A method and system for automatically identifying faults of photovoltaic equipment

Abstract

The invention relates to the field of photovoltaic equipment, and in particular to a method and system for automatically identifying faults of photovoltaic equipment. In order to remove the complex outdoor background image in the initial image, the invention uses regularity and area growth rate to further identify the smoke feature, and then extracts the minimum convex polygon of the smoke feature contour line to obtain the standard smoke feature in order to conveniently establish a regular smoke skeleton, thereby establishing a smoke skeleton and calculating the smoke width of the standard smoke feature at any pixel point on the smoke skeleton, and inferring the smoke source through the smoke width and the smoke skeleton, so as to output a number of devices close to the smoke source according to the smoke source, so as to identify the faulty device and its coordinates, and output them for a robot to quickly reach the corresponding position for processing. The invention solves the problem that when the device in a photovoltaic power station that produces fire smoke due to a fault is blocked by other devices, it may be difficult for a camera to accurately know the faulty device and its position.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic equipment, and in particular to a method and system for automatically identifying faults of photovoltaic equipment. Background Art

[0002] With the advancement of unmanned or few-person photovoltaic power stations, robots are required to be able to handle some emergencies or related maintenance of the power station in a timely manner in addition to daily inspection work, such as cleaning dust on the surface of equipment or extinguishing fires.

[0003] At present, power stations are often monitored through a combination of robots and pan-tilt cameras, but both have certain defects. Robots cannot achieve 24-hour monitoring of specific areas due to the need to charge and the small number of robots. Although pan-tilt cameras can meet 24-hour monitoring requirements, they are difficult to move. Therefore, when a fire occurs due to a malfunction of the equipment, the equipment may be blocked by other equipment, making it difficult for the pan-tilt camera to accurately identify the faulty equipment and its location information when it identifies the fire smoke caused by the malfunction, and it is difficult to send the location information to the processing robot in time. Therefore, it is difficult for the processing robot to reach the location of the faulty equipment in a timely and accurate manner, which can easily expand the danger. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method and system for automatically identifying faults of photovoltaic equipment, which solves the problem that when the equipment in a photovoltaic power station that produces fire smoke due to a fault is blocked by other equipment, it may be difficult for the camera to accurately know the faulty equipment and location.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for automatically identifying faults of photovoltaic equipment, the method comprising the following steps:

[0007] S1, respectively obtaining a plurality of first images and second images of the same area of ​​the corresponding photovoltaic power station including the current moment, and marking the first images and the second images as initial images;

[0008] S2, filtering the initial image for noise to obtain a grayscale image, extracting a number of contrast images from the grayscale image according to a contrast period, and identifying smoke features according to the regularity and area growth rate of the moving targets in the number of contrast images;

[0009] S3, establishing the minimum convex polygon of the smoke feature contour line in the comparison image at the current moment, and marking the convex polygon as a standard smoke feature;

[0010] S4. extracting the smoke skeleton according to the standard smoke feature in the comparison image at the current moment, calculating the smoke width of the standard smoke feature along the smoke skeleton, and calculating the coordinates of the smoke source where the smoke width is 0;

[0011] S5, setting a distance threshold, and determining whether there is a photovoltaic device whose distance from the smoke source is within the distance threshold;

[0012] If yes, the photovoltaic device is marked as a faulty device, and the coordinates of the faulty device are output and the process ends;

[0013] If not, proceed to step S6;

[0014] S6. Predict the coordinates of the intersection of the smoke skeleton and the ground based on the coordinates of each point on the smoke skeleton and the three-dimensional topographic map of the photovoltaic power station, mark several photovoltaic devices closest to the intersection coordinates as faulty devices, and output the coordinates of the faulty devices.

[0015] Preferably, in step S2, the following steps are specifically included:

[0016] S21, converting a plurality of initial images into grayscale images, and removing noise from the current grayscale images;

[0017] S22, setting a comparison period, and obtaining a grayscale image whose image generation time is the comparison period from a plurality of grayscale images, and marking it as a comparison image;

[0018] S23, identifying a moving target in the comparison image and marking it as a suspicious target;

[0019] S24, obtaining the contour length and area of ​​the suspicious target, and calculating its regularity; the calculation formula of the regularity is:

[0020]

[0021] In the above formula, N represents the regularity, C represents the contour length of the suspicious target, and S represents the area of ​​the suspicious target;

[0022] S25. Calculate the area growth rate of the suspicious target according to its area in the plurality of comparison images; the area growth rate calculation formula is:

[0023]

[0024] In the above formula, ΔS i S represents the area growth rate of the suspicious target in the ith comparison image compared with the comparison image at the previous adjacent moment. i represents the area of ​​the suspicious target in the ith comparison image, S i-1It represents the area of ​​the suspicious target in the i-1th contrast image, and the i-1th contrast image is at an adjacent moment before the i-th contrast image;

[0025] S26, judging whether the suspicious target is a smoke feature according to whether the regularity of the suspicious target in the comparison image at the current moment is greater than the regularity threshold, and the area growth rate of each comparison image is greater than 0;

[0026] If so, the suspected target is a smoke feature;

[0027] If not, the suspicious target is not a smoke feature.

