Moire-based power equipment temperature monitoring system and method

Abstract

The invention discloses a moire-based power equipment temperature monitoring system and method. The moire-based power equipment temperature monitoring system comprises a temperature sensing device, an image preprocessing module, a moire feature extraction module and a temperature measurement and calculation module, a camera is used for detecting a moire visual mark which deforms along with the temperature change of the power equipment in the temperature sensing device; according to the invention, the temperature change of the power equipment is described as the change of the ratio of the projection frequency of the moire visual marked one-dimensional periodic grating calculated according to the moire characteristics on the imaging plane of the camera to the reference projection frequency, and the real-time temperature of the measured power equipment can be output.

Description

Technical Field

[0001] This invention belongs to the field of temperature monitoring technology, specifically relating to a temperature monitoring system and method for power equipment based on moiré patterns. Background Technology

[0002] The operating temperature of electrical equipment is typically within a relatively fixed range; a real-time temperature exceeding this range is a precursor to equipment failure. High-voltage busbars and enclosed busbars, among other electrical equipment, may pose safety hazards such as fires due to poor contact or eddy currents in the casing. Failures in these systems can have serious consequences. Therefore, high-precision temperature monitoring of this electrical equipment is necessary to prevent and promptly detect equipment failure risks, ensuring safe and stable operation.

[0003] The mainstream technologies for temperature measurement of power equipment can be divided into contact temperature measurement technology and non-contact temperature measurement technology.

[0004] Contact temperature measurement technology directly contacts the electrical equipment under test, indirectly determining the temperature through changes in the physical parameters of the sensing element. Common contact temperature sensing technologies for electrical equipment mainly include thermocouple temperature measurement, resistance temperature detector (RTD) temperature measurement, and fiber optic temperature measurement. Thermocouple temperature measurement technology is based on the Seebeck effect. Its principle is to weld two conductors of different materials together, with one end as the measuring end and the other end as the reference end. When there is a temperature difference between the measuring end and the reference end, a thermoelectric electromotive force is generated in the circuit. By measuring this electromotive force, the temperature of the measuring end can be determined. However, thermocouple temperature measurement requires cold junction compensation and the signal is susceptible to electromagnetic interference, resulting in relatively low accuracy. Resistance temperature detector (RTD) temperature measurement utilizes the characteristic that the resistance value of a metallic conductor changes regularly with temperature to measure temperature. The most common types include platinum RTDs and copper RTDs. RTD temperature measurement technology faces the problem of self-heating effect. During measurement, current needs to be passed through, which generates heat, causing the measured value to be higher than the actual temperature. Furthermore, both resistance temperature detectors (RTDs) and thermocouples are limited by battery life, requiring timely shutdown and power supply replacement. While using a CT (circuit optic) for power can solve the battery life issue, it increases the cost and complexity of the temperature measurement system. Fiber optic temperature measurement mainly includes fiber Bragg grating (FBG) measurement and distributed fiber optic temperature measurement. The former is based on the Bragg grating principle, while the latter analyzes the backscattered light generated by laser light in the fiber. Fiber optic temperature measurement is extremely expensive, with system costs often reaching tens of thousands of yuan or more.

[0005] Infrared thermometry is the most widely used non-contact temperature measurement technology, which determines the temperature of an object by measuring its infrared radiation energy. Common infrared thermometry devices include point infrared thermometers and area array infrared thermal imagers. The former measures the temperature at a single point, while the latter generates a visualized "thermal image" of a specified measurement range, reflecting the real-time temperature of each sampling area at the pixel level. Infrared thermometry mainly faces the problem of measured values ​​being lower than the true values ​​due to the low emissivity of the object's surface. Furthermore, infrared thermometry requires an unobstructed connection between the sensor and the device being measured; obstructions such as fog or ordinary glass can lead to inaccurate measurements. In complex environments, the absolute temperature measurement accuracy of infrared thermometry is generally lower than that of contact thermometry, making it unsuitable for high-precision temperature monitoring scenarios. Summary of the Invention

[0006] In view of the shortcomings of the prior art, the purpose of this invention is to provide a power equipment temperature monitoring system and method based on moiré patterns, so as to solve the problems of insufficient accuracy, high cost, poor robustness and need for power supply replacement in existing power equipment temperature monitoring methods.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] The present invention provides a temperature monitoring system for power equipment based on moiré patterns, comprising: a temperature sensing device, an image preprocessing module, a moiré pattern feature extraction module, and a temperature measurement module;

[0009] Temperature sensing devices are used to detect temperature changes in electrical equipment and convert these temperature changes into changes in moiré pattern characteristics.

[0010] The image preprocessing module is used to obtain the central region of interest from the moiré image captured by the camera, preprocess the image of the central region of interest, and output it.

[0011] The moiré pattern feature extraction module is used to solve for moiré pattern features based on the image output by the image preprocessing module.

[0012] The temperature measurement module is used to calculate intermediate variables related to temperature based on moiré pattern characteristics, and finally determine the current temperature of the power equipment.

[0013] Furthermore, the temperature sensing device is installed on the surface of the electrical equipment to sense temperature changes in the electrical equipment; it includes: a temperature sensing chamber with a piston, a rectangular baffle with rollers, a cuboid shell with a track, a T-shaped pull rod, a fixing rod, and moiré visual markings.

[0014] A temperature-sensing cavity with a piston is used to sense temperature changes in electrical equipment. It consists of a rigid cavity and a piston. The rigid cavity has a cuboid structure and is connected to a cylindrical needle-shaped piston. The rigid cavity is filled with a temperature-sensing medium with a high coefficient of thermal expansion. As the temperature changes, the temperature-sensing medium in the rigid cavity expands or contracts, and the volume change of the temperature-sensing medium causes the piston to move.

[0015] A rectangular baffle with rollers is connected to the piston in the temperature sensing chamber and the vertical end of the T-shaped lever. It moves along the track in the cuboid shell with track as the piston moves.

[0016] The rectangular shell with tracks has five open faces out of its six faces, while the face in the vertical direction connected to the fixed rod is not open. Tracks are also provided on the four horizontal edges of the rectangular shell to restrict the movement of a rectangular baffle with rollers along the fixed track direction. The tracks are parallel to the object's coordinate system. The X-axis direction;

[0017] A T-shaped pull rod has its vertical end connected to a rectangular baffle with rollers, and its horizontal end connected to one side of a moiré visual marker. It moves along with the rectangular baffle, pulling the connected moiré visual marker and causing it to move along the object coordinate system. Deformation in the X-axis direction;

[0018] The fixing rod is T-shaped, with its vertical end fixed to the non-cutout surface of the cuboid shell with the track, and its horizontal end connected to one side of the moiré visual mark to ensure that one side of the moiré visual mark remains fixed.

[0019] The moiré visual marker is made of elastic material. One end is connected to a fixed rod, and the other end is connected to a T-shaped tie rod. It contains one one-dimensional periodic grating and four positioning marks, which are located at the four corner points of the one-dimensional periodic grating.

[0020] Furthermore, the specific process by which the temperature sensing device converts temperature changes into changes in moiré pattern characteristics is as follows: As the temperature of the power equipment rises or falls, the temperature-sensing medium inside the rigid cavity of the temperature-sensing chamber with the piston expands or contracts; the expansion or contraction of the temperature-sensing medium generates a thrust or suction force on the piston, causing the piston to move; the movement of the piston drives a rectangular baffle with rollers connected to the piston to move along a track on a cuboid shell with a track; the movement of the rectangular baffle with rollers drives a T-shaped pull rod connected to it to move; the T-shaped pull rod pulls the moiré pattern visual mark connected to it, causing the moiré pattern visual mark to be pulled and move along the object coordinate system. Deformation along the X-axis; moiré pattern visual markers along the object coordinate system Deformation along the X-axis causes a change in the frequency of the one-dimensional periodic grating, which in turn causes a change in the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane, ultimately resulting in a change in the moiré pattern characteristics.

[0021] Furthermore, the one-dimensional periodic grating of the moiré visual marker is used to form the moiré pattern. The projection of the one-dimensional periodic grating onto the imaging plane of the camera is superimposed on the color filter array of the camera to form the moiré pattern in the captured moiré pattern image. The propagation direction of the one-dimensional periodic grating is parallel to the track direction of the cuboid shell with track, and its period number is known.

[0022] Furthermore, the four positioning marks of the moiré visual marker are used to crop out the central region of interest in the captured moiré image and in the image coordinate system. The relative spatial frequency of moiré patterns in the region of interest at the center is estimated.

[0023] Furthermore, the object coordinate system The origin in It is the geometric center of the Moore visual mark, that is The projection of a point onto the captured moiré image is located at the midpoint of the image; object coordinate system The X-axis is parallel to the propagation direction of the one-dimensional periodic grating of the moiré visual mark, the Y-axis is parallel to the side of the one-dimensional periodic grating perpendicular to the propagation direction, and the Z-axis is orthogonal to the [missing information - likely a specific point or axis]. flat.

