A multi-modal power equipment detection system

Abstract

The application discloses a kind of multimode power equipment detection systems, it is related to power equipment detection technical field, comprising: obtaining infrared image, visible light image, operating data and environmental data, determine observation area;Observation area is corrected to temperature rise, obtain corrected temperature rise;Obtain the physical height mapping of observation area in visible light image;Observation area axial temperature rise curve is constructed and candidate peak point is determined, obtain hot band intensity and hot band center height;Based on hot band center height, calculate the skip of tapping position, obtain skip value;Tropical intensity and harmonic proportion are statistically analyzed, obtain correlation coefficient and growth amplitude;Linear fitting is carried out to tropical intensity and environmental data, obtain environmental influence coefficient;Skip value, correlation coefficient, growth amplitude and environmental influence coefficient are judged, obtain the detection result of autotransformer.The application can enhance the safety of equipment operation and the reliability of detection system.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, and in particular to a multimodal power equipment testing system. Background Technology

[0002] In power systems, autotransformers are important electrical equipment, widely used due to their compact structure and high transmission efficiency. The low-voltage side, due to the ampere-turn characteristics of the series windings and common winding, becomes a critical area sensitive to concentrated electromagnetic leakage flux and tap switching effects. Infrared hotspots near the low-voltage side, driven by electromagnetic mechanisms such as winding leakage flux, tap position abrupt changes, and harmonic amplification, can continuously induce eddy current losses in the metal structural components, accelerate insulation aging, and even cause winding short circuits due to localized overheating.

[0003] In existing technologies, transformer infrared hotspot detection mainly relies on manual interpretation of thermal images or analysis of single-related tap and harmonic data, which has three major limitations: First, it fails to establish a geometric correlation between tap switching and the highly discrete nature of hotspots, making it impossible to distinguish between tap-driven structural jumps and environmentally caused continuous drifts; second, it lacks quantitative models for harmonic proportions and hotspot intensity, making it difficult to verify the amplification effect of harmonics on eddy current losses; and third, it fails to isolate the dynamic effects of environmental interference, making it impossible to determine whether the hotspot is dominated by electromagnetic losses or environmental noise superposition. Traditional methods struggle to accurately distinguish the true causes of infrared hotspots in multi-factor scenarios, failing to support early and accurate warnings of equipment status and failing to meet the power grid's technical requirements for comprehensive and quantitative assessment of equipment. There is an urgent need to construct a detection system that integrates multimodal data and achieves full-link correlation to overcome the challenge of accurately identifying electromagnetic hotspots. Summary of the Invention

[0004] The purpose of this invention is to solve the problems in the prior art, such as the difficulty in distinguishing between electromagnetic causes and environmental interference due to environmental interference, tap position jumps and harmonic effects superposition at the low-voltage side of the autotransformer infrared hotspot, and the judgment error caused by the time sequence misalignment of multi-source data and the lack of correlation analysis. Therefore, a multi-modal power equipment detection system is proposed.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution:

[0006] A multimodal power equipment testing system, comprising:

[0007] The data acquisition and area positioning module is used to acquire infrared images, visible light images, operating data and environmental data of the autotransformer. The operating data includes: tap position and harmonic ratio, and the environmental data includes: ambient temperature, solar radiation intensity and wind speed. Based on the visible light image, the observation area of ​​the low-voltage side line of the autotransformer is determined.

[0008] The temperature rise correction module is used to correct the temperature rise of the observation area based on infrared images, ambient temperature, solar radiation intensity and wind speed, and obtain the corrected temperature rise.

[0009] The geometric scale calibration module is used to perform height conversion on multiple visible light images to obtain the physical height mapping of the observation area in the visible light image;

[0010] The thermal imaging feature extraction module is used to construct the axial temperature rise curve of the observation area and determine candidate peak points based on the corrected temperature rise and physical height mapping of the observation area, so as to obtain the tropical intensity and tropical center height.

[0011] The tapping position calculation module is used to calculate the tapping position based on the height of the tropical center and obtain the tapping value.

[0012] The harmonic thermal intensity correlation calculation module is used to statistically analyze tropical intensity and harmonic proportion, and obtain the correlation coefficient between tropical intensity and harmonic proportion and the growth rate of tropical intensity with increasing harmonics.

[0013] The environmental disturbance calculation module is used to perform linear fitting of tropical intensity, solar radiation intensity and wind speed to obtain the environmental impact coefficient;

[0014] The comprehensive analysis module is used to determine the jump value, correlation coefficient, growth rate, and environmental impact coefficient to obtain the test results of the autotransformer.

[0015] Preferably, infrared images, visible light images, operational data, and environmental data of the autotransformer are acquired. The operational data includes tap position and harmonic ratio, while the environmental data includes ambient temperature, solar radiation intensity, and wind speed. Based on the visible light images, the observation area at the low-voltage side of the autotransformer is determined, including:

[0016] Infrared images, visible light images, operating data, and environmental data of the autotransformer are acquired at fixed time intervals. The operating data includes tap position and harmonic ratio, while the environmental data includes ambient temperature, solar radiation intensity, and wind speed.

