CN113343998A - Reading monitoring system and method for electric power mechanical meter, computer equipment and application - Google Patents
Reading monitoring system and method for electric power mechanical meter, computer equipment and application Download PDFInfo
- Publication number
- CN113343998A CN113343998A CN202110640232.3A CN202110640232A CN113343998A CN 113343998 A CN113343998 A CN 113343998A CN 202110640232 A CN202110640232 A CN 202110640232A CN 113343998 A CN113343998 A CN 113343998A
- Authority
- CN
- China
- Prior art keywords
- reading
- meter
- image
- alarm
- electromechanical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Alarm Systems (AREA)
Abstract
The invention belongs to the technical field of intelligent monitoring, and discloses a reading monitoring system and method for an electric mechanical meter, computer equipment and application, wherein the reading monitoring method for the electric mechanical meter comprises the following steps: the user inputs 1 image of the front of each meter to be identified in advance, and inputs the type, number and alarm reading information of the meter; collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm; the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment. The incremental training method can effectively reduce reading errors caused by the position of the camera, can adapt to different scenes, is simple and easy to expand, can perform incremental training on the basis of the existing model, and can perform incremental training only by recording a small amount of data by the camera and labeling the data if the recognition effect is poor.
Description
Technical Field
The invention belongs to the technical field of intelligent monitoring, and particularly relates to a reading monitoring system and method for an electric power machinery meter, computer equipment and application.
Background
At present, various mechanical meters, such as a pressure meter, a sulfur hexafluoride meter and the like, exist in an electric power scene and are used for monitoring the running states of various devices in the electric power scene. In the operation process, the mechanical meters are installed in different positions of the power scene, and the operation states of various facilities in the power scene are monitored, so the reading of the mechanical meters directly influences the service life of equipment and the life safety of workers. Currently, the reading of each mechanical meter is usually confirmed by manually observing or presetting a camera to collect video for the traditional image algorithm to recognize. In practical application, the problems of low efficiency, untimely response and the like exist when the reading is purely observed manually, and the traditional image algorithm is lower in robustness of different scenes, so that the safe operation of power scenes is influenced.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) the traditional image recognition mechanical meter reading is greatly influenced by the environment, the problem can be partially solved only by adjusting a large amount of parameters, and the popularization is difficult.
(2) The traditional image algorithm is low in robustness of different scenes, and safe operation of power scenes is influenced.
(3) For the confirmation of the reading of each mechanical meter, the manual observation method has the problems of low efficiency, untimely response, high risk and the like.
The difficulty in solving the above problems and defects is:
(1) the traditional image is obtained by extracting various feature abstract description images of the image, extracting representative features as a reference after analyzing a feature space, and identifying or judging the state of a target in the image, so that the manually determined feature dimension is low, the image cannot adapt to different scenes, and once the scenes are switched, a large amount of parameters are required, so that the image is difficult to popularize;
(2) the workman needs the naked eye to observe when artifical the affirmation, because there are various facilities in the electric power scene, and the mounted position of mechanical meter differs, and some positions are in safe consideration, and the workman is difficult closely to observe, consequently can not obtain effectual reading.
The significance of solving the problems and the defects is as follows: because various facilities exist in the electric power scene, the electric power mechanical meter is an important component for monitoring the working states of the facilities, if the facilities cannot be monitored, serious potential safety hazards exist, if the facilities are monitored simply by manpower, the problems of low efficiency and the like exist, and the traditional algorithm is difficult to popularize due to the limitation of the traditional algorithm, so that after the problem is solved, the working states of all equipment in the electric power scene can be effectively mastered, and the potential safety hazards are reduced.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a reading monitoring system and method for an electromechanical meter, computer equipment and application.