[0024] Furthermore, the image coordinate system The origin in It is the geometric center of the captured moiré image; image coordinate system The X-axis and Y-axis in the image are parallel to the long side and short side, respectively.

[0025] Furthermore, aligning the camera with the direction of the moiré visual markers ensures that the object's coordinate system... The plane is parallel to the camera's imaging plane, and the edges of the moiré visual markers along the propagation direction in the captured moiré image are parallel to the image coordinate system. The X-axis direction is approximately parallel (the included angle does not exceed 5°).

[0026] Furthermore, the image preprocessing module specifically performs the following steps:

[0027] The four corner points of the one-dimensional periodic grating are located based on the positioning marks. Then, the region containing the one-dimensional periodic grating is cropped from the captured moiré image as a coarse selection of the region of interest, and the number of pixels of the side length of the one-dimensional periodic grating perpendicular to the propagation direction is obtained. ;

[0028] After coarsely selecting the region of interest, trim the edges to obtain a side length of... The central region of interest is selected to eliminate the impact of unstable features in the edge regions on accuracy.

[0029] Based on the side length (in pixels) of the central region of interest By comparing the location markers in the captured moiré pattern image, the relative spatial frequency of the moiré pattern in the central region of interest image is estimated. ;

[0030] The R-channel image is separated from the central region of interest. Histogram equalization is performed on the R-channel image to enhance image contrast. Then, the frequency domain three-hole mask algorithm is used to remove noise frequency components from the equalized image. Adaptive binarization is performed on the image with noise frequency components removed to obtain an adaptive binarized image.

[0031] The adaptive binarized image corresponds to the central region of interest in the captured moiré pattern image. It is both the estimated relative spatial frequency of moiré patterns in the central region of interest image and the estimated relative spatial frequency of moiré patterns in the adaptive binarized image. It is both the side length in pixels of the central region of interest and the side length in pixels of the adaptively binarized image; according to and The ratio determines the subsequent image processing used: when When, use the adaptive binarized image as the output; when If the moiré stripes are too dense, a discrete search strategy is used to determine the optimal thumbnail side length. The adaptive binarized image is then downsampled according to the optimal thumbnail side length to generate a thumbnail, which is then used as the output.

[0032] Furthermore, the number of pixels based on the side length of the central region of interest The spatial relative frequency of the moiré pattern in the region of interest image is estimated by comparing the location markers in the captured moiré pattern image with the spatial relative frequency of the moiré pattern. Specifically:

[0033] Based on the positioning marks in the captured moiré image, the four corner points of the one-dimensional periodic grating region are located in the captured moiré image, and then the side length and number of pixels of the one-dimensional periodic grating along the propagation direction are obtained. and its edges along the propagation direction relative to the image coordinate system Inclination angle in the positive X-axis direction ;

[0034] Based on the side length and number of pixels along the propagation direction of the one-dimensional periodic grating Inclination angle and the number of pixels on the side of the central region of interest. To estimate the relative spatial frequency of moiré patterns in the image of the central region of interest. The expression is as follows:

[0035] ;

[0036] in, This represents the pixel unit size of the color filter array; For the color filter array frequency, ; The number of periods contained in a one-dimensional periodic grating.

[0037] Furthermore, the frequency domain three-aperture mask algorithm is specifically as follows:

[0038] Perform a Fast Fourier Transform on the equalized image to obtain the original amplitude spectrum. Phase spectrum ;

[0039] Reproduce the original amplitude spectrum Obtain its copy Create a spectrum similar to the original amplitude spectrum. Masks of the same size with all values ​​equal to 0 ;

[0040] In the copy Find the pixel with the largest amplitude value in the mask, which corresponds to the DC pulse; Lieutenant General centered on DC pulse Change the value of the region to 1, and set the copy... The same in The value of the region is changed to 0, indicating that DC pulses will no longer be searched for.

[0041] Finding a copy For the pixel with the largest amplitude value, calculate the relative spatial frequency of that pixel and compare it with the estimated relative spatial frequency of the moiré pattern in the central region of interest image. The absolute value of the difference; if the absolute value of the difference does not exceed a predefined threshold, then the pixel is considered to correspond to a moiré frequency component, and in the mask... The lieutenant general centered on that pixel Change the value of the region to 1, and set the copy... Centered on this pixel The value of the region is changed to 0, indicating that the pixel has already been found and will not be searched again; otherwise, the pixel is discarded, and it is determined that the pixel is not a moiré frequency component, and only the copy is copied. Centered on this pixel Change the value of the region to 0, indicating that the pixel will not be considered in subsequent searches; repeat the above operation until a pair of moiré frequency components are found;

[0042] The above steps yield a result that is only three. A mask where the value is 1 in the specified area and 0 in all other pixels. ; mask Applied to the original amplitude spectrum This yields an amplitude spectrum with noise frequency components removed. Then, based on the amplitude spectrum after removing noise frequency components... Phase spectrum The spatial domain image is restored using inverse Fourier transform.

[0043] Furthermore, the computational copy The specific method for determining the relative spatial frequency corresponding to the pixel with the largest amplitude value is as follows:

[0044] Establish an amplitude spectrum coordinate system OPQ, with the DC pulse as the origin, and the horizontal and vertical axes representing replicas, respectively. The horizontal rightward direction and the vertical upward direction;

[0045] Set up a copy The coordinates of the pixel with the largest amplitude value are Then the copy The relative spatial frequency corresponding to the pixel with the largest amplitude value is .

[0046] Furthermore, the specific method for determining the optimal thumbnail side length using a discrete search strategy is as follows:

[0047] With sampling rate Downsampling is performed on the adaptively binarized image to generate a thumbnail; sampling rate Equal to the number of pixels of the thumbnail side With the side length and number of pixels of the adaptive binarized image The ratio; the operation of scaling down the adaptive binarized image is equivalent to superimposing the adaptive binarized image with spatial frequencies of... Grid;

[0048] Based on the relative spatial frequency of moiré patterns in the adaptive binarized image Spatial frequency of superimposed grid Thumbnail side length in pixels To estimate the relative spatial frequency of moiré patterns in thumbnails The expression is as follows:

[0049] ;

[0050] because Therefore, the upper bound of the search range is ;

[0051] by Spatial frequency of superimposed grid Relative spatial frequency of moiré patterns in adaptive binarized images Similar judgment conditions are as follows: This determines the lower bound of the search range as follows. ;

[0052] The search range for the optimal thumbnail side length is a discrete interval. The estimated relative spatial frequency of moiré patterns in the adaptively binarized image. Substituting this into the search range yields the specific discrete search interval. Extending the evaluation tolerance interval to the right of the specific discrete search interval, the final search interval for the optimal thumbnail side length in pixels is then determined as follows: , This is the length of the valuation tolerance interval;

[0053] In the final search interval, the optimal thumbnail side length is determined by a joint search algorithm using the Scharr operator and the Laplacian operator.

[0054] Furthermore, the specific steps for determining the optimal thumbnail side length using the joint search algorithm of the Scharr and Laplacian operators are as follows:

[0055] Final search interval Each integer value in the matrix could be the optimal thumbnail side length, called the candidate thumbnail side length; the thumbnail generated from the adaptive binarized image according to the candidate thumbnail side length is called the thumbnail corresponding to the candidate thumbnail side length.

[0056] The Scharr operator is used to calculate the average gradient magnitude of the thumbnail corresponding to each candidate thumbnail side length. The average gradient magnitude is used as the Scharr operator image sharpness of the candidate thumbnail side length. The Scharr operator image sharpness of each candidate thumbnail side length is collected. A difference-based method is used to find the candidate thumbnail side length where the Scharr operator image sharpness reaches the maximum value, resulting in a list of maximum value points.