[0028] Preferably, in step S21, the following steps are specifically included:

[0029] S211, converting a plurality of initial images into grayscale images;

[0030] S212, setting a pixel value change threshold, and using the pixel value of each grayscale image at the same pixel point as the median, establishing a pixel value interval having a size equal to the pixel value change threshold;

[0031] S213, counting the number of pixel values ​​contained in each pixel value interval according to the pixel value of the same pixel point in the plurality of grayscale images;

[0032] S214, calculating the contribution coefficient of the same pixel point of each grayscale image to the pixel value according to the number of pixel values ​​in the pixel value interval; the calculation formula of the contribution coefficient is:

[0033]

[0034] In the above formula, Represents the contribution coefficient of the same pixel point in the i-th grayscale image to the pixel value, N i represents the number of pixel values ​​in the pixel value interval corresponding to the same pixel point in the i-th grayscale image, and m represents the total number of pixel value intervals corresponding to the same pixel point, that is, the number of grayscale images used for calculation;

[0035] S215, updating the pixel value of any pixel in the current grayscale image according to the contribution coefficient of the same pixel of each grayscale image to the pixel value, so as to obtain a first grayscale image, where m represents the number of grayscale images used for calculation;

[0036]

[0037] In the above formula, I0 represents the pixel value of any pixel in the current grayscale image after update. It represents the contribution coefficient of any pixel in the i-th grayscale image to the pixel value, I irepresents the pixel value of the corresponding pixel in the i-th grayscale image, and m represents the number of grayscale images used for calculation.

[0038] Preferably, in step S3, the following steps are specifically included:

[0039] S31, establishing a tangent segment of a rated length at any pixel point on the smoke characteristic contour line, the tangent segment being centered on the pixel point, and the length of the tangent segment being preferably set so as not to penetrate both sides of the smoke characteristic contour line;

[0040] S32, judging whether the pixel point corresponding to the tangent segment is an inner arc point according to whether the tangent segment is located within the smoke feature;

[0041] If so, the pixel point corresponding to the tangent segment is the inner arc point;

[0042] If not, the pixel point corresponding to the tangent segment is an outer arc point;

[0043] S33, marking the contour lines where several adjacent inner arc points are located as inner contour lines, and marking the contour lines where several adjacent outer arc points are located as outer contour lines;

[0044] S34, sequentially acquiring two outer circle contour lines adjacent to each inner circle contour line, and establishing a first adjacent line tangent to both outer circle contour lines;

[0045] S35, connecting two ends of the inner circle contour line adjacent to only one outer circle contour line through a second adjacent line;

[0046] S36, extracting the contour line with the largest area enclosed by the first adjacent line, the second adjacent line and the smoke feature contour line to obtain a minimum convex polygon, and using the minimum convex polygon as a standard smoke feature.

[0047] Preferably, in step S4, the following steps are specifically included:

[0048] S41, extracting a smoke skeleton of standard smoke features;

[0049] S42, respectively obtaining the coordinates of the shooting device used to shoot the two comparison images at the current moment, the lens deflection angle and the viewing angle to generate device data;

[0050] S43, establishing a height line according to the altitude, and marking the height of each point on the smoke skeleton;

[0051] S44, calculating the coordinates of each point on the smoke skeleton according to the device data, the height of each point on the smoke skeleton and the two comparison images at the current moment;

[0052] S45, calculating the width of the standard smoke feature at each pixel point on the smoke skeleton;

[0053] S46, identifying and removing abnormal pixels on the standard smoke feature according to the width change rate of each pixel on the smoke skeleton of the standard smoke feature, so as to obtain a first smoke feature;

[0054] S47. According to the first smoke feature and the width at each point on the smoke skeleton, when the width is predicted to be 0, the coordinates of the pixel point corresponding to the smoke skeleton are the coordinates of the smoke source.

[0055] Preferably, in step S44, the following steps are specifically included:

[0056] S441, establishing a first connection line between two shooting devices according to the device data;

[0057] S442, establishing a second connection line between the camera lens and any height pixel point on the smoke skeleton;

[0058] S443, calculating a first horizontal angle between the projection of the second connecting line and the normal line of the camera lens on the horizontal plane; the calculation formula of the first horizontal angle is:

[0059]

[0060] In the above formula, α1 represents the first horizontal angle, ΔN represents the number of pixels from any height pixel point on the smoke skeleton to the vertical line of the center of the comparison image, N0 is the number of pixels in any row of the comparison image, and α0 represents the viewing angle of the shooting device;

[0061] S444, obtaining a second horizontal angle between the projection of the normal line of the camera lens and the first connecting line on the horizontal plane according to the lens deflection angle;

[0062] S445. Calculate a third horizontal angle of projection of the second connecting line and the first connecting line on the horizontal plane according to the second horizontal angle and the first horizontal angle; the calculation formula of the third horizontal angle is:

[0063] α3=α1+α2

[0064] In the above formula, α3 represents the third horizontal angle, α1 and α2 represent the first horizontal angle and the second horizontal angle respectively. When the pixel point on the smoke skeleton corresponding to the second connecting line is located on the side of the lens normal line close to the other shooting device, the value of the first horizontal angle is taken as the opposite number;

[0065] S446. Calculate the coordinates of each point on the smoke skeleton according to the third horizontal angle and the first connecting line.