The invention is realized in such a way that the reading monitoring method of the electric mechanical meter comprises the following steps:
firstly, a user inputs 1 image of the front side of each meter to be identified in advance, and inputs the type, the number, the counting unit and reading threshold information to be given an alarm, wherein the image is used as a precondition to be recorded by a system and provide a precondition for reading identification;
acquiring an image containing the electromechanical meter through a camera, and inputting the image and the type of the meter into an algorithm, wherein the purpose of the algorithm is to acquire image data of a target meter which is enough to be identified and provide a data source for intelligent reading monitoring;
and step three, automatically identifying the reading of the electric mechanical meter in the graph by the algorithm and comparing the reading with a preset alarm reading, sending alarm information to a background if the reading is larger than or equal to the preset alarm reading for a user to judge the current running state of the equipment, and when the running state is abnormal, the user can timely find the current running state through various alarm modes such as images, sounds and the like, further check the equipment state and timely eliminate the fault and the abnormality.
Further, in step three, the identification of the reading of the electromechanical meter in the graph includes:
identifying the current mechanical meter reading by combining a plurality of deep learning networks; wherein the deep learning network comprises: the system comprises a target detection network, a digital identification network and an image segmentation network.
Further, the identifying, in combination with the plurality of deep learning networks, for the current mechanical meter reading includes:
(1) inputting the image to a target detection network;
(2) reading identifying the detected number: extracting the digital position obtained in the step (1) and inputting the digital position into a digital identification network;
(3) obtaining the position information of scales and pointers in the mechanical meter: extracting the meter position obtained in the step (1) and inputting the meter position into a segmentation network;
(4) after the results of the steps (1) to (3) are obtained, corresponding the detected numbers to the numbers on the template one by one through a preset template to generate a transformation matrix; the image is converted to a positive shooting state through a conversion matrix, and meanwhile, the number, the pointer and the scale are converted together;
(5) calculating principal component of the pointer information to obtain principal component direction information, and calculating intersection positions on a re-circle by combining a scale fitting circle obtained by dividing and the principal component direction of the pointer;
(6) calculating an included angle formed by the intersection point and the scale, and calculating the included angle to obtain a final result; and if the final result is compared with a preset alarm range, sending alarm information if alarm is needed, and if the alarm is not needed, acquiring the next frame of image for continuous identification.
Further, in step (1), the inputting the image into the target detection network includes:
1) images resize to 800 × 800;
2) establishing a top-bottom network, performing continuous convolution, pooling and residual calculation in multiple stages, and constructing feature-maps with three different scales through a connection layer;
3) sending the obtained feature image data of three scales into corresponding activation;
4) and finally, performing post-processing such as non-maximum value inhibition hms, comprehensively analyzing the results of the three scales, and mapping the results into the original image to obtain the type and the corresponding coordinate position of the target.
Further, in step (2), the recognizing the reading of the detected number includes:
1) image resize to 100 x 32;
2) establishing a top-bottom network, constructing a feature-map through continuous convolution pooling of 5 stages, and serializing the feature-map;
3) inputting the serialized data into a double-circulation layer, and determining the label distribution of the characteristic sequence by a circulation network;
4) and inputting the obtained label distribution into a transcription layer, and obtaining a final result through de-duplication and integration operations.
Further, in the step (3), the obtaining of the position information of the scales and the pointer in the mechanical meter includes:
1) images resize to 512 x 512;
2) establishing a top-bottom network, performing convolution, activation and pooling in 5 continuous stages, performing convolution, activation and upsampling in 5 continuous stages, and simultaneously superposing the feature image after convolution into a feature image with the same size obtained by subsequent upsampling to generate a final feature-map;
3) and inputting the output feature map into the softmax activation layer to be converted into a segmentation result.
Further, the method for monitoring the reading of the electromechanical meter further comprises the following steps:
(1) in the operation process, firstly, a camera acquires a current image frame in real time, and the position of an electric mechanical meter in the current image is determined through a detection model;
(2) extracting a roi partial image of the power machinery meter, and identifying a scale position, a pointer position and a number in the roi partial image; after the image is compared with a preset template, the image is corrected to be in a positive shooting state, and image deformation caused by the shooting position is reduced;
(3) analyzing the position information of scales, numbers and pointers in the corrected image to determine the reading; if the reading is larger than or equal to the preset alarm reading, sending information to a background for using the working state of the judgment equipment, and if the reading is smaller than the preset alarm reading, automatically extracting the next frame of image.