[0057] Using the thumbnail corresponding to each maximum point as the input image for the moiré feature extraction module, the relative spatial frequency of the moiré pattern in the thumbnail corresponding to each maximum point is calculated. and direction of dissemination And then according to , and the value of the maximum point Restore the relative spatial frequency of moiré patterns in the adaptive binarized image. The expression is as follows:

[0058] ;

[0059] Find the relative spatial frequency of moiré patterns in the reconstructed adaptive binarized image. Relative spatial frequency of moiré patterns in the estimated adaptive binarized image The closest maximum point is selected as the best thumbnail side length by the Scharr operator;

[0060] Repeat all the operations using the Scharr operator above using the Laplacian operator to obtain the best thumbnail side length selected by the Laplacian operator. Specifically: use the Laplacian operator to calculate the average response intensity of the thumbnail corresponding to each candidate thumbnail side length, and use the average response intensity as the Laplacian operator image sharpness of the candidate thumbnail side length; collect the Laplacian operator image sharpness of each candidate thumbnail side length, and then use a difference-based method to obtain the list of maxima of the Laplacian operator; use the thumbnail corresponding to each maxima of the Laplacian operator's maxima list as the input image of the moiré feature extraction module, calculate the relative spatial frequency and propagation direction of the moiré pattern of the thumbnail corresponding to each maxima of the Laplacian operator's maxima list, and then reconstruct the relative spatial frequency of the moiré pattern of the adaptive binarized image; find the relative spatial frequency of the moiré pattern of the reconstructed adaptive binarized image and... The closest maximum point is selected as the best thumbnail side length by the Laplacian operator;

[0061] Then, a second selection process is performed on the optimal thumbnail side lengths chosen by the two operators: the relative spatial frequency of the moiré patterns in the reconstructed adaptive binarized image is selected. The closest estimated relative spatial frequency of moiré patterns in the adaptively binarized image The thumbnail side length is taken as the final optimal thumbnail side length.

[0062] Furthermore, the moiré pattern feature extraction module specifically performs the following steps:

[0063] The input image is either the image output by the image preprocessing module or a thumbnail corresponding to the maxima point of the joint search algorithm of the Scharr and Laplacian operators. The amplitude spectrum of the input image is obtained by applying a Gaussian window and a Fast Fourier Transform. ;

[0064] Find the amplitude spectrum The point with the largest amplitude value corresponds to the DC pulse; establish an amplitude spectrum coordinate system OUV, with the DC pulse as the origin, and the horizontal and vertical axes representing the amplitude spectrum, respectively. Horizontal to the right and vertical upward; amplitude spectrum Each pixel in the spectrum corresponds to a frequency component, amplitude spectrum The vector from the origin to other pixels is called the frequency vector of the frequency component corresponding to the pixel.

[0065] Find the amplitude spectrum The pixel with the largest amplitude value besides the DC pulse. , This corresponds to the pixel-level moiré frequency components; then, a Gaussian function-based method is used to calculate the sub-pixel-level moiré frequency vector of the input image. If the Gaussian function-based method has no solution, a dynamically weighted fast Fourier transform-based method is used to calculate the sub-pixel-level moiré frequency vector of the input image; let the calculated sub-pixel-level moiré frequency vector of the input image be... ,in and These are the horizontal and vertical components of the subpixel-level moiré frequency vector of the input image, respectively.

[0066] When the input image is a thumbnail corresponding to the maximum point of the joint search algorithm of the Scharr and Laplacian operators, the relative spatial frequency of the moiré pattern of the input image is used. and direction of dissemination As output;

[0067] When the input image is the image output by the image preprocessing module, according to and Determine the subpixel moiré frequency vector of the adaptive binarized image And then according to The moiré pattern features of the adaptive binarized image are calculated as output, including the spatial frequency of the moiré pattern in the adaptive binarized image. and the direction of moiré propagation .

[0068] Furthermore, the method based on Gaussian functions for calculating the sub-pixel moiré frequency vector of the input image specifically involves:

[0069] Use a two-dimensional Gaussian function The amplitude peaks corresponding to the fitted moiré frequency components are expressed as follows:

[0070] ;

[0071] in, The amplitude value. For natural numbers, and These are the x and y coordinates of the amplitude spectrum coordinate system OUV, respectively. It is an unknown quantity;

[0072] Two-dimensional Gaussian function natural logarithm The corresponding expression for a bivariate quadratic function with 6 unknowns is as follows:

[0073] ;

[0074] In Centered In the area, look for besides The 5 pixels with the largest amplitude values ​​outside of the range; use And the fitting of the horizontal and vertical coordinates and amplitude values ​​of the five pixels found in the amplitude spectrum coordinate system OUV. The corresponding quadratic function in two variables is obtained by solving. These 6 unknowns are used to calculate... Coordinates of the maximum value point The expression is as follows;

[0075] ;

[0076] ;

[0077] The coordinates of the calculated maximum value point are: Then the subpixel moiré frequency vector of the input image is .

[0078] Furthermore, the method based on dynamically weighted fast Fourier transform specifically calculates the sub-pixel level moiré frequency vector of the input image as follows:

[0079] Calculate the amplitude spectrum separately China and Israel The sum of the amplitudes of the four 2×2 regions (top left, top right, bottom left, and bottom right) is used to select the 2×2 region with the largest sum as the final weighted region.

[0080] The coordinates of the four pixels in the final weighted region in the amplitude spectrum coordinate system OUV are as follows: , , , The weighted coordinates are obtained by taking a weighted average of the coordinates of the four pixels in the final weighted region, with the magnitude value as the weight. This leads to the sub-pixel moiré frequency vector of the input image. The specific formula is as follows:

[0081] ;

[0082] ;

[0083] in, For the final weighted region, the first The amplitude value of each pixel; and These are the first two terms in the final weighted region. The x and y coordinates of each pixel in the amplitude spectrum coordinate system OUV; and These are the x and y coordinates of the weighted coordinates, respectively.

[0084] Furthermore, when the input image is the image output by the image preprocessing module, according to and Determine the subpixel moiré frequency vector of the adaptive binarized image The specific method is as follows:

[0085] when At that time, the image output by the image preprocessing module is an adaptive binarized image, therefore the sub-pixel level moiré frequency vector of the adaptive binarized image is... ;

[0086] when When the image preprocessing module outputs a thumbnail of the adaptive binarized image, then... It is the moiré frequency vector of the thumbnail, according to Compared to the thumbnail side length in pixels The subpixel-level moiré frequency vector of the adaptive binarized image is calculated as follows: .

[0087] Furthermore, the sub-pixel moiré frequency vector of the adaptively binarized image... The specific expression for calculating the moiré pattern features of an adaptive binarized image is as follows:

[0088] ;

[0089] ;

[0090] in, and , respectively, are the horizontal and vertical components of the sub-pixel moiré frequency vector of the adaptive binarized image, where c is the pixel unit size of the color filter array. To determine the number of pixels per side of the adaptively binarized image. and These represent the spatial frequency and propagation direction of the moiré pattern in the adaptive binarized image, respectively.

[0091] Furthermore, the temperature measurement module specifically performs the following steps:

[0092] A specified temperature is selected as the reference temperature, and the moiré image captured by the camera at this reference temperature is used as the reference moiré image. The frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane at the reference temperature is calculated using the moiré features extracted from the reference moiré image, and this frequency is used as the reference projection frequency.

[0093] Each time temperature is monitored, the moiré pattern image captured at the current temperature is used as the current moiré pattern image; the frequency of the projection of a one-dimensional periodic grating onto the camera imaging plane at the current temperature is calculated using the moiré pattern features extracted from the current moiré pattern image, and this frequency is used as the current projection frequency.

[0094] Calculate the ratio of the reference projection frequency to the current projection frequency, and use it as the current projection frequency ratio; determine the current temperature based on the mapping relationship between the projection frequency ratio and temperature using the current projection frequency ratio.

[0095] The image preprocessing module preprocesses the reference moiré image (or the current moiré image), while the moiré feature extraction module solves for the moiré features of the adaptive binarized image at the reference temperature (or the current temperature).

[0096] Furthermore, the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane... The formula for calculation is:

[0097] ;

[0098] in, For the color filter array frequency, and These represent the spatial frequency and propagation direction of the moiré pattern in the adaptive binarized image, respectively.

[0099] Furthermore, the mapping relationship between the projection frequency ratio and temperature is calculated using the physical properties of the elastic material, specifically as follows:

[0100] The side length of the one-dimensional periodic grating of the moiré visual mark along the propagation direction is pre-calibrated at the reference temperature, and its value is obtained. ;

[0101] The deformation of the one-dimensional periodic grating for moiré visual markings is only affected by piston movement. The volume of the constant-mass sensing medium within the temperature-sensing device at different temperatures is calculated using the PengRob or REFPROP model, thus yielding the volume change of the sensing medium relative to a reference temperature. The positional change of the piston relative to the reference temperature is calculated based on the piston's inner diameter and the volume change of the sensing medium. This positional change of the piston relative to the reference temperature is the deformation of the one-dimensional periodic grating relative to the reference temperature. The deformation of a one-dimensional periodic grating increases as the side length along the propagation direction increases relative to the reference temperature. If the value is positive, the deformation of a one-dimensional periodic grating decreases with respect to the reference temperature as the side length along the propagation direction decreases. It is a negative number;

[0102] Projection frequency ratio and , The relationship is:

[0103] ;

[0104] in, For reference projection frequency, For projection frequencies at other temperatures, The distance from the one-dimensional periodic grating to the camera's imaging plane is [missing information]. The focal length of the camera. and These represent the frequencies of the one-dimensional periodic grating at the reference temperature and at other temperatures, respectively. Let be the number of periods in a one-dimensional periodic grating. and Let be the side lengths of the one-dimensional periodic grating along the propagation direction at the reference temperature and at other temperatures, respectively. The deformation of a one-dimensional periodic grating relative to a reference temperature at other temperatures;

[0105] By calculating the deformation of a one-dimensional periodic grating at other temperatures relative to a reference temperature... The projection frequency ratio at each temperature can be calculated, and finally, a mapping relationship between the projection frequency ratio and temperature can be established.