[0017] Edge detection is performed on the visible light image to extract the edges of the box wall and the first vertical reinforcing rib. The spatial intersection of the two is calculated to obtain the rectangular observation area.

[0018] Preferably, the observation area is temperature-corrected based on infrared images, ambient temperature, solar radiation intensity, and wind speed to obtain the corrected temperature rise, including:

[0019] The infrared image is converted into a temperature image, and the difference between the temperature of each pixel in the temperature image and the ambient temperature is calculated to obtain the relative temperature rise field.

[0020] During the period when the tap position is stable and the harmonic ratio fluctuation is less than the preset harmonic ratio threshold, the relative temperature rise of the relative temperature rise field pixel, solar radiation intensity and wind speed are fitted by linear regression to obtain the environmental impact coefficient.

[0021] The relative temperature rise of the observation area is corrected by the environmental impact coefficient to obtain the corrected temperature rise.

[0022] Preferably, height conversion is performed on multiple visible light images to obtain the physical height mapping of the observation area in the visible light images, including:

[0023] Identify device-inherent reference feature point pairs in visible light images and calculate the pixel spacing between the reference feature point pairs using a distance formula;

[0024] The conversion ratio between pixel distance and physical distance in a visible light image is obtained by referring to the ratio of the pixel spacing of the feature point pair to the prior actual physical distance.

[0025] Using the bottom edge of the observation area in the visible light image as the physical height zero point, the physical height mapping of the observation area in the visible light image is obtained based on the vertical coordinates of the pixels in the observation area and the transformation ratio.

[0026] Preferably, based on the corrected temperature rise and physical height mapping of the observation area, an axial temperature rise curve of the observation area is constructed and candidate peak points are determined to obtain tropical intensity and tropical center height, including:

[0027] The pixels within the observation area are grouped according to their physical height. The average corrected temperature rise of all horizontal pixels in each height group is calculated to obtain the average corrected temperature rise at that height.

[0028] Using the physical height of each group as the horizontal axis and the corrected average temperature rise as the vertical axis, a curve is formed by fitting the data points to obtain the axial temperature rise curve.

[0029] Differentiate the axial temperature rise curve to obtain potential peak points. If the corrected average temperature rise of a potential peak point is greater than or equal to a preset temperature rise threshold, then mark that point as a candidate peak point.

[0030] The point with the highest corrected average temperature rise among the candidate peak points is selected as the tropical center point, and the physical height of the tropical center point is taken as the tropical center height.

[0031] The tropical intensity is obtained by subtracting the median of the axial temperature rise curve from the corrected mean temperature rise at the tropical midpoint.

[0032] Preferably, the jump position is calculated based on the height of the tropical center to obtain the jump position value, including:

[0033] Traverse the tap position sequence of the running data, and mark the time of change when the tap position changes in consecutive time intervals;

[0034] Set up a stability window before the change event and a stability window after the change event, and extract the tropical center height corresponding to the stability window before the change event and the stability window after the change event, respectively;

[0035] The jump value is obtained by comparing the tropical center height in the stable window after the change event with the tropical center height in the stable window before the change event.

[0036] Preferably, statistical analysis is performed on tropical intensity and harmonic proportion to obtain the correlation coefficient between tropical intensity and harmonic proportion, and the increase in tropical intensity with increasing harmonics, including:

[0037] Extract the tropical intensity sequence and harmonic proportion sequence during the low-radiation period, and remove the data during the period of change in the tapping position;

[0038] A univariate linear regression model was fitted to the tropical intensity sequence and harmonic proportion sequence, and the growth rate was calculated by the least squares method.

[0039] The correlation coefficients of the tropical intensity sequence and the harmonic proportion sequence were calculated using the Pearson correlation coefficient formula.

[0040] Preferably, a linear fit is performed on tropical intensity, solar radiation intensity, and wind speed to obtain the environmental impact coefficient, including:

[0041] Extract tropical intensity, solar radiation intensity, and wind speed sequences within periods when the tapping location is stable and the harmonic proportion fluctuation is less than the preset harmonic proportion threshold.

[0042] Linear regression models were constructed for tropical intensity series, solar radiation intensity series, and wind speed series, and the environmental impact coefficients were calculated using the least squares method.

[0043] Preferably, the jump value, correlation coefficient, growth rate, and environmental impact coefficient are determined to obtain the test results of the autotransformer, including:

[0044] Determine the jump value: If the jump value belongs to the preset axial pitch range, it is determined that the change in the tropical position conforms to the winding structure law;

[0045] The growth rate and correlation coefficient are judged: if the growth rate is greater than or equal to the preset growth rate threshold and the correlation coefficient is greater than or equal to the preset coefficient threshold, it is determined that the electromagnetic loss of the tropics is enhanced by harmonic drive.

[0046] The environmental impact coefficient is determined as follows: if the environmental impact coefficient is less than or equal to the preset environmental impact coefficient threshold, the tropical intensity is determined to be unaffected by the environment.