Another object of the present invention is to provide a reading monitoring system for an electromechanical meter, which implements the reading monitoring method for an electromechanical meter, the reading monitoring system comprising: the system comprises a router, a plurality of cameras and a pc serving as a server side;
the router is respectively connected with the cameras and the server, and the router, the cameras and the server pc are connected through the router and are located in the same local area network.
It is a further object of the invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
the user inputs 1 image of the front of each meter to be identified in advance, and inputs the type, number and alarm reading information of the meter;
collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
the user inputs 1 image of the front of each meter to be identified in advance, and inputs the type, number and alarm reading information of the meter;
collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
The invention also aims to provide the electric mechanical meter which is provided with the reading monitoring system of the electric mechanical meter.
Another object of the present invention is to provide an electric power scenario, wherein the electric power scenario is provided with the electric power mechanical meter.
The invention also aims to provide an information data processing terminal which is used for realizing the reading monitoring system of the electric mechanical meter.
By combining all the technical schemes, the invention has the advantages and positive effects that: according to the reading monitoring method of the electric mechanical meter, provided by the invention, the scene image collected in the electric power place is received, the coordinate of the electric mechanical meter is detected for the collected scene image, and the type of the mechanical meter is input by a user; identifying the digital position, digital reading, scale and meter pointer of the detected electromechanical meter; taking a pre-shot meter front image as a template according to the meter type input by a user, and correcting the current image; and finally, synchronously mapping the numbers, the scales and the pointer to the corrected image to determine the reading of the electromechanical meter. The invention has high safety management level for workers in the electric power field; the meter reading can be obtained through the camera, real-time intelligent analysis is carried out and the pushing is given to operation and maintenance managers, and the occurrence of safety accidents can be greatly reduced.
The invention trains models of different tasks through a deep learning method, is used for detecting the positions of the meters and the numbers, identifying the numbers, dividing scales and pointers in the meters, combines the results of the three and a preset meter positive shooting template, corrects the meter positive shooting template into a positive shooting image, obtains meter reading through principal component analysis, scale position and other information after reducing deformation caused by image inclination, and can effectively reduce reading errors caused by the positions of cameras by running the scheme. Meanwhile, the invention adopts a deep learning framework, improves the robustness of the model to the illumination condition, noise and camera position, can effectively adapt to different scenes, is simple and easy to expand, can carry out incremental training on the basis of the existing model under the deep learning framework, and can carry out incremental training only by recording a small amount of data by using the camera and labeling the data if the recognition effect is poor.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments of the present invention will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a reading monitoring method for an electromechanical meter according to an embodiment of the present invention.
Fig. 2 is a flow chart of software in the reading monitoring system of the electromechanical meter according to the embodiment of the present invention.
FIG. 3 is a schematic structural diagram of a reading monitoring system of the electromechanical meter according to an embodiment of the present invention;
in the figure: 1. a camera; 2. a server side; 3. a router.
Fig. 4 is a schematic diagram of a network structure of a power scene image detection mechanical meter and a number according to an embodiment of the present invention.
Fig. 5 is a schematic network structure diagram of an electric power scene image identification number provided by an embodiment of the present invention.
Fig. 6 is a schematic diagram of a network structure of pointers and scales of a power scene image segmentation mechanical meter according to an embodiment of the present invention.
Fig. 7 is a diagram illustrating an application effect of the reading system of the electromechanical meter according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In view of the problems in the prior art, the present invention provides a reading monitoring system, method, computer device and application of an electromechanical meter, and the present invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the reading monitoring method for the electromechanical meter provided by the embodiment of the present invention includes the following steps:
s101, a user inputs 1 image of the front side of each meter to be identified in advance, and inputs the type, the number and the alarm reading information of the meter;
s102, collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
s103, the algorithm automatically identifies the reading of the electric mechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
The technical solution of the present invention is further described with reference to the following examples.
Example 1
The invention is realized in such a way, and the reading monitoring method for the mechanical meter in the electric power scene comprises the following steps:
(1) the user inputs 1 image of the front of each meter to be identified in advance, and inputs the information of the type, number, alarm reading and the like of the meter;
(2) collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
(3) the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
Further, in the step (2), in the operation process, firstly, the camera acquires the current image frame in real time, and the image and the type of the mechanical meter in the image are input into the deep learning model to judge the reading of the mechanical meter in the current image;
further, a plurality of deep learning networks are combined to identify the current mechanical meter reading.