[0106] The power equipment temperature monitoring in this invention refers to the use of a camera to detect the moiré visual marks in a temperature sensing device that deform as the power equipment temperature changes; it is applicable to scenarios where the relative positions of the camera and the power equipment are fixed; this invention characterizes the temperature change of the power equipment as the change in the ratio of the projection frequency of the one-dimensional periodic grating of the moiré visual mark on the camera imaging plane to the reference projection frequency; this invention can output the real-time temperature of the power equipment being measured.

[0107] This invention also provides a method for monitoring the temperature of power equipment based on moiré patterns, which, based on the above system, includes the following steps:

[0108] 1) Install temperature sensing devices on the surface of electrical equipment to detect temperature changes in the electrical equipment;

[0109] 2) Select a reference temperature and use a camera to capture a moiré pattern image at that reference temperature;

[0110] 3) Process the captured moiré image to obtain an adaptive binarized image and the estimated relative spatial frequency of the moiré pattern in the adaptive binarized image. ;according to With the side length and number of pixels of the adaptive binarized image Whether the ratio reaches 24% determines whether to output an adaptive binarized image or a thumbnail;

[0111] 4) Calculate the sub-pixel level moiré frequency vector of the adaptive binarized image based on the output image, and then obtain the moiré spatial frequency of the adaptive binarized image. and the direction of moiré propagation ;

[0112] 5) According to and The frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane is calculated and used as the reference projection frequency.

[0113] 6) During temperature monitoring, acquire the moiré image captured at the current temperature in real time and repeat steps 3) to 4), and calculate the spatial frequency of the moiré pattern in the adaptive binarized image at the current temperature. and the direction of moiré propagation ;according to and Calculate the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane at the current temperature, and use it as the current projection frequency; calculate the ratio of the reference projection frequency to the current projection frequency, and use it as the current projection frequency ratio; use the current projection frequency ratio to determine the current temperature based on the mapping relationship between the projection frequency ratio and the temperature.

[0114] The beneficial effects of this invention are:

[0115] Ultra-high precision: Moiré patterns are formed by superimposing the color filter array of the camera with the projection of the moiré visual markers onto the camera's imaging plane. This results in extremely high sensitivity to relative changes between the camera and the moiré visual markers. Temperature changes in electrical equipment cause the moiré visual markers to deform under forces along their propagation direction, leading to significant changes in the moiré features. Based on a moiré generalization enhancement algorithm (using a frequency domain three-hole mask algorithm to remove noise frequency components, and determining the optimal thumbnail side length based on a discrete search strategy when moiré stripes are too dense), the image is preprocessed to obtain a high-quality preprocessed image. A moiré feature extraction module extracts sub-pixel-level moiré features, and then, combined with a temperature measurement module, calculates the intermediate variable related to temperature (i.e., the ratio of the reference projection frequency to the current projection frequency). This results in an average temperature measurement error of 0.003℃ for the system at a temperature resolution of 0.1℃.

[0116] Robustness: Traditional feature-point-based computer vision methods rely on a limited number of feature points and are easily affected by the environment; while the statistical characteristics of moiré fringes in the frequency domain can provide more accurate and robust information for temperature monitoring. Furthermore, the image preprocessing module of this invention can acquire high-quality preprocessed images, further improving the robustness of the system.

[0117] Low cost: This invention requires only an entry-level industrial camera and a simple temperature sensing device to achieve high-precision temperature monitoring at a low cost.

[0118] Passive: The temperature sensing device of the present invention is passive and does not require shutdown to replace the power supply. Attached Figure Description

[0119] Figure 1 This is the overall flowchart of the system of the present invention;

[0120] Figure 2 This is a scene diagram of the temperature monitoring system of the present invention in an embodiment;

[0121] Figure 3 This is a schematic diagram of the temperature sensing device of the present invention;

[0122] Figure 4 A schematic diagram of the structure of a moiré visual marker;

[0123] Figure 5 This is a flowchart of the frequency domain three-hole mask algorithm of the present invention;

[0124] Figure 6 This is a schematic diagram of the frequency domain three-hole mask algorithm of the present invention;

[0125] Figure 7This is a flowchart of the joint search algorithm for the Scharr and Laplacian operators of the present invention.

[0126] Figure 8 This is a schematic diagram of the temperature measurement module of the present invention. Detailed Implementation

[0127] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.

[0128] Reference Figures 1 to 8 As shown, the present invention provides a power equipment temperature monitoring system based on moiré patterns, which is applied in a camera-connected computing device in a camera-based non-contact temperature monitoring scenario. The system includes: a temperature sensing device, an image preprocessing module, a moiré pattern feature extraction module, and a temperature measurement module.

[0129] Temperature sensing devices are used to detect temperature changes in electrical equipment and convert these temperature changes into changes in moiré pattern characteristics.

[0130] The temperature sensing device is installed on the surface of the power equipment to sense temperature changes in the power equipment; it includes: a temperature sensing chamber 1 with a piston, a rectangular baffle 2 with rollers, a cuboid shell 3 with a track, a T-shaped pull rod 4, a fixing rod 5, and a moiré visual mark 6.

[0131] The temperature sensing cavity 1 with a piston is used to sense the temperature change of the power equipment. It consists of a rigid cavity 11 and a piston 12. The rigid cavity 11 has a cuboid structure and is connected to the cylindrical needle-shaped piston 12. The rigid cavity 11 is filled with a temperature sensing medium with a high coefficient of thermal expansion. As the temperature changes, the temperature sensing medium in the rigid cavity 11 expands or contracts, and the volume change of the temperature sensing medium causes the piston 12 to move.

[0132] A rectangular baffle 2 with rollers is connected to the piston 12 in the temperature sensing chamber 1 with piston and the vertical end of the T-shaped pull rod 4. It moves along the track in the cuboid shell 3 with track as the piston 12 moves.

[0133] The rectangular shell 3 with tracks has five of its six faces that are openwork, while the face that is vertical and connected to the fixing rod 5 is not openwork. Tracks are provided on the four horizontal edges of the rectangular shell 3 to restrict the movement of the rectangular baffle 2 with rollers along the fixed track direction. The tracks are parallel to the object's coordinate system. The X-axis direction;

[0134] The T-shaped pull rod 4 has its vertical end connected to a rectangular baffle 2 with rollers, and its horizontal end connected to one side of the moiré visual mark 6. It moves along with the rectangular baffle 2 with rollers, thereby pulling the moiré visual mark 6 connected to it, causing the moiré visual mark 6 to move along the object coordinate system. Deformation in the X-axis direction;

[0135] The fixing rod 5 is T-shaped, with its vertical end fixed to the non-cutout surface of the cuboid shell 3 with the track, and its horizontal end connected to one side of the moiré visual mark 6 to ensure that one side of the moiré visual mark 6 remains fixed.

[0136] The moiré visual marker 6 is made of elastic material. One end of it is connected to the fixed rod 5, and the other end is connected to the T-shaped tie rod 4. It includes a one-dimensional periodic grating and four positioning marks (ArUco marks). The four positioning marks are located at the four corner points of the one-dimensional periodic grating.

[0137] Specifically, the process by which the temperature sensing device converts temperature changes into changes in moiré pattern features is as follows: As the temperature of the electrical equipment rises or falls, the temperature-sensing medium inside the rigid cavity of the temperature-sensing chamber with the piston expands or contracts; the expansion or contraction of the temperature-sensing medium generates a pushing or suction force on the piston, causing the piston to move; the movement of the piston drives a rectangular baffle with rollers connected to the piston to move along a track on a cuboid shell with a track; the movement of the rectangular baffle with rollers drives a T-shaped pull rod connected to it to move; the T-shaped pull rod pulls the moiré pattern visual mark connected to it, causing the moiré pattern visual mark to be pulled and move along the object coordinate system. Deformation along the X-axis; moiré pattern visual markers along the object coordinate system Deformation along the X-axis causes a change in the frequency of the one-dimensional periodic grating, which in turn causes a change in the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane, ultimately resulting in a change in the moiré pattern characteristics.