[0047] Based on the results of the jump value determination, the growth rate and correlation coefficient determination, and the environmental impact coefficient determination, the test results of the autotransformer are obtained.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] 1. This invention achieves comprehensive coverage of multimodal data by combining infrared images, visible light images, operational data, and environmental data, providing a foundation for subsequent cross-dimensional analysis. It corrects for temperature rise in the observation area based on ambient temperature, solar radiation intensity, and wind speed, effectively eliminating the interference of dynamic environmental factors on equipment surface temperature and avoiding misjudgments caused by environmental noise, thus improving the reliability of temperature rise data.

[0050] 2. This invention converts the pixel scale of visible light images into a physical height mapping, thus establishing a correlation between image pixels and the actual physical dimensions of the equipment. This provides a quantitative spatial benchmark for determining whether changes in tropical location conform to the structural laws of the equipment. By constructing axial temperature rise curves and extracting tropical intensity and tropical center height, complex temperature rise field data is transformed into quantifiable key feature indicators. Based on the difference in tropical center height before and after the tapping position change, the correlation between tapping position changes and tropical spatial location is quantified, effectively distinguishing structural jumps driven by tapping.

[0051] 3. This invention quantifies the impact of harmonics on tropical intensity by analyzing the correlation coefficient and growth rate between tropical intensity and harmonic proportion. It clarifies whether harmonics drive hotspot formation by amplifying electromagnetic losses, filling the gap in existing technologies that lack a quantitative model for harmonic and hotspot intensity. Using linear fitting to obtain environmental impact coefficients quantifies the interference of environmental factors such as solar radiation and wind speed on tropical intensity, directly determining whether tropical intensity is dominated by the environment and solving the problem of existing technologies being unable to isolate the dynamic influence of the environment. By comprehensively judging the causes of hotspots on the low-voltage side of autotransformers through multi-condition determination, it overcomes the limitations of traditional single-data-dimensional analysis, significantly improving the accuracy and reliability of detection results, and providing a decision-making basis for early and accurate warnings and comprehensive quantitative assessments of equipment status. Attached Figure Description

[0052] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0053] Figure 1 This is a functional block diagram of a multimodal power equipment detection system provided in an embodiment of the present invention. Detailed Implementation

[0054] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0055] Example: This example provides a multimodal power equipment detection system. See [link to example]. Figure 1 Specifically, including:

[0056] The data acquisition and area positioning module is used to acquire infrared images, visible light images, operating data and environmental data of the autotransformer. The operating data includes: tap position and harmonic ratio, and the environmental data includes: ambient temperature, solar radiation intensity and wind speed. Based on the visible light image, the observation area of ​​the low-voltage side line of the autotransformer is determined.

[0057] In embodiments of the present invention, infrared images, visible light images, operational data, and environmental data of the autotransformer are acquired. The operational data includes tap position and harmonic ratio, while the environmental data includes ambient temperature, solar radiation intensity, and wind speed. Based on the visible light image, the observation area of ​​the low-voltage side of the autotransformer is determined, including:

[0058] The system acquires infrared images, visible light images, operational data, and environmental data of the autotransformer. The operational data includes tap position and harmonic ratio, while the environmental data includes ambient temperature, solar radiation intensity, and wind speed.

[0059] Specifically, the infrared images, acquired by the multimodal power equipment detection system, are two-dimensional pixel image sequences reflecting the change of surface temperature distribution of the autotransformer over time. They contain infrared intensity information of pixels in different areas of the equipment at different times, used for subsequent conversion into a temperature field and calculation of temperature rise. The visible light images, also acquired by the multimodal power equipment detection system, are two-dimensional pixel image sequences reflecting the change of the autotransformer's external structure over time. They contain visual information of the equipment's physical structure at different times, used to locate the observation area. The operational data, acquired by the multimodal power equipment detection system, includes tap position and harmonic ratio. The tap position records the tap changer's position information at different times, reflecting changes in the effective working turns of the winding, which is used for subsequent calculations. The analysis of the jump pattern of tropical center height with tap position provides an electrical parameter benchmark. The harmonic ratio is the proportion of the 5th harmonic current recorded at different times, reflecting the amplification of electromagnetic loss and providing a quantitative basis for subsequent analysis of the correlation between tropical intensity and harmonics. The ambient temperature is obtained by the multi-mode power equipment detection system, including ambient temperature, solar radiation intensity, and wind speed. Among them, the ambient temperature is used as the reference temperature to calculate the relative temperature rise of the equipment surface in the infrared image and eliminate the interference of the ambient base temperature. The solar radiation intensity quantifies the heating interference of solar radiation on the equipment surface and is used to separate this part of the influence from the temperature rise. The wind speed quantifies the heat dissipation interference of wind speed on the equipment surface and is used to separate this part of the influence from the temperature rise.

[0060] It should be noted that after obtaining the above data, the infrared image, visible light image, operational data, and environmental data need to be time-aligned.