Respectively a target detection network, a digital identification network and an image segmentation network;
firstly, inputting an image into a target detection network;
(1) images resize to 800 × 800;
(2) establishing a top-bottom network, performing continuous convolution, pooling and residual calculation in multiple stages, and constructing feature-maps with three different scales through a connection layer;
(3) sending the obtained feature image data of three scales into corresponding activation;
(4) and finally, performing post-processing such as non-maximum value inhibition hms, comprehensively analyzing the results of the three scales, and mapping the results into the original image to obtain the type and the corresponding coordinate position of the target.
This step detects the meter and the digital position in the meter;
and step two, after extracting the digital position obtained in the step one, inputting the digital position into a digital identification network:
(1) image resize to 100 x 32;
(2) establishing a top-bottom network, constructing a feature-map through continuous convolution pooling of 5 stages, and serializing the feature-map;
(3) inputting the serialized data into a double-circulation layer, and determining the label distribution of the characteristic sequence by a circulation network;
(4) and finally, the obtained labels are distributed and input into a transcription layer, and a final result is obtained through operations such as duplication removal and integration.
This step identifies a reading of the number detected in step (1).
And step three, extracting the meter position obtained in the step one, and inputting the meter position into a segmentation network:
(1) images resize to 512 x 512;
(2) establishing a top-bottom network, performing convolution, activation and pooling in 5 continuous stages, performing convolution, activation and upsampling in 5 continuous stages, and simultaneously superposing the feature image after convolution into a feature image with the same size obtained by subsequent upsampling to generate a final feature-map;
(3) and inputting the output characteristic diagram into the softmax activation layer to be converted into a segmentation result.
This step will obtain the position information of the scale and the pointer in the mechanical meter.
Further, after the results of the three steps are obtained, the detected numbers correspond to the numbers on the template one by one through a preset template, and a transformation matrix is generated.
The image is transformed to the positive state by the transformation matrix, and simultaneously, the number, the pointer and the scale are transformed together.
And calculating the principal component of the pointer information to obtain principal component direction information, and calculating the intersection position on the re-circle by combining a scale fitting circle obtained by dividing and the principal component direction of the pointer.
And finally, calculating an included angle formed by the intersection point and the scale, and calculating the included angle to obtain a final result.
And if the final result is compared with a preset alarm range, sending alarm information if alarm is needed, and if the alarm is not needed, acquiring the next frame of image for continuous identification.
Another object of the present invention is to provide an electric machine meter monitoring system for implementing the electric machine meter monitoring method, the electric machine meter monitoring system including: the system comprises a router, a plurality of cameras and a pc serving as a server side;
the router is respectively connected with the camera and the server.
Further, the router, the cameras and the server pc are all connected through the router and are located in the same local area network.
Example 2
A person skilled in the art can also use other steps to implement the reading monitoring method of the electromechanical meter provided by the present invention, and the reading monitoring method of the electromechanical meter provided by the present invention in fig. 1 is only a specific example.
In S103 provided by the embodiment of the present invention, in the operation process, after the user inputs the meter type corresponding to the shooting of the camera, the camera acquires the current image frame in real time, and after the algorithm receives the image and the meter type, the meter in the image and the number on the meter are detected by the detection model to acquire the position information of the image and the number on the meter; then, identifying the detected number through a number identification model; then, obtaining position information of the scales and the pointer through a segmentation model, and corresponding to a preset meter type template; generating a transformation matrix after the digital positions are in one-to-one correspondence, acting on the obtained result data, and correcting the meter image, the scale position and the pointer position to the positive shot image of the meter; finally, the reading of the current meter is calculated by using the principal component direction of the pointer and combining the scale position information; and if the reading of the meter is within the preset alarm range of the user, sending alarm information, and if the reading of the meter is not beyond the preset alarm range, directly acquiring the next frame of image for continuous judgment. The process is independent of cameras, data of each camera is independently processed, and under the support of multithreading, one server can support simultaneous judgment of reading monitoring of a plurality of electromechanical meters.