[0066] Preferably, in step S446, the following steps are specifically included:

[0067] S4461, projecting pixel points at any height on the shooting device and the smoke skeleton onto a horizontal plane;

[0068] S4462, marking the projection points of the two shooting devices on the horizontal plane as point A and point B respectively, marking the projection point of a pixel point at any height on the smoke skeleton on the horizontal plane as point C, marking the second connecting line directly opposite to point A as line segment a, marking the second connecting line directly opposite to point B as line segment b, and marking the first connecting line as line segment c;

[0069] S4463, calculating the pixel coordinates corresponding to the projection point on the smoke skeleton; the pixel coordinates corresponding to the projection point C on the smoke skeleton satisfy the following relationship:

[0070]

[0071] In the above formula, represents the angle at vertex A, that is, the third horizontal angle at vertex A, represents the angle at vertex B, that is, the third horizontal angle at vertex B, A x and A y are the x-axis coordinate and y-axis coordinate of point A, B x and B y are the x-axis coordinate and y-axis coordinate of point B, C x and C y are the x-axis coordinates and y-axis coordinates of point C, a, b and c are the lengths of line segments a, b and c, respectively. h0 represents the height of the pixel corresponding to the projection point C on the smoke skeleton. C z Indicates the z-axis coordinate of the pixel.

[0072] Preferably, in step S45, the following steps are specifically included:

[0073] S451, establishing radial lines perpendicular to the tangent lines of each point on the smoke skeleton;

[0074] S452, obtaining coordinates of two intersection points where the radial line intersects with the contour line of the standard smoke feature, the two intersection points being located on both sides of the smoke skeleton, and having the smallest vertical distance from the radial line;

[0075] S453. Calculate the width of the standard smoke feature at each point on the smoke skeleton according to the coordinates of the two intersection points; the calculation formula of the width is:

[0076]

[0077] In the above formula, w represents the width of any point on the smoke skeleton, E x 、E y and E zThe x-coordinate, y-coordinate and z-coordinate of one of the intersection points of the radial line and the contour line of the standard smoke feature are represented respectively. x , D y and D z The x-coordinate, y-coordinate and z-coordinate represent the intersection points of the radial line and the contour line of the standard smoke feature.

[0078] Preferably, in step S46, the following steps are specifically included:

[0079] S461, calculating the width change rate of the width at any pixel point on the smoke skeleton relative to the width at the adjacent pixel point on the lower side; the calculation formula of the width change rate is:

[0080]

[0081] In the above formula, Δw i It represents the width change rate of any pixel on the smoke skeleton relative to the width of the adjacent pixel on the lower side, w i and w i-1 They represent the width of any pixel on the smoke skeleton and the width of the adjacent pixel on the lower side of the smoke skeleton;

[0082] S462. Delete pixel points on the standard smoke feature corresponding to pixel points on the smoke skeleton whose width change rate is greater than a width change rate threshold, to obtain a first smoke feature.

[0083] The technical solution also provides a system for implementing a method for automatically identifying faults of photovoltaic equipment. The system includes a processor and a memory. The memory is used to store a computer program. When the computer program is executed by the processor, the method for automatically identifying faults of photovoltaic equipment is implemented.

[0084] Compared with the prior art, the present invention provides a method and system for automatically identifying faults of photovoltaic equipment, which has the following beneficial effects:

[0085] 1. In order to remove the complex outdoor background image in the initial image, the present invention uses regularity and area growth rate to further identify the smoke features. Then, in order to facilitate the establishment of a regular smoke skeleton, the minimum convex polygon of the smoke feature contour line is extracted to obtain the standard smoke feature, thereby establishing a smoke skeleton and calculating the smoke width of the standard smoke feature at any pixel point on the smoke skeleton, and inferring the smoke source through the smoke width and the smoke skeleton, so as to output several devices close to the smoke source according to the smoke source, so as to identify the faulty device and its coordinates, and output them for the robot to quickly reach the corresponding position for processing.

[0086] 2. The present invention sets a contrast period to extract a contrast image, and can better distinguish smoke features from other features in the contrast image according to the morphological change period of the smoke, and further better identify the smoke features by calculating the regularity and area growth rate.

[0087] 3. The present invention achieves the purpose of distinguishing inner arc points and outer arc points by establishing a tangent segment with a rated length at any pixel point on the smoke feature contour line, thereby achieving the purpose of dividing the smoke feature contour line into a plurality of inner circle contour lines and outer circle contour lines, and finally extracting the minimum convex polygon by establishing the first adjacent line and the second adjacent line to obtain a standard smoke feature with a relatively regular shape.

[0088] 4. In order to facilitate the calculation of the coordinates of pixel points on the standard smoke feature, the present invention establishes a smoke skeleton, takes the altitude as a reference, and calculates the coordinates of each point on the smoke skeleton by establishing a triangle, so that the smoke skeleton can reflect the flow direction of the smoke more clearly.

[0089] 5. The present invention calculates the width of the smoke and combines it with the smoke skeleton to determine the abnormal pixel points based on the width change rate of the smoke width, thereby identifying the position where the smoke is blocked by the equipment, thereby eliminating its influence on the calculation of the smoke source, so that the coordinates of the smoke source can be calculated more accurately, thereby identifying the blocked faulty equipment and the equipment coordinates. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0091] Figure 1 A flow chart of a method for automatically identifying faults of photovoltaic equipment according to the present invention;

[0092] Figure 2 is a schematic diagram of smoke characteristics of the present invention;

[0093] Figure 3 A schematic diagram of a first adjacent line and a second adjacent line on the smoke feature of the present invention;

[0094] Figure 4 Schematic diagram of the standard smoke characteristics and smoke skeleton of the present invention. DETAILED DESCRIPTION

[0095] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods, so that the implementation process of how the present application uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0096] Those skilled in the art can understand that all or part of the steps in the following embodiments can be completed by instructing the relevant hardware through a program, so the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0097] In order to solve the problem that when a device in a photovoltaic power station that generates fire smoke due to a fault is blocked by other devices, it may be difficult for a camera to accurately know the faulty device and its location, which makes it difficult for a fire-fighting robot to receive corresponding location information and arrive at the fault point in time and accurately to extinguish the fire, the present invention provides a method for automatically identifying faults of photovoltaic devices, which is used to identify and locate the faulty device through fire smoke when the faulty device is blocked by other devices, so as to improve the automatic processing capability of the photovoltaic power station when encountering a fire, such as Figure 1 As shown, the method comprises the following steps:

[0098] S1. Obtain a plurality of first images and second images of the same area of ​​the corresponding photovoltaic power station at the current moment, and mark the first image and the second image as initial images. The first image and the second image may be obtained by a shooting device such as a pan / tilt camera set in the photovoltaic power station. At least two initial images obtained from different positions are used to facilitate more accurate positioning of smoke.