[0061] Based on the visible light image, determine the observation area of ​​the low-voltage side line terminal of the autotransformer;

[0062] Specifically: The Canny edge detection algorithm is used to process the visible light image, extracting the edges of the enclosure wall and the first vertical reinforcing rib. The enclosure wall edge is the longitudinal contour line of the low-voltage side shell of the autotransformer, appearing as a continuous high-contrast pixel sequence. The edge of the first vertical reinforcing rib is the contour line of the first longitudinal metal reinforcing structure on the enclosure wall near the low-voltage side line end, also presented as a high-contrast edge. In the pixel coordinate system, the horizontal coordinate range of the enclosure wall edge and the horizontal coordinate range of the first vertical reinforcing rib edge are determined. The vertical coordinate range of both is the longitudinal coverage area of ​​the low-voltage side line end of the equipment. The rectangular area formed by the left and right horizontal boundaries and the upper and lower vertical boundaries is the spatial intersection of the two, obtaining the pixel coordinates of the rectangular observation area. This observation area focuses on the area between the enclosure wall and the first vertical reinforcing rib at the low-voltage side line end of the autotransformer, eliminating interference from other structures and defining a precise spatial range for subsequent temperature rise calculation and feature extraction.

[0063] The temperature rise correction module is used to correct the temperature rise of the observation area based on infrared images, ambient temperature, solar radiation intensity and wind speed, and obtain the corrected temperature rise.

[0064] In an embodiment of the present invention, temperature rise correction is performed on the observation area based on infrared images, ambient temperature, solar radiation intensity, and wind speed to obtain the corrected temperature rise, including:

[0065] The infrared image is converted into a temperature image, and the difference between the temperature of each pixel in the temperature image and the ambient temperature is calculated to obtain the relative temperature rise field.

[0066] Specifically, based on the factory calibration parameters of the infrared thermal imager, the grayscale value of each pixel in the observation area of ​​the infrared image is converted to obtain the absolute temperature corresponding to that pixel, forming a converted temperature image. The ambient temperature at the same time as the temperature image is extracted, and for each pixel in the observation area, the difference between its absolute temperature and the ambient temperature is calculated to obtain the relative temperature of each pixel. The relative temperature rise of all pixels constitutes a relative temperature rise field, which is used to eliminate the influence of the ambient base temperature on the overall equipment and focus on the local temperature rise of the equipment surface relative to the environment, laying the foundation for the subsequent removal of environmental dynamic interference.

[0067] During the period when the tap position is stable and the harmonic ratio fluctuation is less than the preset harmonic ratio threshold, the relative temperature rise of the relative temperature rise field pixel, solar radiation intensity and wind speed are fitted by linear regression to obtain the environmental impact coefficient.

[0068] Specifically, the stability criterion for the tap position is set as no change in the tap position for 30 consecutive minutes, and the preset harmonic proportion threshold is 2%, meaning the fluctuation range of the 5th harmonic proportion is less than 2%. According to electromagnetic principles, the eddy current loss of metal structural components is proportional to the square of the current frequency. The loss of the 5th harmonic (frequency 250Hz) is 25 times that of the fundamental frequency (50Hz). When the ratio of the 5th harmonic current to the fundamental current is less than 2%, the additional electromagnetic loss it generates is only 1% of the fundamental frequency loss. This level of loss is within the error range of infrared thermometry and its impact on temperature rise is negligible, ensuring that temperature rise changes during this period are mainly dominated by environmental factors. The time period that simultaneously meets both of the above conditions is designated as the clean period for environmental impact analysis. During this period, the impact of electromagnetic factors on temperature rise is negligible, ensuring that the fitting results only reflect the effects of environmental factors. Within the clean period, linear regression is performed on the relative temperature rise, solar radiation intensity, and wind speed of each pixel within the observation area. The fitting formula is:

[0069]

[0070] In the formula, For pixels within the observation area At any moment The relative temperature rise, For pixels The solar radiation influence coefficient. For a moment The intensity of solar radiation, For pixels The wind speed influence coefficient, For a moment wind speed, For pixels At any moment The fitting residuals;

[0071] The arithmetic mean of the solar radiation influence coefficient and wind speed influence coefficient of all pixels in the observation area is taken to obtain the regional average solar radiation influence coefficient and the regional average wind speed influence coefficient.

[0072] The relative temperature rise of the observation area is corrected by the environmental impact coefficient to obtain the corrected temperature rise;

[0073] Specifically, regarding time For each pixel within the observation area, the corrected temperature rise is obtained by subtracting the influence components of solar radiation and wind speed from the relative temperature rise of each pixel. The corrected temperature rise for all pixels is then calculated by subtracting the influence components of solar radiation from the average solar radiation influence coefficient. The product of solar radiation intensity at time, the wind speed influence component is the average wind speed influence coefficient and The product of wind speed at that time.

[0074] The geometric scale calibration module is used to perform height conversion on multiple visible light images to obtain the physical height mapping of the observation area in the visible light image;

[0075] In embodiments of the present invention, height conversion is performed on multiple visible light images to obtain a physical height mapping of the observation area in the visible light images, including:

[0076] Identify device-inherent reference feature point pairs in visible light images and calculate the pixel spacing between the reference feature point pairs using a distance formula;

[0077] Specifically, within the observation area of ​​the visible light image, inherent physical markers on the surface of the equipment are selected as reference feature point pairs. Among them, structures with stable positions, high contrast, and known actual dimensions are preferred, such as the center of the fixing bolt on the wall of the low-voltage side line box. The center pixel coordinates of the reference feature point pair are recorded, and the pixel distance between the two points is calculated using the Euclidean distance formula.

[0078] The conversion ratio between pixel distance and physical distance in a visible light image is obtained by referring to the ratio of the pixel spacing of the feature point pair to the prior actual physical distance.

[0079] Specifically, referring to the equipment design drawings of the autotransformer, the actual physical straight-line distance between the above-mentioned reference feature point pairs is extracted, and the actual physical straight-line distance is divided by the pixel pitch to obtain the conversion ratio between pixel distance and physical distance.

[0080] Using the bottom edge of the observation area in the visible light image as the physical height zero point, the physical height mapping of the observation area in the visible light image is obtained based on the vertical coordinates of the pixels in the observation area and the transformation ratio.

[0081] Specifically, in the pixel coordinate system of the visible light image, the bottom edge of the observation area is set as the zero point of the physical height, and the positive direction of the physical height is upward along the vertical axis of the image. The physical height is calculated by subtracting the vertical coordinate of the bottom edge of the observation area from the vertical coordinate of any pixel in the observation area and then multiplying it by the transformation ratio. The physical height of all pixels in the observation area is used to construct the physical height mapping of the observation area. This provides a unified physical scale benchmark for subsequent tropical center height positioning and axial pitch correlation of tap joint sections.

[0082] The thermal imaging feature extraction module is used to construct the axial temperature rise curve of the observation area and determine candidate peak points based on the corrected temperature rise and physical height mapping of the observation area, so as to obtain the tropical intensity and tropical center height.

[0083] In an embodiment of the present invention, based on the corrected temperature rise and physical height mapping of the observation area, an axial temperature rise curve of the observation area is constructed and candidate peak points are determined to obtain tropical intensity and tropical center height, including:

[0084] The pixels within the observation area are grouped according to their physical height. The average corrected temperature rise of all horizontal pixels in each height group is calculated to obtain the average corrected temperature rise at that height.

[0085] Specifically, based on the physical height mapping of the observation area, the bottom edge of the observation area is taken as the zero point of height, and the physical height of the top edge is taken as the maximum interval value to determine the physical height range of the observation area. The range is divided into continuous height groups with a fixed step size. All pixels in the observation area are assigned to the corresponding height group according to their physical height. The corrected average temperature rise of all pixels in each height group is calculated as the corrected average temperature rise of that height.

[0086] Using the physical height of each group as the horizontal axis and the corrected average temperature rise as the vertical axis, a curve is formed by fitting the data points to obtain the axial temperature rise curve.

[0087] Specifically, the midpoint of the physical height interval of each height group is taken as the horizontal axis data point, and the corrected average temperature rise of each height group is taken as the vertical axis data point to form a data point set. The data points are then fitted using cubic spline interpolation to obtain the axial temperature rise curve.

[0088] Differentiate the axial temperature rise curve to obtain potential peak points. If the corrected average temperature rise of a potential peak point is greater than or equal to a preset temperature rise threshold, then mark that point as a candidate peak point.

[0089] Specifically, the first and second derivatives of the axial temperature rise curve are calculated using the numerical differentiation method. Points that simultaneously satisfy the condition that the first derivative value is 0 and the second derivative value is less than 0 are retained and marked as potential peak points. Based on the maximum temperature rise during normal operation of the equipment, a preset temperature rise threshold is set, and only points whose corrected average temperature rise value of potential peak points is greater than or equal to the preset temperature rise threshold are retained as candidate peak points.

[0090] The point with the highest corrected average temperature rise among the candidate peak points is selected as the tropical center point, and the physical height of the tropical center point is taken as the tropical center height.

[0091] Specifically, the candidate peak points are sorted in descending order according to the corrected average temperature rise, and the point ranked first is taken as the tropical center point. The physical height of the tropical center point is taken as the tropical center height.

[0092] The tropical intensity is obtained by subtracting the median of the axial temperature rise curve from the corrected mean temperature rise at the tropical midpoint.

[0093] Specifically, the median of the axial temperature rise curve is calculated across all corrected mean temperature rise ranges. The difference between the corrected mean temperature rise at the tropical center and the median of the axial temperature rise curve is taken as the tropical intensity.

[0094] The tapping position calculation module is used to calculate the tapping position based on the height of the tropical center and obtain the tapping value.

[0095] In an embodiment of the present invention, the jump position is calculated based on the tropical center height to obtain the jump position value, including:

[0096] Traverse the tap position sequence of the running data, and mark the time of change when the tap position changes in consecutive time intervals;

[0097] Specifically, extract the sequence of tap position changes over time from the running data, traverse the tap positions of two consecutive moments in the sequence, calculate the absolute value of the difference, and if the absolute value of the difference is greater than 0, mark the next moment as the tap position change event moment.

[0098] Set up a stability window before the change event and a stability window after the change event, and extract the tropical center height corresponding to the stability window before the change event and the stability window after the change event, respectively;

[0099] Specifically, taking the moment of the tap change event as the endpoint, a 3-minute period is extracted before the event as the pre-event window, ensuring that the tap position remains constant within this window. Taking the moment of the tap change event as the starting point, a 3-minute period is extracted after the event as the post-event window, ensuring that the tap position remains constant within this window. Tropical center height data are extracted from the stable windows before and after the event, and their arithmetic mean is calculated to serve as the tropical center height for the stable windows before and after the event.

[0100] The jump value is obtained by comparing the tropical center height in the stable window after the change event with the tropical center height in the stable window before the change event.

[0101] Specifically, the jump value is calculated by subtracting the tropical center height of the stable window before the change event from the tropical center height of the stable window after the change event; the jump value quantifies the correlation between the change in tapping position and the jump in tropical spatial location.

[0102] The harmonic thermal intensity correlation calculation module is used to statistically analyze tropical intensity and harmonic proportion, and obtain the correlation coefficient between tropical intensity and harmonic proportion and the growth rate of tropical intensity with increasing harmonics.

[0103] In embodiments of the present invention, statistical analysis is performed on tropical intensity and harmonic proportion to obtain the correlation coefficient between tropical intensity and harmonic proportion, and the increase in tropical intensity with increasing harmonics, including:

[0104] Extract the tropical intensity sequence and harmonic proportion sequence during the low-radiation period, and remove the data during the period of change in the tapping position;

[0105] Specifically, based on the critical value of the radiation intensity of the equipment surface directly exposed to the sun, a solar radiation intensity threshold is set, and the time period with a solar radiation intensity less than the solar radiation intensity threshold is selected from the solar radiation intensity sequence as a low radiation period. In the low radiation period, the transient interval period of one minute before and after the moment when the tap position changes is removed to obtain the effective time period when the tap position is stable. The tropical intensity data and the proportion data of the 5th harmonic are extracted in the effective time period.

[0106] A univariate linear regression model was fitted to the tropical intensity sequence and harmonic proportion sequence, and the growth rate was calculated by the least squares method.

[0107] Specifically, a linear relationship model is established between tropical intensity and the proportion of the 5th harmonic, with the following formula:

[0108]

[0109] In the formula, For tropical intensity, For the growth rate, The proportion of the 5th harmonic. For constant terms, The growth rate is determined by the least squares method, which represents the random error term. .

[0110] The correlation coefficients of the tropical intensity sequence and the harmonic proportion sequence were calculated using the Pearson correlation coefficient formula.

[0111] Specifically, based on the tropical intensity sequence and harmonic proportion sequence, a Pearson correlation coefficient formula is established to obtain the correlation coefficient between the tropical intensity sequence and the harmonic proportion sequence. The specific formula is as follows:

[0112]

[0113] In the formula, The Pearson correlation coefficient between the tropical intensity sequence and the harmonic proportion sequence. For the number of valid data, The first in the effective sequence of tropical intensity One element, The first harmonic proportion in the effective sequence Each element.

[0114] The environmental disturbance calculation module is used to perform linear fitting of tropical intensity, solar radiation intensity and wind speed to obtain the environmental impact coefficient;

[0115] In an embodiment of the present invention, a linear fit is performed on tropical intensity, solar radiation intensity, and wind speed to obtain an environmental impact coefficient, including:

[0116] Extract tropical intensity, solar radiation intensity, and wind speed sequences within periods when the tapping location is stable and the harmonic proportion fluctuation is less than the preset harmonic proportion threshold.

[0117] Specifically, the stability criterion for the tap position is set as no change in the tap position within 30 consecutive minutes, and the preset harmonic proportion threshold is 2%, that is, the fluctuation range of the 5th harmonic proportion is less than 2%. The time period that meets both of the above conditions is recorded as the clean time period for environmental impact analysis. During this period, the influence of electromagnetic factors on temperature rise can be ignored, ensuring that the fitting results only reflect the effects of environmental factors; the tropical intensity sequence, solar radiation intensity sequence and wind speed sequence are extracted within the clean time period.

[0118] Linear regression models were constructed for tropical intensity sequences, solar radiation intensity sequences, and wind speed sequences, and environmental impact coefficients were calculated using the least squares method.

[0119] Specifically, using tropical intensity as the dependent variable and solar radiation intensity and wind speed as independent variables, a multiple linear regression model is constructed, with the following formula:

[0120]

[0121] In the formula, For the first Tropical intensity data This represents the solar radiation influence coefficient. Solar radiation intensity, The wind speed influence coefficient, For wind speed, For constant terms, The random error term is represented by the solar radiation influence coefficient and the wind speed influence coefficient, which are obtained by the least squares method.

[0122] The comprehensive analysis module is used to determine the jump value, correlation coefficient, growth rate and environmental impact coefficient to obtain the test results of the autotransformer;

[0123] In embodiments of the present invention, the jump value, correlation coefficient, growth rate, and environmental impact coefficient are determined to obtain the detection results of the autotransformer, including:

[0124] Determine the jump value: If the jump value belongs to the preset axial pitch range, it is determined that the change in the tropical position conforms to the winding structure law;

[0125] Specifically, based on the autotransformer winding design drawings, the standard value of the winding axial displacement corresponding to each tap change is extracted. Considering manufacturing errors and measurement deviations, the axial pitch range is set as follows: Divide the jump value by the tap position change to obtain the jump value per unit position. If the jump value per unit position belongs to the preset axial pitch range, it is determined that the tropical position change conforms to the winding structure law; otherwise, it is determined that it does not conform. This determination verifies the correlation between the tropical spatial position jump and the winding physical structure, and excludes random position offsets caused by non-electromagnetic factors.