The software flow in the reading monitoring system of the electromechanical meter provided by the embodiment of the invention is shown in fig. 2.
As shown in fig. 3, the system for monitoring the operating state of the disconnecting switch according to the embodiment of the present invention includes: the system comprises a plurality of cameras 1, a pc serving as a server 2 and a router 3, wherein all devices are connected through the router and are located in the same local area network. The router 3 is respectively connected with the camera 1 and the server 2.
When the system runs, a program in the server 2 can access a preset camera 1 to identify the reading of the electromechanical meter in the camera; if the reading in the camera 1 reaches the preset alarm range, the alarm information is directly sent.
The neural network structure for identifying the electromechanical meter in real time is shown in fig. 4, 5 and 6.
Fig. 4 shows a detection network, which is mainly composed of a series of convolutional layers, pooling layers, and residuals, and finally obtains position information of a meter and numbers by using non-maximum suppression (nms).
Fig. 5 shows an identification network, which mainly includes a convolutional layer, a bi-cyclic layer, and a transcription layer, where the convolutional layer extracts a feature sequence through a series of convolution and pooling operations, inputs the feature sequence into the bi-cyclic layer to obtain a corresponding tag, and finally performs operations such as sorting and de-duplication on the tag in the transcription layer to obtain a final identification result.
Fig. 6 shows a segmented network, which mainly consists of a convolutional layer, a pooling layer and a deconvolution, and obtains position information of the scale and the pointer in the meter after being activated by an activation function (softmax) through a series of transformations.
The working principle of the invention is as follows: the method comprises the steps of collecting templates needing meter identification in advance, setting types of the meters, preparing a plurality of cameras, enabling each camera to correspond to an electric mechanical meter, presetting the types of the meters and the alarm range in the cameras, connecting the cameras and a service end to a router, setting the cameras and the service end to be in the same network segment, and then starting algorithm service. The algorithm service can be automatically connected with the cameras to acquire image data, the reading of the electric power mechanical meter corresponding to each camera is judged through the deep learning neural network, whether the current reading needs to give an alarm or not is judged, and if the reading needs to give an alarm, alarm information is output.
The algorithm is trained by more than 2000 meters of different types commonly seen in the production of the power industry, wherein the meters comprise a thermometer, a hygrometer, a pressure gauge, an ammeter, a sulfur hexafluoride meter, a lightning arrester meter and the like, and the accuracy test and algorithm optimization are carried out by more than 500 images, so that the identification effect of more than 90% accuracy of various meters can be finally achieved.
Figure 7 shows the effect of the algorithm operation. As shown in fig. 7, the system can accurately identify the information such as the measurement range, the pointer disk area, the pointer position, etc., and combine them to obtain the reading corresponding to the current pointer. When the reading exceeds the alarm threshold value, the reading can be accurately identified and an alarm can be given.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When used in whole or in part, can be implemented in a computer program product that includes one or more computer instructions. When loaded or executed on a computer, cause the flow or functions according to embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center via wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL), or wireless (e.g., infrared, wireless, microwave, etc.)). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A reading monitoring method for an electric power machine meter is characterized by comprising the following steps:
the user inputs 1 image of the front of each meter to be identified in advance, and inputs the type, number and alarm reading information of the meter;
collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
2. The method of monitoring reading of an electromechanical meter according to claim 1, wherein the identification of the electromechanical meter reading in the graph comprises: identifying the current mechanical meter reading by combining a plurality of deep learning networks; wherein the deep learning network comprises: the system comprises a target detection network, a digital identification network and an image segmentation network.