[0099] S2, filtering the initial image for noise to obtain a grayscale image, extracting several contrast images from the grayscale image according to the contrast period, and identifying the smoke features according to the regularity and area growth rate of the moving targets in the several contrast images; since there are a large number of changing background images in the wild, in order to accurately identify the smoke features, firstly, through actual experiments, understand the dynamic change law of smoke outdoors, and then design a corresponding algorithm according to the dynamic change law to separate it from the outdoor scene, so as to accurately identify the smoke features in the initial image. In step S2, the following steps are specifically included:

[0100] S21, converting several initial images into grayscale images, and removing noise in the current grayscale image; since there are many interference sources outdoors, there may be noise in the initial grayscale image obtained. In order to reduce the effect of noise on the recognition of smoke features, step S21 specifically includes the following steps:

[0101] S211, converting a plurality of initial images into grayscale images;

[0102] S212, obtaining pixel values ​​of the same pixel point in a plurality of grayscale images, extracting the median of the pixel values, and marking the median pixel value. The plurality of grayscale images need to correspond to the same shooting device to ensure the consistency of the scene, so two groups of grayscale images need to be obtained respectively, and the two groups of grayscale images correspond to two shooting devices respectively;

[0103] S212, setting a pixel value change threshold, and using the pixel value of each grayscale image at the same pixel point as the median, establishing a pixel value interval having a size equal to the pixel value change threshold;

[0104] S213, counting the number of pixel values ​​contained in each pixel value interval according to the pixel value of the same pixel point in the plurality of grayscale images;

[0105] S214, calculating the contribution coefficient of the same pixel point of each grayscale image to the pixel value according to the number of pixel values ​​in the pixel value interval; the calculation formula of the contribution coefficient is:

[0106]

[0107] In the above formula, Represents the contribution coefficient of the same pixel point in the i-th grayscale image to the pixel value, N i represents the number of pixel values ​​in the pixel value interval corresponding to the same pixel point in the i-th grayscale image, and m represents the total number of pixel value intervals corresponding to the same pixel point, that is, the number of grayscale images used for calculation;

[0108] S215, updating the pixel value of any pixel in the current grayscale image according to the contribution coefficient of the same pixel of each grayscale image to the pixel value, so as to obtain a first grayscale image, where m represents the number of grayscale images used for calculation;

[0109]

[0110] In the above formula, I0 represents the pixel value of any pixel in the current grayscale image after update. It represents the contribution coefficient of any pixel in the i-th grayscale image to the pixel value, I iRepresents the pixel value of the corresponding pixel in the i-th grayscale image, and m represents the number of grayscale images used for calculation. Through this method, the noise points that appear for the first time with a sudden change in pixel value can be suppressed. However, once there are many pixel points with a sudden change in pixel value, the contribution coefficient can increase rapidly, and the feature can be retained, thereby avoiding the removal of smoke features.

[0111] S22, setting a comparison period, and obtaining a grayscale image whose image generation time is the comparison period from a plurality of grayscale images, and marking it as a comparison image. Since the frequency of image capture by the shooting device is high, in order to increase the distinction between images at different times, the comparison period is set so that the features in the selected comparison image can change during the comparison period, so as to expand the variation of the smoke feature, thereby more accurately distinguishing the smoke feature. The comparison period is generally set to 1s to 10s. When it is set to 2s in actual application, it can be better distinguished from other features, the probability of missing the smoke feature is relatively the lowest, and the probability of correctly identifying the smoke feature is relatively the highest;

[0112] S23, identifying a moving target in the comparison image and marking it as a suspicious target;

[0113] S24, obtaining the contour length and area of ​​the suspicious target, and calculating its regularity; the calculation formula of the regularity is:

[0114]

[0115] In the above formula, N represents regularity, C represents the contour length of the suspicious target, and S represents the area of ​​the suspicious target. The shape of the smoke feature is relatively complex. In order to accurately identify it, the circle is used as a reference. The regularity of the circle is 1. The closer it is to a circle, the closer its regularity is to 1. Through experiments, the regularity of different smoke features at different stages is calculated, and the regularity results are: 2.51, 2.18, 1.96, 3.21, 2.89, 2.15, 2.43, 2.36, 2.12, 2.21, 3.18, 2.39. Therefore, when setting the regularity threshold later, it is generally set between 1.8 and 1.9.

[0116] S25. Calculate the area growth rate of the suspicious target according to its area in the plurality of comparison images; the area growth rate calculation formula is:

[0117]

[0118] In the above formula, ΔS i S represents the area growth rate of the suspicious target in the ith comparison image compared with the comparison image at the previous adjacent moment. i represents the area of ​​the suspicious target in the ith comparison image, S i-1It represents the area of ​​the suspicious target in the i-1th contrast image. The i-1th contrast image is at an adjacent moment on the i-th contrast image. This is because, in the early stage when the equipment generates fire smoke due to a failure, the range of the smoke is constantly expanding during the rising process due to the expansion of the fire, so its area is in a state of continuous growth.