[0126] The growth rate and correlation coefficient are judged: if the growth rate is greater than or equal to the preset growth rate threshold and the correlation coefficient is greater than or equal to the preset coefficient threshold, it is determined that the electromagnetic loss of the tropics is enhanced by harmonic drive.

[0127] Specifically, based on experimental data of electromagnetic loss of similar equipment under harmonic interference, a threshold of 0.5 was set for the increase in intensity, which corresponds to the minimum significant increase in tropical intensity when the proportion of harmonics increases by 1%. According to statistical standards, a threshold of 0.7 was set for the correlation coefficient, which corresponds to the critical level of strong linear correlation, ensuring that the association between harmonics and tropical intensity is statistically significant. If the increase in intensity is greater than or equal to 0.5 and the correlation coefficient is greater than or equal to 0.7, it is determined that the electromagnetic loss of the tropics is enhanced by harmonics; otherwise, it is determined that it is not significantly driven by harmonics. This determination quantifies the amplification effect of harmonics on tropical intensity and verifies the dominance of the electromagnetic loss mechanism.

[0128] The environmental impact coefficient is determined as follows: if the environmental impact coefficient is less than or equal to the preset environmental impact coefficient threshold, the tropical intensity is determined to be unaffected by the environment.

[0129] Specifically, environmental impact coefficient thresholds are set, with the solar radiation impact coefficient threshold at 0.3 and the wind speed impact coefficient threshold at 0.2. These thresholds are based on the maximum impact of environmental factors on the surface temperature of the equipment when there are no electromagnetic hotspots, ensuring that environmental interference is within a negligible range. If the solar radiation impact coefficient is less than or equal to 0.3 and the wind speed impact coefficient is less than or equal to 0.2, it is determined that the tropical intensity is not dominated by the environment; otherwise, it is determined that the tropical intensity is significantly affected by environmental factors. This determination excludes the interference of environmental factors such as solar radiation and wind speed on the tropical intensity, ensuring that the detection results focus on electromagnetic causes.

[0130] Based on the results of the jump value determination, the growth rate and correlation coefficient determination, and the environmental impact coefficient determination, the test results of the autotransformer are obtained.

[0131] Specifically, if the determination result is that the change in the tropical position conforms to the winding structure law, and the electromagnetic loss of the tropical region is enhanced by harmonic drive, and the intensity of the tropical region is not dominated by the environment, then the detection result is that there is an electromagnetic infrared hot spot on the low-voltage side of the autotransformer. This result indicates that the formation of the hot spot is directly related to electromagnetic factors such as winding leakage flux, tap position jump and harmonic amplification, and environmental interference can be ignored, providing a clear basis for fault location and cause diagnosis for equipment operation and maintenance. Conversely, if any one of the determinations is not met, it is determined to be a non-electromagnetic hot spot or the cause of the hot spot is unknown.

[0132] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multimodal power equipment testing system, characterized in that, include: The data acquisition and area positioning module is used to acquire infrared images, visible light images, operating data and environmental data of the autotransformer. The operating data includes: tap position and harmonic ratio, and the environmental data includes: ambient temperature, solar radiation intensity and wind speed. Based on the visible light image, the observation area of ​​the low-voltage side line of the autotransformer is determined. The temperature rise correction module is used to correct the temperature rise of the observation area based on infrared images, ambient temperature, solar radiation intensity and wind speed, and obtain the corrected temperature rise. The geometric scale calibration module is used to perform height conversion on multiple visible light images to obtain the physical height mapping of the observation area in the visible light image; The thermal imaging feature extraction module is used to construct the axial temperature rise curve of the observation area and determine candidate peak points based on the corrected temperature rise and physical height mapping of the observation area, so as to obtain the tropical intensity and tropical center height. The tapping position calculation module is used to calculate the tapping position based on the height of the tropical center and obtain the tapping value. The harmonic thermal intensity correlation calculation module is used to statistically analyze tropical intensity and harmonic proportion, and obtain the correlation coefficient between tropical intensity and harmonic proportion and the growth rate of tropical intensity with increasing harmonics. The environmental disturbance calculation module is used to perform linear fitting of tropical intensity, solar radiation intensity and wind speed to obtain the environmental impact coefficient; The comprehensive analysis module is used to determine the jump value, correlation coefficient, growth rate, and environmental impact coefficient to obtain the test results of the autotransformer.

2. The multimodal power equipment testing system according to claim 1, characterized in that, Acquire infrared images, visible light images, operational data, and environmental data of the autotransformer. Operational data includes tap position and harmonic content; environmental data includes ambient temperature, solar radiation intensity, and wind speed. Based on the visible light images, determine the observation area of ​​the low-voltage side of the autotransformer, including: Infrared images, visible light images, operating data, and environmental data of the autotransformer are acquired at fixed time intervals. The operating data includes tap position and harmonic ratio, while the environmental data includes ambient temperature, solar radiation intensity, and wind speed. Edge detection is performed on the visible light image to extract the edges of the box wall and the first vertical reinforcing rib. The spatial intersection of the two is calculated to obtain the rectangular observation area.