3. The method of monitoring reading of an electromechanical meter of claim 1, wherein the identifying a current mechanical meter reading in conjunction with a plurality of deep learning networks comprises:
(1) inputting the image to a target detection network;
(2) reading identifying the detected number: extracting the digital position obtained in the step (1) and inputting the digital position into a digital identification network;
(3) obtaining the position information of scales and pointers in the mechanical meter: extracting the meter position obtained in the step (1) and inputting the meter position into a segmentation network;
(4) after the results of the steps (1) to (3) are obtained, corresponding the detected numbers to the numbers on the template one by one through a preset template to generate a transformation matrix; the image is converted to a positive shooting state through a conversion matrix, and meanwhile, the number, the pointer and the scale are converted together;
(5) calculating principal component of the pointer information to obtain principal component direction information, and calculating intersection positions on a re-circle by combining a scale fitting circle obtained by dividing and the principal component direction of the pointer;
(6) calculating an included angle formed by the intersection point and the scale, and calculating the included angle to obtain a final result; and if the final result is compared with a preset alarm range, sending alarm information if alarm is needed, and if the alarm is not needed, acquiring the next frame of image for continuous identification.
4. The reading monitoring method of the electromechanical meter according to claim 3, wherein in step (1), the inputting the image to the target detection network comprises:
1) images resize to 800 × 800;
2) establishing a top-bottom network, performing continuous convolution, pooling and residual calculation in multiple stages, and constructing feature-maps with three different scales through a connection layer;
3) sending the obtained feature image data of three scales into corresponding activation;
4) and finally, performing post-processing such as inhibiting nms by a non-maximum value, comprehensively analyzing results of three scales, and mapping the results into an original image to obtain the type of the target and a corresponding coordinate position.
5. The electromechanical meter reading monitoring method of claim 3, wherein the identifying the reading of the detected number in step (2) comprises:
1) image resize to 100 x 32;
2) establishing a top-bottom network, constructing a feature-map through continuous convolution pooling of 5 stages, and serializing the feature-map;
3) inputting the serialized data into a double-circulation layer, and determining the label distribution of the characteristic sequence by a circulation network;
4) and inputting the obtained label distribution into a transcription layer, and obtaining a final result through de-duplication and integration operations.
6. The reading monitoring method for the electromechanical meter according to claim 3, wherein in the step (3), the obtaining of the position information of the scale and the pointer in the electromechanical meter comprises:
1) images resize to 512 x 512;
2) establishing a top-bottom network, performing convolution, activation and pooling in 5 continuous stages, performing convolution, activation and upsampling in 5 continuous stages, and simultaneously superposing the feature image after convolution into a feature image with the same size obtained by subsequent upsampling to generate a final feature-map;
3) and inputting the output feature map into the softmax activation layer to be converted into a segmentation result.
7. The method of monitoring reading of an electromechanical meter according to claim 1, further comprising:
(1) in the operation process, firstly, a camera acquires a current image frame in real time, and the position of an electric mechanical meter in the current image is determined through a detection model;
(2) extracting a roi partial image of the power machinery meter, and identifying a scale position, a pointer position and a number in the roi partial image; after the image is compared with a preset template, the image is corrected to be in a positive shooting state, and image deformation caused by the shooting position is reduced;
(3) analyzing the position information of scales, numbers and pointers in the corrected image to determine the reading; if the reading is larger than or equal to the preset alarm reading, sending information to a background for using the working state of the judgment equipment, and if the reading is smaller than the preset alarm reading, automatically extracting the next frame of image.
8. An electromechanical meter reading monitoring system for implementing the electromechanical meter reading monitoring method according to any one of claims 1 to 7, wherein the electromechanical meter reading monitoring system comprises: the system comprises a router, a plurality of cameras and a pc serving as a server side;
the router is respectively connected with the cameras and the server, and the router, the cameras and the server pc are connected through the router and are located in the same local area network.
9. A computer device, characterized in that the computer device comprises a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to carry out the steps of:
the user inputs 1 image of the front of each meter to be identified in advance, and inputs the type, number and alarm reading information of the meter;
collecting an image containing an electric mechanical meter through a camera, and inputting the image and the type of the meter into an algorithm;
the algorithm automatically identifies the reading of the electromechanical meter in the graph and compares the reading with a preset alarm reading, and if the reading is larger than or equal to the preset alarm reading, alarm information is sent to a background for a user to judge the running state of the current equipment.