[0119] S26, judging whether the suspicious target is a smoke feature according to whether the regularity of the suspicious target in the comparison image at the current moment is greater than the regularity threshold, and the area growth rate of each comparison image is greater than 0;

[0120] If so, the suspected target is a smoke feature;

[0121] If not, the suspicious target is not a smoke feature. The combination of the two features can be used to determine whether it is a smoke feature, which is more accurate.

[0122] S3. Establish the minimum convex polygon of the smoke feature contour line in the comparison image at the current moment, and mark the convex polygon as the standard smoke feature. In order to use the contour of the smoke feature to subsequently infer the coordinates of the smoke source, it is necessary to regularize the smoke features of complex shapes, so as to facilitate the sorting out of the extension trend of the smoke features, and thus infer the smoke source, such as Figure 2 , Figure 3 and Figure 4 As shown, in step S3, the following steps are specifically included:

[0123] S31, establishing a tangent segment with a rated length at any pixel point on the smoke characteristic contour line, with the pixel point as the center;

[0124] S32, judging whether the pixel point corresponding to the tangent segment is an inner arc point according to whether the tangent segment is located within the smoke feature;

[0125] If so, the pixel point corresponding to the tangent segment is the inner arc point;

[0126] If not, the pixel point corresponding to the tangent segment is an outer arc point;

[0127] S33, marking the contour lines where several adjacent inner arc points are located as inner contour lines, and marking the contour lines where several adjacent outer arc points are located as outer contour lines;

[0128] S34, sequentially acquiring two outer circle contour lines adjacent to each inner circle contour line, and establishing a first adjacent line tangent to both outer circle contour lines;

[0129] S35, connecting two ends of the inner circle contour line adjacent to only one outer circle contour line through a second adjacent line;

[0130] S36, extracting the contour line with the largest area enclosed by the first adjacent line, the second adjacent line and the smoke feature contour line to obtain a minimum convex polygon, and using the minimum convex polygon as a standard smoke feature.

[0131] S4. Extract the smoke skeleton according to the standard smoke feature in the comparison image at the current moment, calculate the smoke width of the standard smoke feature along the smoke skeleton, and calculate the coordinates of the smoke source where the smoke width is 0; because the shapes of the smoke cross sections captured by the two gimbal cameras may be different when capturing smoke images in the same area, in order to calculate the coordinate value of a certain position on the smoke feature, it is not possible to directly locate a certain position on the smoke feature, and it is necessary to correspond each position on the smoke feature by altitude, so that the coordinate value at a certain height on the smoke coordinate can be calculated by comparing the images corresponding to the two initial images captured from different angles. In step S4, the following steps are specifically included:

[0132] S41, extracting a smoke skeleton of standard smoke features, where the smoke skeleton is used to comprehensively reflect the extension direction of the smoke. Since the smoke tends to rise, at any height, there is generally one pixel point on the smoke skeleton;

[0133] S42, respectively obtaining the coordinates of the shooting device used to shoot the two comparison images at the current moment, the lens deflection angle and the viewing angle to generate device data;

[0134] S43, establishing a height line according to the altitude, and marking the height of each point on the smoke skeleton;

[0135] S44, calculating the coordinates of each point on the smoke skeleton according to the equipment data, the height of each point on the smoke skeleton and the two comparison images at the current moment; when the height value of any point on the smoke skeleton is determined, in order to more conveniently calculate the coordinates of each point on the smoke skeleton, it is necessary to construct a graph for calculation, in step S44, specifically including the following steps:

[0136] S441, establishing a first connection line between two shooting devices according to the device data;

[0137] S442, establishing a second connection line between the camera lens and any height pixel point on the smoke skeleton;

[0138] S443, calculating a first horizontal angle between the projection of the second connecting line and the normal line of the camera lens on the horizontal plane; the calculation formula of the first horizontal angle is:

[0139]

[0140] In the above formula, α1 represents the first horizontal angle, ΔN represents the number of pixels from any height pixel point on the smoke skeleton to the vertical line of the center of the comparison image, N0 is the number of pixels in any row of the comparison image, and α0 represents the viewing angle of the shooting device;

[0141] S444, obtaining a second horizontal angle between the projection of the normal line of the camera lens and the first connecting line on the horizontal plane according to the lens deflection angle;

[0142] S445. Calculate a third horizontal angle of projection of the second connecting line and the first connecting line on the horizontal plane according to the second horizontal angle and the first horizontal angle; the calculation formula of the third horizontal angle is:

[0143] α3=α1+α2

[0144] In the above formula, α3 represents the third horizontal angle, α1 and α2 represent the first horizontal angle and the second horizontal angle respectively. When the pixel point on the smoke skeleton corresponding to the second connecting line is located on the side of the lens normal line close to the other shooting device, the value of the first horizontal angle is taken as the opposite number;

[0145] S446, calculating the coordinates of each point on the smoke skeleton according to the third horizontal angle and the first connecting line; the figure actually constructed is a triangle, and a method for calculating the coordinates of any point on the smoke skeleton by trigonometric functions is provided here, and in step S446, the following steps are specifically included:

[0146] S4461, projecting pixel points at any height on the shooting device and the smoke skeleton onto a horizontal plane;

[0147] S4462, marking the projection points of the two shooting devices on the horizontal plane as point A and point B respectively, marking the projection point of a pixel point at any height on the smoke skeleton on the horizontal plane as point C, marking the second connecting line directly opposite to point A as line segment a, marking the second connecting line directly opposite to point B as line segment b, and marking the first connecting line as line segment c;

[0148] S4463, calculating the pixel coordinates corresponding to the projection point on the smoke skeleton; the pixel coordinates corresponding to the projection point C on the smoke skeleton satisfy the following relationship:

[0149]

[0150] In the above formula, represents the angle at vertex A, that is, the third horizontal angle at vertex A, represents the angle at vertex B, that is, the third horizontal angle at vertex B, A x and A y are the x-axis coordinate and y-axis coordinate of point A, B x and B yare the x-axis coordinate and y-axis coordinate of point B, C x and C y are the x-axis coordinates and y-axis coordinates of point C, a, b and c are the lengths of line segments a, b and c, respectively. h0 represents the height of the pixel corresponding to the projection point C on the smoke skeleton. C z Indicates the z-axis coordinate of the pixel.

[0151] S45, calculating the width of the standard smoke feature at each pixel point on the smoke skeleton. When calculating the width of the standard smoke feature at each pixel point on the smoke skeleton, the direction of the width should be perpendicular to the direction of the tangent of the smoke skeleton at that point. Therefore, in order to further clarify the calculation method of the width and prevent the situation where it is difficult to accurately identify abnormal pixel points through the width change rate due to different width calculation methods, step S45 specifically includes the following steps:

[0152] S451, establish radial lines perpendicular to the tangent lines of each point on the smoke skeleton; the vector of the radial line can be expressed as: (X i+1 -X i-1 , Y i+1 -Y i-1 , Z i+1 -Z i-1 ), in the above formula, if the pixel point corresponding to the radial line is the i-th pixel point on the smoke skeleton, then (X i+1 , Y i+1 , Z i+1 ) and (X i-1 , Y i-1 , Z i-1 ) are the x-axis, y-axis, and z-axis coordinates of the i+1th and i-1th pixel points adjacent to it, respectively.

[0153] S452, obtaining coordinates of two intersection points where the radial line intersects with the contour line of the standard smoke feature, the two intersection points being located on both sides of the smoke skeleton, and having the smallest vertical distance from the radial line;

[0154] S453. Calculate the width of the standard smoke feature at each point on the smoke skeleton according to the coordinates of the two intersection points; the calculation formula of the width is:

[0155]

[0156] In the above formula, w represents the width of any point on the smoke skeleton, E x 、E y and E z The x-coordinate, y-coordinate and z-coordinate of one of the intersection points of the radial line and the contour line of the standard smoke feature are represented respectively. x , D y and D zThe x-coordinate, y-coordinate and z-coordinate represent the intersection points of the radial line and the contour line of the standard smoke feature.

[0157] S46, according to the width change rate of the standard smoke feature at each pixel point on the smoke skeleton, identify and remove abnormal pixel points on the standard smoke feature to obtain the first smoke feature; when the smoke source is blocked, the width of the smoke exposed near the smoke source will decrease rapidly, so its width change rate is large, and it is necessary to remove the abnormal pixel points in this part to avoid affecting the subsequent calculation of the coordinates of the smoke source. In step S46, the following steps are specifically included:

[0158] S461, calculating the width change rate of the width at any pixel point on the smoke skeleton relative to the width at the adjacent pixel point on the lower side; the calculation formula of the width change rate is:

[0159]

[0160] In the above formula, Δw i It represents the width change rate of any pixel on the smoke skeleton relative to the width of the adjacent pixel on the lower side, w i and w i-1 They represent the width of any pixel on the smoke skeleton and the width of the adjacent pixel on the lower side of the smoke skeleton;

[0161] S462. Delete pixel points on the standard smoke feature corresponding to pixel points on the smoke skeleton whose width change rate is greater than a width change rate threshold, to obtain a first smoke feature.

[0162] S47. According to the first smoke feature and the width at each point on the smoke skeleton, when the predicted width is 0, the pixel coordinates corresponding to the smoke skeleton are the coordinates of the smoke source. The prediction method can adopt algorithms such as a neural network model or a linear regression equation.

[0163] S5, setting a distance threshold, and determining whether there is a photovoltaic device whose distance from the smoke source is within the distance threshold;

[0164] If yes, the photovoltaic device is marked as a faulty device, and the coordinates of the faulty device are output and the process ends;

[0165] If not, proceed to step S6;

[0166] S6. Predict the coordinates of the intersection of the smoke skeleton and the ground based on the coordinates of each point on the smoke skeleton and the three-dimensional topographic map of the photovoltaic power station, mark several photovoltaic devices closest to the intersection coordinates as faulty devices, and output the coordinates of the faulty devices.

[0167] The present invention first obtains the first image and the second image through a shooting device, and then in order to remove the complex outdoor background image, the method mentioned in step S2 is used to further identify the smoke characteristics by using regularity and area growth rate, and then further establishes a tangent segment to determine whether any point on the smoke feature contour line is an inner arc point and an outer arc point, thereby further achieving the purpose of identifying the inner circle contour line and the outer circle contour line on the smoke feature contour line, thereby converting the smoke characteristics into standard smoke characteristics, and then establishing a smoke skeleton to reflect the flow trend of the smoke, so that it is also convenient to calculate the coordinate points of the smoke skeleton by comparing the images, and to realize the estimation of the smoke source according to the smoke width. When there is a device near the coordinates of the smoke source, it is identified that the corresponding device may be a faulty device, and the corresponding device coordinates are output. When there is no device near the coordinates of the smoke source, it is possible that there is a certain height difference between the source and the ground, then the smoke source is extended to the ground according to the smoke skeleton, and a number of devices and coordinates closest to the coordinates of the smoke source are output, so that the fire-fighting robot can quickly reach the corresponding position.