3. The multimodal power equipment testing system according to claim 1, characterized in that, Temperature rise correction was performed on the observed area based on infrared images, ambient temperature, solar radiation intensity, and wind speed to obtain the corrected temperature rise, including: The infrared image is converted into a temperature image, and the difference between the temperature of each pixel in the temperature image and the ambient temperature is calculated to obtain the relative temperature rise field. During the period when the tap position is stable and the harmonic ratio fluctuation is less than the preset harmonic ratio threshold, the relative temperature rise of the relative temperature rise field pixel, solar radiation intensity and wind speed are fitted by linear regression to obtain the environmental impact coefficient. The relative temperature rise of the observation area is corrected by the environmental impact coefficient to obtain the corrected temperature rise.

4. The multimodal power equipment testing system according to claim 1, characterized in that, Height transformation is performed on multiple visible light images to obtain the physical height mapping of the observed region in the visible light image, including: Identify device-inherent reference feature point pairs in visible light images and calculate the pixel spacing between the reference feature point pairs using a distance formula; The conversion ratio between pixel distance and physical distance in a visible light image is obtained by referring to the ratio of the pixel spacing of the feature point pair to the prior actual physical distance. Using the bottom edge of the observation area in the visible light image as the physical height zero point, the physical height mapping of the observation area in the visible light image is obtained based on the vertical coordinates of the pixels in the observation area and the transformation ratio.

5. The multimodal power equipment testing system according to claim 1, characterized in that, Based on the corrected temperature rise and physical height mapping of the observation area, an axial temperature rise curve for the observation area is constructed and candidate peak points are determined, yielding tropical intensity and tropical center height, including: The pixels within the observation area are grouped according to their physical height. The average corrected temperature rise of all horizontal pixels in each height group is calculated to obtain the average corrected temperature rise at that height. Using the physical height of each group as the horizontal axis and the corrected average temperature rise as the vertical axis, a curve is formed by fitting the data points to obtain the axial temperature rise curve. Differentiate the axial temperature rise curve to obtain potential peak points. If the corrected average temperature rise of a potential peak point is greater than or equal to a preset temperature rise threshold, then mark that point as a candidate peak point. The point with the highest corrected average temperature rise among the candidate peak points is selected as the tropical center point, and the physical height of the tropical center point is taken as the tropical center height. The tropical intensity is obtained by subtracting the median of the axial temperature rise curve from the corrected mean temperature rise at the tropical midpoint.

6. The multimodal power equipment testing system according to claim 1, characterized in that, Based on the tropical center altitude, the tapping position is calculated to obtain the tapping value, including: Traverse the tap position sequence of the running data, and mark the time of the change event when the tap position changes in consecutive moments; Set up a stability window before the change event and a stability window after the change event, and extract the tropical center height corresponding to the stability window before the change event and the stability window after the change event, respectively; The jump value is obtained by comparing the tropical center height in the stable window after the change event with the tropical center height in the stable window before the change event.

7. The multimodal power equipment testing system according to claim 1, characterized in that, Statistical analysis of tropical intensity and harmonic proportion yielded the correlation coefficient between tropical intensity and harmonic proportion, as well as the increase in tropical intensity with increasing harmonics, including: Extract the tropical intensity sequence and harmonic proportion sequence during the low-radiation period, and remove the data during the period of change in the tapping position; A univariate linear regression model was fitted to the tropical intensity sequence and harmonic proportion sequence, and the growth rate was calculated by the least squares method. The correlation coefficients of the tropical intensity sequence and the harmonic proportion sequence were calculated using the Pearson correlation coefficient formula.

8. The multimodal power equipment testing system according to claim 1, characterized in that, Linear fitting was performed on tropical intensity, solar radiation intensity, and wind speed to obtain environmental impact coefficients, including: Extract tropical intensity, solar radiation intensity, and wind speed sequences within periods when the tapping location is stable and the harmonic proportion fluctuation is less than the preset harmonic proportion threshold. Linear regression models were constructed for tropical intensity sequences, solar radiation intensity sequences, and wind speed sequences, and the environmental impact coefficients were calculated using the least squares method.

9. A multimodal power equipment testing system according to claim 1, characterized in that, The test results of the autotransformer are obtained by determining the jump value, correlation coefficient, growth rate, and environmental impact coefficient, including: Determine the jump value: If the jump value belongs to the preset axial pitch range, it is determined that the change in the tropical position conforms to the winding structure law; The growth rate and correlation coefficient are judged: if the growth rate is greater than or equal to the preset growth rate threshold and the correlation coefficient is greater than or equal to the preset coefficient threshold, it is determined that the electromagnetic loss of the tropics is enhanced by harmonic drive. The environmental impact coefficient is determined as follows: if the environmental impact coefficient is less than or equal to the preset environmental impact coefficient threshold, the tropical intensity is determined to be unaffected by the environment. The test results of the autotransformer are obtained based on the results of the jump value determination, the growth rate and correlation coefficient determination, and the environmental impact coefficient determination.

Images