range of edge pixels, 、 、 is the variance of the red channel pixel value, the green channel pixel value, and the blue channel pixel value of all pixels in the search range. For the The sum of the gradient values ​​of all pixels within the search range of edge pixels, All are custom parameters. is a natural constant.

5. The intelligent substation integrated monitoring platform according to claim 2, characterized in that: The method for selecting seed points in the thermal image includes: Randomly select several pixels as initial seed points and calculate the importance of the seed points: ； Where, For the The importance of the seed points, 、 、 Respectively The pixel value of the red channel, the pixel value of the green channel, and the pixel value of the blue channel of the seed point, To obtain the maximum value function, Function is used for normalization; The seed points whose importance is less than the preset importance threshold are discarded, and the seed points whose importance is greater than or equal to the preset importance threshold are retained as seed points for subsequent region growing.

6. The intelligent substation integrated monitoring platform according to claim 2, characterized in that: The method for judging whether there is a local high temperature area in the thermal image according to the result of region growing is: In the thermal image, calculate the average value of the red channel pixel values ​​of all pixels in each grown area; If the average value of the red channel pixel values ​​of all pixels in a grown area is greater than a preset pixel value, and the maximum value of the importance of all seed points corresponding to the grown area is greater than a preset importance threshold, the grown area is a local high temperature area existing in the thermal imaging image.

7. The intelligent substation integrated monitoring platform according to claim 1, characterized in that: The method for obtaining the variance consideration value is: calculating the sum of the squares of the variances of the pixel values ​​of the three channels of red, green and blue, and taking the arithmetic square root of the sum of the squares as the variance consideration value.

8. The intelligent substation integrated monitoring platform according to claim 3, characterized in that: The search range of the edge pixel point and the search range of the seed point are both centered on themselves. pixel range, where is the default value.

9. The intelligent substation integrated monitoring platform according to claim 6, characterized in that: The method for determining whether an early warning is needed based on the judgment results is: Continuous acquisition The thermal image of the minute, calculate the ratio of the pixel point of the local high temperature area of ​​each thermal image to all the pixel points of the thermal image. If the ratio is minutes, and an early warning is issued. If the number of increases does not increase sequentially within minutes, no warning will be issued. is the default value.

Images