[0168] The technical solution also provides a system for implementing a method for automatically identifying faults of photovoltaic equipment. The system includes a processor and a memory. The memory is used to store a computer program. When the computer program is executed by the processor, the method for automatically identifying faults of photovoltaic equipment is implemented.

[0169] The above implementation methods have been described in detail. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for automatically identifying faults of photovoltaic equipment, characterized in that: The method comprises the following steps: S1, respectively obtaining a plurality of first images and second images of the same area of ​​the corresponding photovoltaic power station including the current moment, and marking the first images and the second images as initial images; S2, filtering the initial image for noise to obtain a grayscale image, extracting a number of contrast images from the grayscale image according to a contrast period, and identifying smoke features according to the regularity and area growth rate of the moving targets in the number of contrast images; S3, establishing the minimum convex polygon of the smoke feature contour line in the comparison image at the current moment, and marking the convex polygon as a standard smoke feature; In step S3, the following steps are specifically included: S31, establishing a tangent segment with a rated length at any pixel point on the smoke characteristic contour line, with the pixel point as the center; S32, judging whether the pixel point corresponding to the tangent segment is an inner arc point according to whether the tangent segment is located within the smoke feature; If so, the pixel corresponding to the tangent segment is an inner arc point; If not, the pixel point corresponding to the tangent segment is an outer arc point; S33, marking the contour lines where several adjacent inner arc points are located as inner contour lines, and marking the contour lines where several adjacent outer arc points are located as outer contour lines; S34, sequentially acquiring two outer circle contour lines adjacent to each inner circle contour line, and establishing a first adjacent line tangent to both outer circle contour lines; S35, connecting two ends of the inner circle contour line adjacent to only one outer circle contour line through a second adjacent line; S36, extracting the contour line with the largest area in the area enclosed by the first adjacent line, the second adjacent line and the smoke feature contour line to obtain a minimum convex polygon, and using the minimum convex polygon as a standard smoke feature; S4. extracting the smoke skeleton according to the standard smoke feature in the comparison image at the current moment, calculating the smoke width of the standard smoke feature along the smoke skeleton, and calculating the coordinates of the smoke source where the smoke width is 0; S5, setting a distance threshold, and determining whether there is a photovoltaic device whose distance from the smoke source is within the distance threshold; If yes, the photovoltaic device is marked as a faulty device, and the coordinates of the faulty device are output and the process ends; If not, proceed to step S6; S6. Predict the coordinates of the intersection of the smoke skeleton and the ground based on the coordinates of each point on the smoke skeleton and the three-dimensional topographic map of the photovoltaic power station, mark several photovoltaic devices closest to the intersection coordinates as faulty devices, and output the coordinates of the faulty devices.

2. The identification method according to claim 1, characterized in that: In step S2, the following steps are specifically included: S21, converting a plurality of initial images into grayscale images, and removing noise from the current grayscale images; S22, setting a comparison period, and obtaining a grayscale image whose image generation time is the comparison period from a plurality of grayscale images, and marking it as a comparison image; S23, identifying a moving target in the comparison image and marking it as a suspicious target; S24, obtaining the contour length and area of ​​the suspicious target, and calculating its regularity; the calculation formula of the regularity is: In the above formula, N represents the regularity, C represents the contour length of the suspicious target, and S represents the area of ​​the suspicious target; S25. Calculate the area growth rate of the suspicious target according to its area in the plurality of comparison images; the area growth rate calculation formula is: In the above formula, ΔS i S represents the area growth rate of the suspicious target in the ith comparison image compared with the comparison image at the previous adjacent moment. i represents the area of ​​the suspicious target in the ith comparison image, S i-1 It represents the area of ​​the suspicious target in the i-1th contrast image, and the i-1th contrast image is at an adjacent moment before the i-th contrast image; S26, judging whether the suspicious target is a smoke feature according to whether the regularity of the suspicious target in the comparison image at the current moment is greater than the regularity threshold, and the area growth rate of each comparison image is greater than 0; If so, the suspected target is a smoke feature; If not, the suspicious target is not a smoke feature.

3. The identification method according to claim 2, characterized in that: In step S21, the following steps are specifically included: S211, converting a plurality of initial images into grayscale images; S212, setting a pixel value change threshold, and using the pixel value of each grayscale image at the same pixel point as the median, establishing a pixel value interval having a size equal to the pixel value change threshold; S213, counting the number of pixel values ​​contained in each pixel value interval according to the pixel value of the same pixel point in the plurality of grayscale images; S214, calculating the contribution coefficient of the same pixel point of each grayscale image to the pixel value according to the number of pixel values ​​in the pixel value interval; the calculation formula of the contribution coefficient is: In the above formula, Represents the contribution coefficient of the same pixel point in the i-th grayscale image to the pixel value, N i represents the number of pixel values ​​in the pixel value interval corresponding to the same pixel point in the i-th grayscale image, and m represents the total number of pixel value intervals corresponding to the same pixel point, that is, the number of grayscale images used for calculation; S215, updating the pixel value of any pixel in the current grayscale image according to the contribution coefficient of the same pixel of each grayscale image to the pixel value, so as to obtain a first grayscale image, where m represents the number of grayscale images used for calculation; In the above formula, I0 represents the pixel value of any pixel in the current grayscale image after update. It represents the contribution coefficient of any pixel in the i-th grayscale image to the pixel value, I i represents the pixel value of the corresponding pixel in the i-th grayscale image, and m represents the number of grayscale images used for calculation.

4. The identification method according to claim 1, characterized in that: In step S4, the following steps are specifically included: S41, extracting a smoke skeleton of standard smoke features; S42, respectively obtaining the coordinates of the shooting device used to shoot the two comparison images at the current moment, the lens deflection angle and the viewing angle to generate device data; S43, establishing a height line according to the altitude, and marking the height of each point on the smoke skeleton; S44, calculating the coordinates of each point on the smoke skeleton according to the device data, the height of each point on the smoke skeleton and the two comparison images at the current moment; S45, calculating the width of the standard smoke feature at each pixel point on the smoke skeleton; S46, identifying and removing abnormal pixels on the standard smoke feature according to the width change rate of each pixel on the smoke skeleton of the standard smoke feature, so as to obtain a first smoke feature; S47. According to the first smoke feature and the width at each point on the smoke skeleton, when the width is predicted to be 0, the coordinates of the pixel point corresponding to the smoke skeleton are the coordinates of the smoke source.

5. The identification method according to claim 4, characterized in that: In step S44, the following steps are specifically included: S441, establishing a first connection line between two photographing devices according to the device data; S442, establishing a second connection line between the camera lens and any height pixel point on the smoke skeleton; S443, calculating a first horizontal angle between the projection of the second connecting line and the normal line of the camera lens on the horizontal plane; the calculation formula of the first horizontal angle is: In the above formula, α1 represents the first horizontal angle, ΔN represents the number of pixels from any height pixel point on the smoke skeleton to the vertical line of the center of the comparison image, N0 is the number of pixels in any row of the comparison image, and α0 represents the viewing angle of the shooting device; S444, obtaining a second horizontal angle between the projection of the normal line of the camera lens and the first connecting line on the horizontal plane according to the lens deflection angle; S445. Calculate a third horizontal angle of projection of the second connecting line and the first connecting line on the horizontal plane according to the second horizontal angle and the first horizontal angle; the calculation formula of the third horizontal angle is: α3＝α1+α2 In the above formula, α3 represents the third horizontal angle, α1 and α2 represent the first horizontal angle and the second horizontal angle respectively. When the pixel point on the smoke skeleton corresponding to the second connecting line is located on the side of the lens normal line close to the other shooting device, the value of the first horizontal angle is taken as the opposite number; S446. Calculate the coordinates of each point on the smoke skeleton according to the third horizontal angle and the first connecting line.

6. The identification method according to claim 5, characterized in that: In step S446, the following steps are specifically included: S4461, projecting pixel points at any height on the shooting device and the smoke skeleton onto a horizontal plane; S4462, marking the projection points of the two shooting devices on the horizontal plane as point A and point B respectively, marking the projection point of a pixel point at any height on the smoke skeleton on the horizontal plane as point C, marking the second connecting line directly opposite to point A as line segment a, marking the second connecting line directly opposite to point B as line segment b, and marking the first connecting line as line segment c; S4463, calculating the coordinates of the pixel points on the smoke skeleton corresponding to the projection point; the coordinates of the pixel points on the smoke skeleton corresponding to the projection point C satisfy the following relationship: In the above formula, represents the angle at vertex A, that is, the third horizontal angle at vertex A, represents the angle at vertex B, that is, the third horizontal angle at vertex B, A x and A y are the x-axis and y-axis coordinates of point A, B x and B y are the x-axis coordinate and y-axis coordinate of point B, C x and C y are the x-axis coordinates and y-axis coordinates of point C, a, b and c are the lengths of line segments a, b and c, respectively. h0 represents the height of the pixel corresponding to the projection point C on the smoke skeleton. C z Indicates the z-axis coordinate of the pixel.

7. The identification method according to claim 4, characterized in that: In step S45, the following steps are specifically included: S451, establishing radial lines perpendicular to the tangent lines of each point on the smoke skeleton; S452, obtaining coordinates of two intersection points where the radial line intersects with the contour line of the standard smoke feature, the two intersection points being located on both sides of the smoke skeleton, and having the smallest vertical distance from the radial line; S453. Calculate the width of the standard smoke feature at each point on the smoke skeleton according to the coordinates of the two intersection points; the calculation formula of the width is: In the above formula, w represents the width of any point on the smoke skeleton, E x 、E y and E z represents the x-coordinate, y-coordinate and z-coordinate of one of the intersection points where the radial line intersects the contour line of the standard smoke feature, D x , D y and D z The x-coordinate, y-coordinate and z-coordinate represent the intersection point of the radial line with the contour line of the standard smoke feature.

8. The identification method according to claim 4, characterized in that: In step S46, the following steps are specifically included: S461, calculating the width change rate of the width at any pixel point on the smoke skeleton relative to the width at the adjacent pixel point on the lower side; the calculation formula of the width change rate is: In the above formula, Δw i It represents the width change rate of any pixel on the smoke skeleton relative to the width of the adjacent pixel on the lower side, w i and w i-1 They represent the width of any pixel on the smoke skeleton and the width of the adjacent pixel on the lower side of the smoke skeleton; S462. Delete pixel points on the standard smoke feature corresponding to pixel points on the smoke skeleton whose width change rate is greater than a width change rate threshold, to obtain a first smoke feature.

9. An identification system for implementing the identification method according to any one of claims 1 to 8, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, the method for automatically identifying faults of a photovoltaic device according to any one of claims 1 to 8 is implemented.

Images