[0138] The one-dimensional periodic grating of the moiré visual marker is used to form the moiré pattern. The projection of the one-dimensional periodic grating onto the imaging plane of the camera is superimposed on the color filter array (CFA) of the camera to form the moiré pattern in the captured moiré pattern image. The propagation direction of the one-dimensional periodic grating is parallel to the track direction of the cuboid shell with track, and its period number is known.

[0139] The four positioning markers of the moiré visual marker are used to crop out the central region of interest in the captured moiré image and in the image coordinate system. The relative spatial frequency of moiré patterns in the region of interest at the center is estimated.

[0140] The object coordinate system The origin in It is the geometric center of the Moore visual mark, that is The projection of a point onto the captured moiré image is located at the midpoint of the image; object coordinate system The X-axis is parallel to the propagation direction of the one-dimensional periodic grating of the moiré visual mark, the Y-axis is parallel to the side of the one-dimensional periodic grating perpendicular to the propagation direction, and the Z-axis is orthogonal to the [missing information - likely a specific point or axis]. flat.

[0141] The image coordinate system The origin in It is the geometric center of the captured moiré image; image coordinate system The X-axis and Y-axis in the image are parallel to the long side (horizontal side) and the short side (vertical side), respectively.

[0142] In this case, the camera is aligned with the direction of the moiré visual marker, so that the object coordinate system... The plane is parallel to the camera's imaging plane, and the edges of the moiré visual markers along the propagation direction in the captured moiré image are parallel to the image coordinate system. The X-axis direction is approximately parallel (the included angle does not exceed 5°).

[0143] The image preprocessing module is used to obtain the central region of interest from the moiré image captured by the camera, preprocess the image of the central region of interest, and output it.

[0144] The image preprocessing module specifically performs the following steps:

[0145] The four corner points of the one-dimensional periodic grating are located based on the positioning marks. Then, the region containing the one-dimensional periodic grating is cropped from the captured moiré image as a coarse selection of the region of interest, and the number of pixels of the side length of the one-dimensional periodic grating perpendicular to the propagation direction is obtained. ;

[0146] After coarsely selecting the region of interest, trim the edges to obtain a side length of... The central region of interest is selected to eliminate the impact of unstable features in the edge regions on accuracy.

[0147] Based on the side length (in pixels) of the central region of interest By comparing the location markers in the captured moiré pattern image, the relative spatial frequency of the moiré pattern in the central region of interest image is estimated. ;

[0148] The R-channel image is separated from the central region of interest. Histogram equalization is performed on the R-channel image to enhance image contrast. Then, the frequency domain three-hole mask algorithm is used to remove noise frequency components from the equalized image. Adaptive binarization is performed on the image with noise frequency components removed to obtain an adaptive binarized image.

[0149] The adaptive binarized image corresponds to the central region of interest in the captured moiré pattern image. It is both the estimated relative spatial frequency of moiré patterns in the central region of interest image and the estimated relative spatial frequency of moiré patterns in the adaptive binarized image. It is both the side length in pixels of the central region of interest and the side length in pixels of the adaptively binarized image; according to and The ratio determines the subsequent image processing used: when When, use the adaptive binarized image as the output; when If the moiré stripes are too dense, a discrete search strategy is used to determine the optimal thumbnail side length. The adaptive binarized image is then downsampled according to the optimal thumbnail side length to generate a thumbnail, which is then used as the output.

[0150] Specifically, the number of pixels based on the side length of the central region of interest The spatial relative frequency of the moiré pattern in the region of interest image is estimated by comparing the location markers in the captured moiré pattern image with the spatial relative frequency of the moiré pattern. Specifically:

[0151] Based on the positioning marks in the captured moiré image, the four corner points of the one-dimensional periodic grating region are located in the captured moiré image, and then the side length and number of pixels of the one-dimensional periodic grating along the propagation direction are obtained. and its edges along the propagation direction relative to the image coordinate system Inclination angle in the positive X-axis direction ;

[0152] Based on the side length and number of pixels along the propagation direction of the one-dimensional periodic grating Inclination angle and the number of pixels on the side of the central region of interest. To estimate the relative spatial frequency of moiré patterns in the image of the central region of interest. The expression is as follows:

[0153] ;

[0154] in, This represents the pixel unit size of the color filter array; For the color filter array frequency, ; The number of periods contained in a one-dimensional periodic grating.

[0155] Specifically, the frequency domain three-aperture mask algorithm is as follows:

[0156] Perform a Fast Fourier Transform on the equalized image to obtain the original amplitude spectrum. Phase spectrum ;

[0157] Reproduce the original amplitude spectrum Obtain its copy Create a spectrum similar to the original amplitude spectrum. Masks of the same size with all values ​​equal to 0 ;

[0158] In the copy Find the pixel with the largest amplitude value in the mask, which corresponds to the DC pulse; Lieutenant General centered on DC pulse Change the value of the region to 1, and set the copy... The same in The value of the region is changed to 0, indicating that DC pulses will no longer be searched for.

[0159] Finding a copy For the pixel with the largest amplitude value, calculate the relative spatial frequency of that pixel and compare it with the estimated relative spatial frequency of the moiré pattern in the central region of interest image. The absolute value of the difference; if the absolute value of the difference does not exceed a predefined threshold (usually set to 10 pixels), then the pixel is considered to correspond to a moiré frequency component, and in the mask... The lieutenant general centered on that pixel Change the value of the region to 1, and set the copy... Centered on this pixel The value of the region is changed to 0, indicating that the pixel has already been found and will not be searched again; otherwise, the pixel is discarded, and it is determined that the pixel is not a moiré frequency component, and only the copy is copied. Centered on this pixel Change the value of the region to 0, indicating that the pixel will not be considered in subsequent searches; repeat the above operation until a pair of moiré frequency components are found;

[0160] The above steps yield a result that is only three. A mask where the value is 1 in the specified area and 0 in all other pixels. ; mask Applied to the original amplitude spectrum This yields an amplitude spectrum with noise frequency components removed. Then, based on the amplitude spectrum after removing noise frequency components... Phase spectrum The spatial domain image is restored using inverse Fourier transform.

[0161] Wherein, the computed copy The specific method for determining the relative spatial frequency corresponding to the pixel with the largest amplitude value is as follows:

[0162] Establish an amplitude spectrum coordinate system OPQ, with the DC pulse as the origin, and the horizontal axis (P-axis) and vertical axis (Q-axis) as replicas. The horizontal rightward direction and the vertical upward direction;

[0163] Set up a copy The coordinates of the pixel with the largest amplitude value are Then the copy The relative spatial frequency corresponding to the pixel with the largest amplitude value is .

[0164] The specific method for determining the optimal thumbnail side length using a discrete search strategy is as follows:

[0165] With sampling rate Downsampling is performed on the adaptively binarized image to generate a thumbnail; sampling rate Equal to the number of pixels of the thumbnail side With the side length and number of pixels of the adaptive binarized image The ratio; the operation of scaling down the adaptive binarized image is equivalent to superimposing the adaptive binarized image with spatial frequencies of... Grid;

[0166] Based on the relative spatial frequency of moiré patterns in the adaptive binarized image Spatial frequency of superimposed grid Thumbnail side length in pixels To estimate the relative spatial frequency of moiré patterns in thumbnails The expression is as follows:

[0167] ;

[0168] because Therefore, the upper bound of the search range is ;

[0169] by Spatial frequency of superimposed grid Relative spatial frequency of moiré patterns in adaptive binarized images Similar judgment conditions are as follows: This determines the lower bound of the search range as follows. ;

[0170] The search range for the optimal thumbnail side length is a discrete interval. The estimated relative spatial frequency of moiré patterns in the adaptively binarized image. Substituting this into the search range yields the specific discrete search interval. Extending the evaluation tolerance interval to the right of the specific discrete search interval, the final search interval for the optimal thumbnail side length in pixels is then determined as follows: , The length of the error tolerance interval for valuation is generally set to 30;

[0171] In the final search interval, the optimal thumbnail side length is determined by a joint search algorithm using the Scharr operator and the Laplacian operator.

[0172] Specifically, the method of using a joint search algorithm of the Scharr operator and the Laplacian operator to determine the optimal thumbnail side length is as follows:

[0173] Final search interval Each integer value in the matrix could be the optimal thumbnail side length, called the candidate thumbnail side length; the thumbnail generated from the adaptive binarized image according to the candidate thumbnail side length is called the thumbnail corresponding to the candidate thumbnail side length.

[0174] The Scharr operator is used to calculate the average gradient magnitude of the thumbnail corresponding to each candidate thumbnail side length. The average gradient magnitude is used as the Scharr operator image sharpness of the candidate thumbnail side length. The Scharr operator image sharpness of each candidate thumbnail side length is collected. A difference-based method is used to find the candidate thumbnail side length where the Scharr operator image sharpness reaches the maximum value, resulting in a list of maximum value points.