10. An electromechanical meter, characterized in that the electromechanical meter is equipped with an electromechanical meter reading monitoring system according to claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110640232.3A CN113343998A (en) | 2021-06-08 | 2021-06-08 | Reading monitoring system and method for electric power mechanical meter, computer equipment and application |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110640232.3A CN113343998A (en) | 2021-06-08 | 2021-06-08 | Reading monitoring system and method for electric power mechanical meter, computer equipment and application |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113343998A true CN113343998A (en) | 2021-09-03 |
Family
ID=77475547
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110640232.3A Pending CN113343998A (en) | 2021-06-08 | 2021-06-08 | Reading monitoring system and method for electric power mechanical meter, computer equipment and application |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113343998A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743397A (en) * | 2021-11-08 | 2021-12-03 | 深圳市信润富联数字科技有限公司 | Data processing method and system for industrial instrument, terminal device and storage medium |
| CN113837312A (en) * | 2021-09-30 | 2021-12-24 | 南方电网电力科技股份有限公司 | Method and device for evaluating running state of zinc oxide arrester |
| CN115496807A (en) * | 2022-11-18 | 2022-12-20 | 南方电网数字电网研究院有限公司 | Meter pointer positioning method and device, computer equipment and storage medium |
-
2021
- 2021-06-08 CN CN202110640232.3A patent/CN113343998A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113837312A (en) * | 2021-09-30 | 2021-12-24 | 南方电网电力科技股份有限公司 | Method and device for evaluating running state of zinc oxide arrester |
| CN113743397A (en) * | 2021-11-08 | 2021-12-03 | 深圳市信润富联数字科技有限公司 | Data processing method and system for industrial instrument, terminal device and storage medium |
| CN115496807A (en) * | 2022-11-18 | 2022-12-20 | 南方电网数字电网研究院有限公司 | Meter pointer positioning method and device, computer equipment and storage medium |
| CN115496807B (en) * | 2022-11-18 | 2023-01-20 | 南方电网数字电网研究院有限公司 | Meter pointer positioning method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113343998A (en) | Reading monitoring system and method for electric power mechanical meter, computer equipment and application | |
| CN111555716B (en) | Method, device, equipment and storage medium for determining working state of photovoltaic array | |
| CN110826538B (en) | Abnormal off-duty identification system for electric power business hall | |
| CN108416294B (en) | Fan blade fault intelligent identification method based on deep learning | |
| CN110084165B (en) | Intelligent identification and early warning method for abnormal events in open scene of power field based on edge calculation | |
| CN111179249A (en) | Power equipment detection method and device based on deep convolutional neural network | |
| CN111582235A (en) | Alarm method, system and equipment for monitoring abnormal events in station in real time | |
| CN111401418A (en) | Employee dressing specification detection method based on improved Faster r-cnn | |
| CN113688817A (en) | Instrument identification method and system for automatic inspection | |
| CN113963154A (en) | Method, system, storage medium and equipment for detecting gas leakage of VOCs (volatile organic compounds) based on Faster R-CNN (fast R-CNN) | |
| CN114119518A (en) | Method and system for detecting temperature abnormal point in infrared image of current transformer | |
| CN115239646A (en) | Defect detection method and device for power transmission line, electronic equipment and storage medium | |
| CN117114420B (en) | Image recognition-based industrial and trade safety accident risk management and control system and method | |
| WO2024119983A1 (en) | Anomality detection method and apparatus for cable terminal of rail vehicle | |
| CN116311034A (en) | Robot inspection system based on contrast detection | |
| CN114140731B (en) | Traction substation abnormality detection method | |
| CN116091976A (en) | Station room defect identification detection method, system, device and storage medium | |
| CN115187880A (en) | Communication optical cable defect detection method and system based on image recognition and storage medium | |
| CN114821806A (en) | Method and device for determining behavior of operator, electronic equipment and storage medium | |
| CN113850773A (en) | Detection method, device, equipment and computer readable storage medium | |
| CN115018332B (en) | Visual management system of thermal power plant based on virtual reality | |
| CN114429421A (en) | Method for detecting state of disconnecting link facing scheduling service scene | |
| CN116432078A (en) | Building electromechanical equipment monitoring system | |
| CN112819183A (en) | Algorithm and system for intelligently distinguishing heating defects of power transmission and distribution line | |
| CN117218436A (en) | Transformer appearance defect identification method based on oblique photographing, three-dimensional reconstruction and yolov5 algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |