A method for identifying electrical meters in substations based on PANet network and prior knowledge

Abstract

The invention discloses a method for identifying electric meters in a substation based on a PANet network and prior knowledge, comprising: adjusting the image brightness of the electric meter based on an image illumination compensation algorithm of perception theory to obtain an adjusted image; identifying key information of the adjusted image using a PANet network; correcting the key information using prior knowledge; determining the dial scale using the corrected key information, and obtaining the reading of the electric meter according to the dial scale; the method improves the accuracy of identifying the reading of the electric meter based on illumination compensation of the PANet network, prior knowledge and perception theory.

Description

Technical Field

[0001] The invention relates to the technical field of substation electrical meter identification, and in particular to a substation electrical meter identification method based on a PANet network and prior knowledge. Background Art

[0002] In recent years, with the development of social economy, the requirements of enterprises and users for the reliability of power supply of power system have gradually increased; as an important part of the power system, the safe and stable operation of high-voltage equipment in substations has become the key to operation and maintenance; various electrical instruments are often configured for the above-mentioned key equipment to display the working status of the equipment; therefore, when performing inspection tasks within the station, the importance of identifying and recording meter readings is self-evident; due to the influence of the complex on-site environment, manual inspections are labor-intensive, long working hours, and the inspection quality cannot be guaranteed.

[0003] In addition, in some areas, substations are scattered, the number of inspection personnel does not match the assigned workload, and a few inspection personnel even face safety issues; therefore, the demand for automated substation inspection technology is increasing. Although many substations have put into use inspection robots to solve the problem of inspection image shooting, electrical meter recognition in outdoor environments is still a difficult point due to interference factors such as lighting and changes in shooting angles.

[0004] Existing research on electrical meter image recognition is mainly divided into two types: traditional image processing methods and deep learning algorithms; the traditional image processing method first extracts the image in the HSV (Hue, Saturation, Value) color domain based on the color difference between the pointer and the dial, and then uses the dial contour and the pointer connected domain to determine the dial center and the pointer angle respectively, and finally calculates the reading; there is also a method that converts the dial image into a grayscale image, uses median filtering to denoise, and then uses edge detection and Hough transform to detect the pointer angle and calculate the reading. The traditional method is more friendly to the field of small sample learning, but it itself has strong prerequisites, and the robustness of the algorithm cannot be guaranteed, such as the inability to handle changes in shooting angles.

[0005] The other is an image recognition method based on deep learning, which directly regresses the readings, or calculates the readings after identifying key information such as the dial, pointer, and scale numbers; a regression model based on a convolutional neural network is proposed based on the AlexNet network structure. For the located dial image, the final reading can be obtained by directly inputting the model; a deep regression algorithm is used to identify the seven key points of the pointer instrument, and the pointer rotation angle is calculated using the above key points to finally obtain the reading. Although deep learning models generally have strong generalization performance, they have a large demand for data. How to use prior knowledge in the electrical field to reduce the model data volume demand becomes the key. Summary of the invention

[0006] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0007] In view of the above existing problems, the present invention is proposed.

[0008] Therefore, the present invention provides a method for identifying electrical meters in a substation based on a PANet network and prior knowledge, which can avoid (1) the problem of inaccurate readings of electrical meters caused by too dark images;

[0009] (2) The problem of inaccurate readings of electrical meters caused by changes in shooting angles.

[0010] To solve the above technical problems, the present invention provides the following technical solutions: including: adjusting the image brightness of an electric meter based on an image illumination compensation algorithm of perception theory to obtain an adjusted image; identifying key information of the adjusted image using a PANet network; correcting the key information using prior knowledge; determining the dial scale using the corrected key information, and obtaining the reading of the electric meter according to the dial scale.

[0011] As a preferred solution of the substation electrical meter recognition method based on PANet network and prior knowledge described in the present invention, the image illumination compensation algorithm of the perception theory includes:

[0012] For a single channel, the image brightness of the electric meter is adjusted using the single channel compensation J(L);

[0013] For multiple channels, use multi-channel compensation Adjust the image brightness of the electrical meter.

[0014] As a preferred solution of the substation electrical meter identification method based on the PANet network and prior knowledge of the present invention, the single-channel compensation J(L) includes:

[0015]

[0016] Among them, ρ(x,y) is the weight of each pixel position, I(x,y) is the image brightness, L(x,y) is the illumination, x and y are the two directions of the smoothness constraint, and Calculate the partial derivatives for x and y respectively, λ is the coefficient, and Ω is the entire image.

[0017] As a preferred solution of the substation electrical meter identification method based on PANet network and prior knowledge described in the present invention, wherein: the multi-channel compensation include:

[0018]

[0019] Among them, G(x,y) is the gray value obtained by transformation, L G (x,y) is the brightness reconstructed based on the grayscale value.

[0020] As a preferred solution of the substation electrical meter identification method based on the PANet network and prior knowledge described in the present invention, the key information includes: a dial and a pointer.

[0021] As a preferred solution of the substation electrical meter identification method based on PANet network and prior knowledge described in the present invention, wherein: Includes: Based on the center of the circle (h o ,j o ), rotation angle θ, semi-major axis l a and semi-minor axis l b , establish the internal equation of the dial

[0022] u=(hh o )cosθ+(jj o )sinθ

[0023] v=(hh o )sinθ-(jj o )cosθ

[0024]

[0025] Where (h, j) is any point inside the dial.

[0026] As a preferred solution of the substation electrical meter identification method based on PANet network and prior knowledge described in the present invention, the pointer Including: Set the shape of the pointer to be a triangle with three vertices and the interior points of the triangle Based on the three vertices of a triangle and the interior points of the triangle Establish the interior equation of a triangle

[0027]

[0028] Among them, (h′,j′) is any point inside the pointer,

[0029] As a preferred embodiment of the substation electrical meter recognition method based on the PANet network and prior knowledge of the present invention, the corrected key information includes:

[0030]

[0031] Among them, Ω0 is the pixel position of the dial area, and Ω1 is the pixel position of the pointer area.

[0032] As a preferred embodiment of the substation electrical meter recognition method based on the PANet network and prior knowledge of the present invention, the dial scale: by correcting the key information, determine the pointer size and the ellipse size; calculate the perimeter C of the ellipse:

[0033] C = 2πl b + 4(l a - l b )

[0034] Calculate the scales m and n on both sides of the grid where the pointer is located according to the perimeter of the ellipse:

[0035] n = (2πl b + 4(l a - l b )) * N / i

[0036] m = (2πl b + 4(l a - l b )) * M / i

[0037] Among them, m < n, i is the total number of scales, N is the Nth scale on the right side where the pointer is located, and M is the Mth scale on the left side where the pointer is located.

[0038] As a preferred embodiment of the substation electrical meter recognition method based on the PANet network and prior knowledge of the present invention, the reading r of the electrical meter includes:

[0039]

[0040] Among them, θ m and θ n are the radians corresponding to the scales m and n respectively, and θ r is the radian corresponding to the pointer.

[0041] The beneficial effects of the present invention: (1) The present invention solves the interference problem caused by light changes during outdoor reading recognition based on the light compensation of the perception theory, and at the same time aims to improve the local contrast, thereby improving the recognition accuracy;

[0042] (2) The present invention is based on the PANet network and prior knowledge, which reduces the data requirement while ensuring the recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0044] Figure 1 A flowchart of a method for identifying electrical meters in a substation based on a PANet network and prior knowledge according to a first embodiment of the present invention;

[0045] Figure 2 This is a diagram showing the effect before illumination compensation of the method for identifying electrical meters in a substation based on the PANet network and prior knowledge according to the first embodiment of the present invention;

[0046] Figure 3 This is a diagram showing the effect of illumination compensation of the substation electrical meter identification method based on the PANet network and prior knowledge described in the first embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0050] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0051] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0052] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] Example 1

[0054] Reference Figures 1 to 3 , which is the first embodiment of the present invention, provides a method for identifying electrical meters in a substation based on a PANet network and prior knowledge, comprising:

[0055] S1: An image illumination compensation algorithm based on perception theory adjusts the image brightness of the electrical meter to obtain an adjusted image.

[0056] (1) Two conclusions in perception theory: a) Humans are more sensitive to environmental reflectivity than lighting conditions; b) In the perception process, local contrast plays a greater role than overall brightness.

[0057] (2) The image illumination compensation algorithm of the perception theory includes: for a single channel, using the single channel compensation J(L) to adjust the image brightness of the electric meter; for multiple channels, using the multi-channel compensation Adjust the image brightness of the electrical meter.

[0058] (3) Based on the above conclusions, refer to Figures 2-3The following are the effects before and after illumination compensation. The illumination compensation task can be described in the following mathematical language: given the image brightness I(x, y), the reconstructed illumination L(x, y) enhances the local contrast. According to Weber's law, we can establish the following optimization objective function: single channel compensation J(L):

[0059]

[0060] Among them, ρ(x,y) is the weight of each pixel position, x and y are the two directions of the smoothness constraint, and Calculate the partial derivatives for x and y respectively, λ is the coefficient, and Ω is the entire image.

[0061] (3) For multi-channel images, the image is first converted into a grayscale image, the brightness is restored using the grayscale image, and then the image is proportionally applied to the original image. Right now:

[0062]

[0063] Among them, G(x,y) is the gray value obtained by transformation, L G (x,y) is the brightness reconstructed based on the grayscale value.

[0064] Preferably, the present invention eliminates the influence of external brightness and darkness by using an image illumination compensation algorithm based on perception theory, while aiming to improve local contrast, which is beneficial to subsequent image recognition.

[0065] S2: Using the PANet network to identify key information of the adjusted image.

[0066] Key information includes: dial and hands.

[0067] Preferably, since the target to be identified is relatively small, the present invention enhances the recognition accuracy of targets of different small sizes by utilizing the PANet network, accurately identifies the dial, pointer and scale numbers, and improves the accuracy of recognition.

[0068] S3: Use prior knowledge to correct key information.

[0069] (1) For the prior knowledge, the shapes of the dial and pointer are fixed as ellipse and triangle, and all possible scale distributions are stored in advance. Finally, the optimal shape parameters are estimated using the identified pixel positions of the dial and pointer, and the most likely distribution is determined using the identified scale positions and digital information.

[0070] (2) The dial includes: based on the center of the circle (h o ,j o ), rotation angle θ, semi-major axis l a and semi-minor axis l b, establish the internal equation of the dial

[0071] u=(hh o )cosθ+(jj o )sinθ

[0072] v=(hh o )sinθ-(jj o )cosθ

[0073]

[0074] Where (h, j) is any point inside the dial.

[0075] (3) The pointer includes: Assume that the shape of the pointer is a triangle with three vertices and the interior points of the triangle Based on the three vertices of a triangle and the interior points of the triangle Establish the interior equation of a triangle

[0076]

[0077] Among them, (h′,j′) is any point inside the pointer.

[0078] (4) The revised key information includes:

[0079]

[0080] Among them, Ω0 is the pixel position of the dial area, and Ω1 is the pixel position of the pointer area.

[0081] Preferably, the present invention reduces the data requirement while ensuring the generalization ability by utilizing prior knowledge; in addition, the reliability of the reading is further ensured by correcting the data through prior knowledge.

[0082] S4: Determine the dial scale using the corrected key information, and obtain the reading of the electrical meter according to the dial scale.

[0083] (1) Dial scale:

[0084] By correcting the key information, the pointer size and the ellipse size are determined;

[0085] Calculate the circumference C of the ellipse:

[0086] C=2πl b +4(l a -l b )

[0087] Calculate the scales m and n on both sides of the grid where the pointer is located according to the circumference of the ellipse respectively:

[0088] n = (2πl b +4(l a -l b ))*N / i

[0089] m = (2πl b +4(l a -l b ))*M / i

[0090] Where m < n, i is the total number of scales, N is the Nth scale on the right side where the pointer is located, and M is the Mth scale on the left side where the pointer is located.

[0091] (2) The reading r of the electrical meter includes:

[0092]

[0093] Where θ m and θ n are the radians corresponding to the scales m and n respectively, and θ r is the radian corresponding to the pointer.

[0094] Preferably, by using the relative radian in the present invention, the final reading is independent of the starting position of the radian calculation, further enhancing the transferability of the method.

[0095] Example 2

[0096] In order to verify and illustrate the technical effects adopted in this method, different methods selected in this example are compared and tested with this method, and the experimental results are compared by means of scientific demonstration to verify the real effects of this method.

[0097] This method pre-trains the PANet network model on the MS COCO dataset. The fixed layer is the basic network, and its function is to extract image features. The basic network uses ResNet50. The region candidate network feature pyramid network is trained separately, and its total number of samples is 512, and the positive-negative sample ratio is 1:3; the network weight decay parameter is set to 0.0001, and the momentum parameter is set to 0.9. As the number of iterations increases, the learning rate changes from 0.02 to 0.002. In addition, in order to avoid overfitting, a dropout layer is added, and the parameter is 0.2. Secondly, there is feature path enhancement based on convolution and summation operations, followed by adaptive feature pooling based on multi-level feature fusion, and finally there is a prediction including a mask branch and a classification branch.

[0098] Traditional technical solutions: For outdoor electricity meter identification, there are interference factors such as lighting and shooting angle changes. Although the lighting compensation algorithm can be integrated, the problem of shooting angle changes cannot be solved. This is because the traditional method cannot obtain scale information, and reading errors will inevitably occur when the surface rotates.

[0099] In order to verify that the present method has higher image recognition accuracy and better reading recognition than traditional image processing algorithm, illumination compensation algorithm (ICA) and traditional image processing algorithm, convolutional neural network regression method, illumination compensation algorithm (ICA) and convolutional neural network regression method, Mask RCNN, illumination compensation algorithm (ICA) and Mask RCNN, in this embodiment, the traditional image processing algorithm, illumination compensation algorithm (ICA) and traditional image processing algorithm, convolutional neural network regression method, illumination compensation algorithm (ICA) and convolutional neural network regression method, Mask RCNN, illumination compensation algorithm (ICA) and Mask RCNN are used to perform 200 real-time comparisons on the same image with the present method.

[0100] Table 1: Comparison of image recognition and reading accuracy of different models.

[0101] Model Image Recognition reading Traditional image processing algorithms / 63.3％ ICA+ traditional image processing algorithm / 68.2％ Convolutional Neural Network Regression / 53.6％ ICA+convolutional neural network regression method / 65.8％ Mask RCNN 81.7％ 78.0％ ICA+Mask RCNN 84.5％ 82.9％ This method 89.4％ 95.1％

[0102] It can be seen from the above table that the image recognition accuracy of the present invention is better than that of other methods, and the accuracy of reading is significantly greater than that of other methods.

[0103] It should be appreciated that embodiments of the present invention may be implemented or enforced by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable memory. The method may be implemented in a computer program using standard programming techniques - including a non-transitory computer-readable storage medium configured with a computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the methods and drawings described in the specific embodiments. Each program may be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if desired, the program may be implemented in an assembly or machine language. In any case, the language may be a compiled or interpreted language. In addition, the program may be run on a programmed ASIC for this purpose.

[0104] Furthermore, the operations of the processes described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes described herein (or variations and / or combinations thereof) may be performed under the control of one or more computer systems configured with executable instructions, and may be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) that is executed collectively on one or more processors, by hardware, or a combination thereof. The computer program includes a plurality of instructions that may be executed by one or more processors.

[0105] Further, the method can be implemented in any type of computing platform that is operably connected to a suitable computer, including but not limited to a personal computer, a minicomputer, a mainframe, a workstation, a network or distributed computing environment, a separate or integrated computer platform, or in communication with a charged particle tool or other imaging device, etc. Aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, an optical read and / or write storage medium, a RAM, a ROM, etc., so that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the process described herein. In addition, the machine-readable code, or part thereof, can be transmitted via a wired or wireless network. When such media includes instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor, the invention described herein includes these and other different types of non-transitory computer-readable storage media. When programmed according to the methods and techniques described in the present invention, the present invention also includes the computer itself. The computer program can be applied to input data to perform the functions described herein, thereby converting the input data to generate output data stored in a non-volatile memory. The output information can also be applied to one or more output devices such as a display. In a preferred embodiment of the present invention, the converted data represents a physical and tangible object, including a specific visual depiction of the physical and tangible object produced on a display.

[0106] As used in this application, the terms "component", "module", "system", etc. are intended to refer to a computer-related entity, which can be hardware, firmware, a combination of hardware and software, software, or software in operation. For example, a component can be, but is not limited to: a process running on a processor, a processor, an object, an executable file, a thread in execution, a program, and / or a computer. As an example, an application running on a computing device and the computing device can both be components. One or more components can exist in a process and / or thread in execution, and a component can be located in a computer and / or distributed between two or more computers. In addition, these components can be executed from various computer-readable media having various data structures thereon. These components can communicate in a local and / or remote process manner, such as according to a signal having one or more data packets (e.g., data from a component that interacts with another component in a local system, a distributed system, and / or interacts with other systems in a signal manner through a network such as the Internet).

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for identifying electrical meters in a substation based on a PANet network and prior knowledge, characterized in that: Including: Adjust the image brightness of the electrical meter using the image illumination compensation algorithm based on the perception theory to obtain the adjusted image; Use the PANet network to identify the key information of the adjusted image; Use prior knowledge to correct the key information; Use the corrected key information to determine the dial scale, and obtain the reading of the electrical meter according to the dial scale; The key information includes: the dial and the pointer; The dial includes: Based on the center (h o ,j o ), rotation angle θ, semi-major axis l a and semi-minor axis l b , establish the internal equation of the dial u=(h-h o )cosθ+(j-j o )sinθ v=(h-h o )sinθ-(j-j o )cosθ where (h, j) is any point inside the dial; The pointer includes: Assume the shape of the pointer is a triangle with three vertices and the interior points of the triangle Based on the three vertices of a triangle and the interior points of the triangle Establish the interior equation of a triangle where (h′, j′) is any point inside the pointer; The dial scale: Determine the pointer size and the ellipse size by correcting the key information; Calculate the perimeter C of the ellipse: C＝2πl b +4(l a -l b ) Calculate the scales m and n on both sides of the grid where the pointer is located according to the perimeter of the ellipse: n=(2πl b +4(l a -l b ))*N / i m=(2πl b +4(l a -l b ))*M / i where m < n, i is the total number of scales, N is the Nth scale on the right side where the pointer is located, and M is the Mth scale on the left side where the pointer is located.

2. The method for identifying electrical meters in a substation based on a PANet network and prior knowledge as claimed in claim 1, characterized in that: The image illumination compensation algorithm based on the perception theory includes For a single channel, use the single-channel compensation J(L) to adjust the image brightness of the electrical meter; For multiple channels, use multi-channel compensation Adjust the image brightness of the electrical meter.

3. The method for identifying electrical meters in a substation based on a PANet network and prior knowledge as claimed in claim 2, characterized in that: The single-channel compensation J(L) includes: Among them, ρ(x,y) is the weight of each pixel position, I(x,y) is the image brightness, L(x,y) is the illumination, x and y are the two directions of the smoothness constraint, and Calculate the partial derivatives for x and y respectively, λ is the coefficient, and Ω is the entire image.

4. The method for identifying electrical meters in a substation based on a PANet network and prior knowledge as claimed in any one of claims 2 or 3, characterized in that: The multi-channel compensation include: Among them, G(x,y) is the gray value obtained by transformation, L G (x,y) is the brightness reconstructed based on the grayscale value.

5. The method for identifying electrical meters in a substation based on a PANet network and prior knowledge as claimed in claim 1, characterized in that: The corrected key information includes: where Ω0 is the pixel point position of the dial area, and Ω1 is the pixel point position of the pointer area.

6. The method for identifying electrical meters in a substation based on a PANet network and prior knowledge as claimed in claim 1, characterized in that: The reading r of the electrical meter includes: Among them, θ m and θ n are the radians corresponding to the scales m and n, θ r is the radian corresponding to the pointer.

Images