[0175] Using the thumbnail corresponding to each maximum point as the input image for the moiré feature extraction module, the relative spatial frequency of the moiré pattern in the thumbnail corresponding to each maximum point is calculated. and direction of dissemination And then according to , and the value of the maximum point Restore the relative spatial frequency of moiré patterns in the adaptive binarized image. The expression is as follows:

[0176] ;

[0177] Find the relative spatial frequency of moiré patterns in the reconstructed adaptive binarized image. Relative spatial frequency of moiré patterns in the estimated adaptive binarized image The closest maximum point is selected as the best thumbnail side length by the Scharr operator;

[0178] Repeat all the operations using the Scharr operator above using the Laplacian operator to obtain the best thumbnail side length selected by the Laplacian operator. Specifically: use the Laplacian operator to calculate the average response intensity of the thumbnail corresponding to each candidate thumbnail side length, and use the average response intensity as the Laplacian operator image sharpness of the candidate thumbnail side length; collect the Laplacian operator image sharpness of each candidate thumbnail side length, and then use a difference-based method to obtain the list of maxima of the Laplacian operator; use the thumbnail corresponding to each maxima of the Laplacian operator's maxima list as the input image of the moiré feature extraction module, calculate the relative spatial frequency and propagation direction of the moiré pattern of the thumbnail corresponding to each maxima of the Laplacian operator's maxima list, and then reconstruct the relative spatial frequency of the moiré pattern of the adaptive binarized image; find the relative spatial frequency of the moiré pattern of the reconstructed adaptive binarized image and... The closest maximum point is selected as the best thumbnail side length by the Laplacian operator;

[0179] Then, a second selection process is performed on the optimal thumbnail side lengths chosen by the two operators: the relative spatial frequency of the moiré patterns in the reconstructed adaptive binarized image is selected. The closest estimated relative spatial frequency of moiré patterns in the adaptively binarized image The thumbnail side length is taken as the final optimal thumbnail side length.

[0180] The moiré feature extraction module is used to solve the moiré features (including the spatial frequency and propagation direction of the moiré) of the adaptive binarized image or to calculate the relative spatial frequency and propagation direction of the moiré in the thumbnail corresponding to the maximum point of the joint search algorithm of the Scharr operator and the Laplacian operator.

[0181] The moiré pattern feature extraction module specifically performs the following steps:

[0182] The input image is either the image output by the image preprocessing module or a thumbnail corresponding to the maxima point of the joint search algorithm of the Scharr and Laplacian operators. The amplitude spectrum of the input image is obtained by applying a Gaussian window and a Fast Fourier Transform. ;

[0183] Find the amplitude spectrum The point with the largest amplitude value corresponds to the DC pulse; establish an amplitude spectrum coordinate system OUV, with the DC pulse as the origin, and the horizontal axis (U-axis) and vertical axis (V-axis) representing the amplitude spectrum. Horizontal to the right and vertical upward; amplitude spectrum Each pixel in the spectrum corresponds to a frequency component, amplitude spectrum The vector from the origin to other pixels is called the frequency vector of the frequency component corresponding to the pixel.

[0184] Find the amplitude spectrum The pixel with the largest amplitude value besides the DC pulse. , This corresponds to the pixel-level moiré frequency components; then, a Gaussian function-based method is used to calculate the sub-pixel-level moiré frequency vector of the input image. If the Gaussian function-based method has no solution, a dynamically weighted fast Fourier transform-based method is used to calculate the sub-pixel-level moiré frequency vector of the input image; let the calculated sub-pixel-level moiré frequency vector of the input image be... ,in and These are the horizontal and vertical components of the subpixel-level moiré frequency vector of the input image, respectively.

[0185] When the input image is a thumbnail corresponding to the maximum point of the joint search algorithm of the Scharr and Laplacian operators, the relative spatial frequency of the moiré pattern of the input image is used. and direction of dissemination As output;

[0186] When the input image is the image output by the image preprocessing module, according to and Determine the subpixel moiré frequency vector of the adaptive binarized image And then according to The moiré pattern features of the adaptive binarized image are calculated as output, including the spatial frequency of the moiré pattern in the adaptive binarized image. and the direction of moiré propagation .

[0187] Specifically, the method based on Gaussian functions for calculating the sub-pixel moiré frequency vector of the input image is as follows:

[0188] Use a two-dimensional Gaussian function The amplitude peaks corresponding to the fitted moiré frequency components are expressed as follows:

[0189] ;

[0190] in, The amplitude value. For natural numbers, and These are the x and y coordinates of the amplitude spectrum coordinate system OUV, respectively. It is an unknown quantity;

[0191] Two-dimensional Gaussian function natural logarithm Corresponding to a variable containing 6 unknowns (i.e.) The expression for a quadratic function in two variables is as follows:

[0192] ;

[0193] In (amplitude spectrum) Centered on the pixel with the largest amplitude value other than the DC pulse. In the area, look for besides The 5 pixels with the largest amplitude values ​​outside of the range; use And the fitting of the horizontal and vertical coordinates and amplitude values ​​of the five pixels found in the amplitude spectrum coordinate system OUV. The corresponding quadratic function in two variables is obtained by solving. These 6 unknowns are used to calculate... Coordinates of the maximum value point The expression is as follows;

[0194] ;

[0195] ;

[0196] The coordinates of the calculated maximum value point are: Then the subpixel moiré frequency vector of the input image is .

[0197] Specifically, the method based on dynamically weighted fast Fourier transform for calculating the sub-pixel moiré frequency vector of the input image is as follows:

[0198] Calculate the amplitude spectrum separately China and Israel (amplitude spectrum) The pixel with the largest amplitude value (excluding the DC pulse) is the sum of the amplitudes of four 2×2 regions: the top left, top right, bottom left, and bottom right. The 2×2 region with the largest amplitude sum is selected as the final weighted region.

[0199] The coordinates of the four pixels in the final weighted region in the amplitude spectrum coordinate system OUV are as follows: , , , The weighted coordinates are obtained by taking a weighted average of the coordinates of the four pixels in the final weighted region, with the magnitude value as the weight. This leads to the sub-pixel moiré frequency vector of the input image. The specific formula is as follows:

[0200] ;

[0201] ;

[0202] in, For the final weighted region, the first The amplitude value of each pixel; and These are the first two terms in the final weighted region. The x and y coordinates of each pixel in the amplitude spectrum coordinate system OUV; and These are the x and y coordinates of the weighted coordinates, respectively.

[0203] Specifically, when the input image is the image output by the image preprocessing module, according to and Determine the subpixel moiré frequency vector of the adaptive binarized image The specific method is as follows:

[0204] when At that time, the image output by the image preprocessing module is an adaptive binarized image, therefore the sub-pixel level moiré frequency vector of the adaptive binarized image is... ;

[0205] when When the image preprocessing module outputs a thumbnail of the adaptive binarized image, then... It is the moiré frequency vector of the thumbnail, according to Compared to the thumbnail side length in pixels The subpixel-level moiré frequency vector of the adaptive binarized image is calculated as follows: .

[0206] Specifically, the sub-pixel moiré frequency vector of the adaptive binarized image... Calculate the moiré features of the adaptive binarized image (including the spatial frequency of the moiré pattern in the adaptive binarized image). and the direction of moiré propagation The specific expression for ) is as follows:

[0207] ;

[0208] ;

[0209] in, and , respectively, are the horizontal and vertical components of the sub-pixel moiré frequency vector of the adaptive binarized image, where c is the pixel unit size of the color filter array. To determine the number of pixels per side of the adaptively binarized image. and These represent the spatial frequency and propagation direction of the moiré pattern in the adaptive binarized image, respectively.

[0210] The temperature measurement module is used to calculate intermediate variables related to temperature based on moiré pattern characteristics, and finally determine the current temperature of the power equipment.

[0211] The temperature measurement module specifically performs the following steps:

[0212] A specified temperature is selected as the reference temperature, and the moiré image captured by the camera at this reference temperature is used as the reference moiré image. The frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane at the reference temperature is calculated using the moiré features extracted from the reference moiré image, and this frequency is used as the reference projection frequency.

[0213] Each time temperature is monitored, the moiré pattern image captured at the current temperature is used as the current moiré pattern image; the frequency of the projection of a one-dimensional periodic grating onto the camera imaging plane at the current temperature is calculated using the moiré pattern features extracted from the current moiré pattern image, and this frequency is used as the current projection frequency.

[0214] Calculate the ratio of the reference projection frequency to the current projection frequency, and use it as the current projection frequency ratio; determine the current temperature based on the mapping relationship between the projection frequency ratio and temperature using the current projection frequency ratio.

[0215] The image preprocessing module preprocesses the reference moiré image (or the current moiré image), while the moiré feature extraction module solves for the moiré features of the adaptive binarized image at the reference temperature (or the current temperature).

[0216] Specifically, the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane... The formula for calculation is:

[0217] ;

[0218] in, For the color filter array frequency, and These represent the spatial frequency and propagation direction of the moiré pattern in the adaptive binarized image, respectively.

[0219] Specifically, the mapping relationship between the projection frequency ratio and temperature is calculated from the physical properties of the elastic material, using the following method:

[0220] The side length of the one-dimensional periodic grating of the moiré visual mark along the propagation direction is pre-calibrated at the reference temperature, and its value is obtained. ;

[0221] The deformation of the one-dimensional periodic grating for moiré visual markings is only affected by piston movement. The volume of the constant-mass sensing medium within the temperature-sensing device at different temperatures is calculated using the PengRob or REFPROP model, thus yielding the volume change of the sensing medium relative to a reference temperature. The positional change of the piston relative to the reference temperature is calculated based on the piston's inner diameter and the volume change of the sensing medium. This positional change of the piston relative to the reference temperature is the deformation of the one-dimensional periodic grating relative to the reference temperature. The deformation of a one-dimensional periodic grating increases as the side length along the propagation direction increases relative to the reference temperature. If the value is positive, the deformation of a one-dimensional periodic grating decreases with respect to the reference temperature as the side length along the propagation direction decreases. It is a negative number;

[0222] Projection frequency ratio and , The relationship is:

[0223] ;

[0224] in, For reference projection frequency, For projection frequencies at other temperatures, The distance from the one-dimensional periodic grating to the camera's imaging plane is [missing information]. The focal length of the camera. and These represent the frequencies of the one-dimensional periodic grating at the reference temperature and at other temperatures, respectively. Let be the number of periods in a one-dimensional periodic grating. and Let be the side lengths of the one-dimensional periodic grating along the propagation direction at the reference temperature and at other temperatures, respectively. The deformation of a one-dimensional periodic grating relative to a reference temperature at other temperatures;

[0225] The deformation of a one-dimensional periodic grating relative to a reference temperature at different temperatures was calculated. The projection frequency ratio at each temperature can be calculated, and finally, a mapping relationship between the projection frequency ratio and temperature can be established.

[0226] The power equipment temperature monitoring in this invention refers to the use of a camera to detect the moiré visual marks in a temperature sensing device that deform as the power equipment temperature changes; it is applicable to scenarios where the relative positions of the camera and the power equipment are fixed; this invention characterizes the temperature change of the power equipment as the change in the ratio of the projection frequency of the one-dimensional periodic grating of the moiré visual mark on the camera imaging plane to the reference projection frequency; this invention can output the real-time temperature of the power equipment being measured.

[0227] This invention also provides a method for monitoring the temperature of power equipment based on moiré patterns, which, based on the above system, includes the following steps:

[0228] 1) Install temperature sensing devices on the surface of electrical equipment to detect temperature changes in the electrical equipment;

[0229] 2) Select a reference temperature and use a camera to capture a moiré pattern image at that reference temperature;

[0230] 3) Process the captured moiré image to obtain an adaptive binarized image and the estimated relative spatial frequency of the moiré pattern in the adaptive binarized image. ;according to With the side length and number of pixels of the adaptive binarized image Whether the ratio reaches 24% determines whether to output an adaptive binarized image or a thumbnail;

[0231] 4) Calculate the sub-pixel level moiré frequency vector of the adaptive binarized image based on the output image, and then obtain the moiré spatial frequency of the adaptive binarized image. and the direction of moiré propagation ;

[0232] 5) According to and The frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane is calculated and used as the reference projection frequency.

[0233] 6) During temperature monitoring, acquire the moiré image captured at the current temperature in real time and repeat steps 3) to 4), and calculate the spatial frequency of the moiré pattern in the adaptive binarized image at the current temperature. and the direction of moiré propagation ;according to and Calculate the frequency of the projection of the one-dimensional periodic grating onto the camera imaging plane at the current temperature, and use it as the current projection frequency; calculate the ratio of the reference projection frequency to the current projection frequency, and use it as the current projection frequency ratio; use the current projection frequency ratio to determine the current temperature based on the mapping relationship between the projection frequency ratio and the temperature.

[0234] This invention has many specific applications. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A moire-based power equipment temperature monitoring system, characterized by, The application relates to a temperature sensing device for power equipment, which comprises a temperature sensing device, an image preprocessing module, a moire feature extraction module and a temperature calculation module. The temperature sensing device is installed on the surface of the power equipment to sense the temperature change of the power equipment; the temperature sensing device comprises a temperature sensing cavity with a piston, a rectangular baffle with a roller, a cuboid shell with a track, a T-shaped pull rod, a fixed rod and a moire visual marker. The temperature sensing cavity with the piston is used for sensing the temperature change of the power equipment and is composed of a rigid cavity and a piston; the rigid cavity is a cuboid structure which is connected with a cylindrical needle barrel-shaped piston; the rigid cavity is filled with a temperature sensing medium with a large thermal expansion coefficient; with the temperature change, the temperature sensing medium in the rigid cavity expands or shrinks, and the volume change of the temperature sensing medium causes the piston to move. The rectangular baffle with the roller is connected with the piston in the temperature sensing cavity with the piston and the vertical end of the T-shaped pull rod and moves along the track in the cuboid shell with the track with the movement of the piston. The fixed rod is designed in a T shape, the vertical end is fixed on the non-hollowed face of the cuboid shell with the track, and the horizontal end is connected with one side of the moire visual marker to ensure that one side of the moire visual marker remains fixed. The moire visual marker is made of an elastic material, one end of the moire visual marker is connected with the fixed rod, and the other end of the moire visual marker is connected with the T-shaped pull rod; the moire visual marker comprises one one-dimensional periodic grating and four positioning markers, and the four positioning markers are respectively located at four corner points of the one-dimensional periodic grating.

2. The moire-based power equipment temperature monitoring system of claim 1, wherein, The image preprocessing module specifically performs the following steps: The R channel image is separated from the central region of interest, the histogram equalization is performed on the R channel image to enhance the image contrast, the frequency domain three-hole mask algorithm is used to remove the noise frequency components of the equalized image, and the self-adaptive binary image is obtained through self-adaptive binaryzation of the image with the noise frequency components removed. The frequency domain three-hole mask algorithm specifically comprises: The cuboid shell with track includes five of six surfaces in hollow design, the surface in vertical direction and connected with fixed rod is not in hollow design, four edges of the cuboid shell in horizontal direction are provided with tracks for limiting the rectangular baffle with roller to move along the fixed track direction, and the track is parallel to the X axis direction of object coordinate system ​ A T-shaped pull rod, whose vertical end is connected with a rectangular baffle with a roller, and whose horizontal end is connected with one side of a moire visual mark; it moves with the movement of the rectangular baffle with the roller, and then pulls the moire visual mark connected therewith, so that the moire visual mark is deformed along the X-axis direction of the object coordinate system by being pulled ; The specific method for determining the optimal thumbnail length by using the discrete search strategy is as follows: In the final search interval, the Scharr operator and the Laplacian operator joint search algorithm are used to determine the optimal thumbnail length.

3. The moire-based power equipment temperature monitoring system of claim 2, wherein, The specific process of the temperature sensing device converting temperature change into change of moire features is as follows: as the temperature of the power equipment rises or falls, the temperature sensing medium in the rigid cavity of the temperature sensing cavity with a piston expands or shrinks; the expansion or shrinkage of the temperature sensing medium generates a pushing force or a suction force on the piston, causing the piston to move; the movement of the piston drives the rectangular baffle with a roller connected to the piston to move along the track on the cuboid shell with a track; the movement of the rectangular baffle with a roller drives the T-shaped pull rod connected thereto to move; the T-shaped pull rod pulls the moire visual marker connected thereto, so that the moire visual marker is pulled to deform along the X-axis direction of the object coordinate system ; the deformation of the moire visual marker along the X-axis direction of the object coordinate system causes the frequency of the one-dimensional periodic grating to change, further causing the frequency of the projection of the one-dimensional periodic grating on the camera imaging plane to change, and finally causing the moire features to change.

4. The moire-based power equipment temperature monitoring system of claim 2, wherein, The temperature calculation module specifically performs the following steps: According to the positioning of the positioning marks, four corner points of the one-dimensional periodic grating are positioned, and then a region where the one-dimensional periodic grating is located is cropped from the photographed moire image as a rough selection region of interest, and the number of pixels of the side length of the one-dimensional periodic grating perpendicular to the propagation direction is obtained ; The edge of the rough selection region of interest is cropped to obtain a central region of interest with a side length of to eliminate the influence of unstable features in the edge region on the accuracy; According to the number of pixels of the side length of the center region of interest Estimating the relative spatial frequency of the moire pattern of the center region of interest image with the positioning mark in the captured moire pattern image ; A specified temperature is selected as a reference temperature, the moire image taken by the camera at the reference temperature is taken as a reference moire image, the frequency of the projection of the one-dimensional periodic grating on the camera imaging plane at the reference temperature is calculated by using the moire feature extracted from the reference moire image, and the frequency is taken as a reference projection frequency; The adaptive binarized image corresponds to the central region of interest in the captured moiré pattern image. It is both the estimated relative spatial frequency of moiré patterns in the central region of interest image and the estimated relative spatial frequency of moiré patterns in the adaptive binarized image. It is both the side length in pixels of the central region of interest and the side length in pixels of the adaptively binarized image; according to and The ratio determines the subsequent image processing used: when When, use the adaptive binarized image as the output; when If the moiré stripes are too dense, a discrete search strategy is used to determine the optimal thumbnail side length. The adaptive binarized image is then downsampled according to the optimal thumbnail side length to generate a thumbnail, which is then used as the output.

5. The moire-based power equipment temperature monitoring system of claim 4, wherein, The length of the side of the center region of interest in pixels Estimating the spatial frequency of the moire pattern in the center region of interest image from the positioning mark in the captured moire pattern image Specifically: According to the positioning mark in the photographed moire image, four corner points of the one-dimensional periodic grating region in the photographed moire image are positioned, and then the pixel number of the edge length of the one-dimensional periodic grating along the propagation direction is obtained and the tilt angle of the X-axis positive direction of the edge thereof along the propagation direction relative to the image coordinate system of the image coordinate system ; According to the number of pixels of the edge length of the one-dimensional periodic grating along the propagation direction , the tilt angle , and the number of pixels of the edge length of the central region of interest to estimate the relative spatial frequency of the moire of the central region of interest image The expression is as follows: ； wherein, is a pixel cell size of the color filter array; is a color filter array frequency, ; is a number of periods contained in the one-dimensional periodic grating.

6. The moire-based power equipment temperature monitoring system of claim 5, wherein, In each temperature monitoring, the moire image taken at the current temperature is taken as a current moire image. performing a fast Fourier transform on the equalized image to obtain an original amplitude spectrum and a phase spectrum ; copying the original magnitude spectrum , obtaining a copy thereof ; creating a mask of the same size as the original magnitude spectrum , with all values being zero ; In the copy Find the pixel with the largest amplitude value in the mask, which corresponds to the DC pulse; Lieutenant General centered on DC pulse Change the value of the region to 1, and set the copy... The same in The value of the region is changed to 0, indicating that DC pulses will no longer be searched for. Finding a copy For the pixel with the largest amplitude value, calculate the relative spatial frequency of that pixel and compare it with the estimated relative spatial frequency of the moiré pattern in the central region of interest image. The absolute value of the difference; if the absolute value of the difference does not exceed a predefined threshold, then the pixel is considered to correspond to a moiré frequency component, and in the mask... The lieutenant general centered on that pixel Change the value of the region to 1, and set the copy... Centered on this pixel The value of the region is changed to 0, indicating that the pixel has already been found and will not be searched again; Otherwise, discard the pixel, determine that the pixel is not a moiré frequency component, and only copy it. Centered on this pixel Change the value of the region to 0, indicating that the pixel will not be considered in subsequent searches; repeat the above operation until a pair of moiré frequency components are found; The above steps yield a result that is only three. A mask where the value is 1 in the specified area and 0 in all other pixels. ; mask Applied to the original amplitude spectrum This yields an amplitude spectrum with noise frequency components removed. Then, based on the amplitude spectrum after removing noise frequency components... Phase spectrum The spatial domain image is restored using inverse Fourier transform; ​ at a sampling rate down-sampling the adaptively binarized image to generate a thumbnail; the sampling rate is equal to the number of pixels of the side length of the thumbnail of the adaptively binarized image; and the ratio of the number of pixels of the side length of the adaptively binarized image. The operation of thumbnailing an adaptive binarized image is equivalent to superimposing a grid with spatial frequency of 1 / 2 onto the adaptive binarized image; Moiré relative spatial frequency of an adaptive binarized image , spatial frequency of the superimposed grid , thumbnail side length in pixels to estimate the moiré relative spatial frequency of the thumbnail , expressed as follows: ； Because , the upper bound of the search range is ; With the spatial frequency of the superimposed grid the moire opposing spatial frequency of the adaptive binarized image the decision condition for proximity is then: and the lower bound of the search range is determined as ; The search range of the optimal thumbnail side length is a discrete interval ; the estimated relative spatial frequency of moire of the adaptive binarization image The search range is brought into the specific discrete search interval ; a certain estimated fault tolerance interval is extended to the right side of the specific discrete search interval, and the final search interval of the optimal thumbnail side length pixel number is , The estimated fault tolerance interval length ​ 7. The moire-based power equipment temperature monitoring system of claim 6, wherein, ​ The input image is an image output by the image preprocessing module or a thumbnail corresponding to a maximum value point in a Scharr operator and Laplacian operator joint search algorithm, and a Gaussian window and a fast Fourier transform are used on the input image to obtain an amplitude spectrum corresponding to the input image ; Finding the amplitude spectrum The point with the largest amplitude value in the amplitude spectrum corresponds to the DC impulse; an amplitude spectrum coordinate system OUV is established, with the DC impulse as the origin, the horizontal and vertical axes as the horizontal right direction and vertical up direction of the amplitude spectrum respectively; each pixel in the amplitude spectrum corresponds to a frequency component, and the vector from the origin to other pixels in the amplitude spectrum is called the frequency vector of the frequency component corresponding to the pixel. finding the amplitude spectrum the pixel with the largest amplitude value except for the direct current impulse , corresponds to the pixel-level moire frequency component; the sub-pixel level moire frequency vector of the input image is calculated using the method based on the Gaussian function, and if the method based on the Gaussian function has no solution, the method based on the dynamic weighted fast Fourier transform is used to calculate the sub-pixel level moire frequency vector of the input image; the calculated sub-pixel level moire frequency vector of the input image is , wherein and are the horizontal axis component and the vertical axis component of the sub-pixel level moire frequency vector of the input image respectively; When the input image is a thumbnail corresponding to a maximum point in a Scharr operator and Laplacian operator joint search algorithm, the moire relative spatial frequency of the input image is used with the direction of propagation as output; When the input image is the image output by the image preprocessing module, according to With determining a sub-pixel level moire frequency vector of the adaptive binarized image , and according to calculating the moire features of the adaptive binarized image as output, including the moire spatial frequency of the adaptive binarized image and the moire propagation direction .

8. The moire-based power equipment temperature monitoring system of claim 7, wherein, ​ ​ ​ The frequency of the projection of the one-dimensional periodic grating on the camera imaging plane at the current temperature is calculated by using the moire features extracted from the current moire image, and is taken as the current projection frequency; The ratio of the reference projection frequency to the current projection frequency is calculated as the current projection frequency ratio; and the current temperature is determined according to the mapping relationship between the projection frequency ratio and the temperature by using the current projection frequency ratio.

9. The moire-based power equipment temperature monitoring system of claim 8, wherein, a frequency of a projection of the one-dimensional periodic grating on an imaging plane of the camera The calculation formula is: ； wherein, is the color filter array frequency, and are the moire spatial frequency and moire propagation direction of the adaptively binarized image, respectively.

10. A moire-based power equipment temperature monitoring method based on the system of any of claims 1-9, characterized by, The method comprises the following steps: 1) setting a temperature sensing device on the surface of the power equipment to sense the temperature change of the power equipment; 2) selecting a reference temperature, and shooting a moire image at the reference temperature by using a camera; 3) processing the captured moire image to obtain an adaptive binary image and an estimated relative spatial frequency of the moire image of the adaptive binary image ; according to whether the ratio of the number of pixels of the side length of the adaptive binary image reaches 24% determines whether to output the adaptive binary image or the thumbnail image; 4) calculating the sub-pixel level moire frequency vector of the adaptive binarized image from the output image, and further obtaining the moire spatial frequency of the adaptive binarized image and the moire propagation direction ; 5) according to With The frequency of the projection of the one-dimensional periodic grating on the imaging plane of the camera is calculated as a reference projection frequency; 6) Real-time acquisition of the moire image of the current temperature shot during temperature monitoring and repetition of steps 3) to 4) to calculate the moire spatial frequency of the adaptive binary image at the current temperature and the moire propagation direction ; according to and the frequency of the one-dimensional periodic grating projected on the camera imaging plane at the current temperature is calculated as the current projection frequency; The ratio of the reference projection frequency to the current projection frequency is calculated as the current projection frequency ratio; and the current temperature is determined according to the mapping relationship between the projection frequency ratio and the temperature by using the current projection frequency ratio.

Images