CN113033322A - Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning - Google Patents
Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning Download PDFInfo
- Publication number
- CN113033322A CN113033322A CN202110233358.9A CN202110233358A CN113033322A CN 113033322 A CN113033322 A CN 113033322A CN 202110233358 A CN202110233358 A CN 202110233358A CN 113033322 A CN113033322 A CN 113033322A
- Authority
- CN
- China
- Prior art keywords
- image
- oil leakage
- equipment
- oil
- model
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention discloses a method for identifying hidden oil leakage hazards of transformer substation oil filling equipment based on deep learning, which comprises the following steps of: step A, collecting potential oil leakage hazard image samples of the power transformation equipment; b, expanding an image sample of the potential oil leakage hazard of the power transformation equipment; c, marking the severity of the sample with the oil leakage hidden danger of the power transformation equipment; step D, training a detection model for the potential oil leakage hazard of the power transformation equipment; e, deploying a detection model for the potential oil leakage hazard of the power transformation equipment; and F, detecting the severity of the oil leakage hidden danger of the patrol image of the transformer substation. The method can automatically, quickly and accurately identify the oil leakage condition of the oil filling equipment in the substation patrol image, and greatly improve the substation patrol working efficiency.
Description
Technical Field
The invention relates to a method for identifying hidden oil leakage hazards of oil filling equipment of a transformer substation.
Background
The transformer substation is an important part in a power system, and only can ensure the stable operation of the whole power grid as a basic fulcrum of the whole power grid framework, so that the safe operation of the whole power grid can be ensured, at present, many important devices such as a transformer, a mutual inductor, a capacitor and the like in the transformer substation are oil-filled devices, oil leakage can occur in the operation process of the devices due to the reasons of improper construction and installation, operation aging, environmental factors and the like, slight oil leakage can be developed into emergency hidden dangers to force the non-planned power failure of the transformer substation, particularly, the potential safety hazards of the operation of the oil-filled devices of an unattended transformer substation are large, the devices can be insufficiently insulated and tripped due to grounding short circuit if the situation is not timely processed, and even the devices.
In order to ensure the reliable operation of a power grid, a substation is required to be regularly patrolled to find equipment defects and hidden dangers threatening the safe operation of the equipment in time, the automation degree of the conventional substation patrol system is limited, patrol images are judged and identified by human intervention, however, the slight oil leakage condition is difficult to find by manual judgment, the oil leakage condition can only be subjectively estimated, a reasonable quantitative evaluation is difficult to achieve, meanwhile, the number of substation equipment is large, the patrol workload is large, and the hidden dangers are difficult to find by manual judgment in time.
Disclosure of Invention
The invention aims to provide a method for identifying hidden oil leakage danger of oil filling equipment of a transformer substation based on deep learning, which can automatically, quickly and accurately identify the oil leakage situation of the oil filling equipment in a patrol image of the transformer substation and greatly improve the patrol working efficiency of the transformer substation.
The technical solution of the invention is as follows:
a method for identifying hidden oil leakage hazards of transformer substation oil filling equipment based on deep learning is characterized by comprising the following steps: the method comprises the following steps:
step A, collecting potential oil leakage hazard image samples of the power transformation equipment;
b, expanding an image sample of the potential oil leakage hazard of the power transformation equipment;
c, marking the severity of the sample with the oil leakage hidden danger of the power transformation equipment;
step D, training a detection model for the potential oil leakage hazard of the power transformation equipment;
e, deploying a detection model for the potential oil leakage hazard of the power transformation equipment;
and F, detecting the severity of the oil leakage hidden danger of the patrol image of the transformer substation.
The step A comprises the following steps:
acquiring a patrol image and a video of the power transformation equipment, storing a patrol image sample with the oil leakage condition of the equipment, extracting an oil leakage image frame from the video containing the oil leakage condition of the equipment, storing the image sample, and removing a data sample with poor quality; the method comprises the following steps: (1) a sample in which the pixel area occupied by the oil leakage area in the scene image is less than 5%; (2) only less than 10% of the leakage oil area is displayed in the scene image, thereby constituting a preliminary leakage oil image sample set.
The step B comprises the following steps:
step B-1: simulating oil leakage scene images of oil filling equipment, simulating slight, medium and severe oil leakage conditions of the equipment by smearing insulating oil on potential oil leakage parts of the oil filling equipment during safety shutdown, shooting and collecting the oil leakage, ensuring that the oil leakage falls in a central area of a visual field of the camera equipment when shooting the oil leakage, respectively shooting the oil leakage from different distances and angles, and obtaining images with different brightness by adjusting an aperture of the equipment to improve the number and diversity of oil leakage sample image samples;
step B-2: increasing the number of the obtained oil leakage samples of the oil-filled equipment by using a turning, cutting and scaling method; the method for processing the oil leakage trace image through overturning comprises the following steps: turning the image every 30 degrees to obtain a new image; the method for processing the oil leakage body image by cutting comprises the following steps: cutting the obtained oil leakage body image randomly, cutting 20%, 40% and 60% of the image from the upper, lower, left and right sides of the image respectively, and finally obtaining 12 new cut images from each oil leakage image; the method of performing image processing of the oil leakage sample by scaling is as follows: respectively changing the leakage oil image into 25%, 50% and 200% of the original size to obtain new images;
step B-3: for the obtained oil leakage sample image, image noise is introduced to increase the number of image samples, and the specific method is as follows: the image is processed by additive noise, the formula is as follows: f (x, y) ═ g (x, y) + q (), where x and y represent the horizontal and vertical coordinates of the pixels in the image, g (x, y) represents the true value of each pixel, g (x, y) represents the vector corresponding to each pixel for the multichannel image, q () represents the noise function, and f (x, y) represents the value of each pixel of the image after noise introduction, where the noise is introduced by the probability density function of gaussian random variables, represented by the following formula:
wherein: z is a variable, P (z) is the probability density function of z, μ is the expectation of z, and σ is the variance of z.
The step C comprises the following steps:
for each image, marking out all oil leakage areas by using a minimum rectangle under the condition that the target of the oil leakage area of the equipment is completely displayed in the image; and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that only one part of the oil leakage area of the equipment is displayed in the image, and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that the oil leakage area of the equipment is partially shielded, wherein each oil leakage area in the image is marked with a slight label, a moderate label and a severe label according to the severity of the oil leakage.
The step D comprises the following steps:
the model training is an iterative loop process, and each iteration comprises the main steps of data splitting, algorithm parameter adjustment, model training, model verification and the like;
step D-1: data splitting: dividing all images into three parts, namely a training set, a verification set and a test set respectively, and enabling the number of image samples of the three subsets to be 70%, 20% and 10% respectively;
step D-2: adjusting parameters in an algorithm: selecting a target detection algorithm based on a deep learning algorithm, and carrying out preliminary adjustment on parameters of the algorithm;
step D-3: training a model: training the algorithm on the basis of the training set images to obtain a target detection model of the oil leakage of the substation equipment;
step D-4: and (3) model verification: on the basis of the verification set, the accuracy of the target detection model is evaluated, if the accuracy does not meet the requirement, the parameters of the target detection algorithm are adjusted, the training set is used for model training, and the verification set is used for evaluating the accuracy of the model; if the accuracy meets the requirement, the next step is carried out, if the accuracy does not meet the requirement, the algorithm parameter adjusting step is returned again, and the operation in the previous step is repeated until the accuracy meets the condition; and B, evaluating the accuracy of the model again on the basis of the test set, if the accuracy meets the requirement, keeping the model, if the accuracy cannot meet the requirement, re-expanding and optimizing the quality of the sample through the step B, and finishing the training of the model according to the flow in sequence until the model finally meets the accuracy requirement.
The step E comprises the following steps:
the trained model is deployed on hardware equipment to provide service for oil leakage oil recognition of the transformer substation equipment in the later period, and the model is deployed mainly on a transformer substation inspection robot body and a transformer substation fixed camera body.
The step F comprises the following steps:
the method mainly comprises the steps of acquiring a robot patrol video and a fixed camera video in real time, extracting frames from the videos, processing the images and detecting the images;
step F-1: real-time collection of a robot tour video and a fixed camera video; in the actual process of tour and monitoring, no matter the robot or the fixed camera shoots real-time videos, the real-time videos need to be transmitted to the intelligent identification module;
step F-2: video frame extraction: the module then frames the video at fixed intervals, the specific length of the intervals depends on the performance of the module and the computing power of the hardware carrier;
step F-3: image processing: the frame image needs image enhancement, and the specific method comprises linear gray level enhancement and logarithmic function nonlinear transformation; due to the fact that the quality of the frame image is poor and the identification accuracy of the image intelligent module is high in the case of the image under the shooting condition, image enhancement needs to be performed on each frame image, and the specific method comprises linear gray scale enhancement and logarithmic function nonlinear transformation;
the method of linear gray scale enhancement is as follows,
wherein g (x, y) is the true value of each pixel, [ a, b ] is the gray scale range of the original image, [ c, d ] is the expected range of the gray scale of the new image, and f (x, y) is the pixel value of the changed image;
the logarithmic function non-linear transformation formula is as follows,
a, b and c are adjustable parameters, and the gray scale range of a newly generated image can be within an ideal range after iterative correction; if the processed image is a color image, the conversion from color to gray needs to be carried out before the image enhancement, and the conversion can be realized by using a built-in method of opencv, and the enhanced gray image can be converted back to the color image by using a method provided by opencv;
step F-4: and transmitting the corrected picture to an image identification module, detecting whether the image contains the oil leakage condition of the equipment, the specific pixel position of the oil leakage area and the severity level of the oil leakage condition, and storing an oil leakage detection result image.
The invention has the following beneficial effects: (1) according to the invention, an image sample set containing three levels of slight, moderate and severe oil leakage situations is constructed by methods of historical images, field simulation acquisition, turning, cutting, scaling, noise introduction and the like, so that the sufficiency of training samples is ensured, and the robustness of a training model is ensured. (2) The invention well divides the severity of the oil leakage condition of the equipment into three grades of slight, moderate and severe, and is convenient for inspection personnel to adopt different treatment schemes according to the oil leakage condition of the equipment. (3) The method deploys the trained equipment oil leakage hidden danger identification model on equipment such as a fixed camera, an inspection robot and the like, and can automatically, quickly and accurately identify the oil leakage hidden danger of the equipment.
Drawings
The invention is further illustrated by the following figures and examples.
FIG. 1 is a schematic diagram of the working steps of the identification method of the present invention.
Detailed Description
A method for identifying hidden oil leakage hazards of transformer substation oil filling equipment based on deep learning comprises the following steps:
step A, collecting potential oil leakage hazard image samples of the power transformation equipment;
b, expanding an image sample of the potential oil leakage hazard of the power transformation equipment;
c, marking the severity of the sample with the oil leakage hidden danger of the power transformation equipment;
step D, training a detection model for the potential oil leakage hazard of the power transformation equipment;
e, deploying a detection model for the potential oil leakage hazard of the power transformation equipment;
and F, detecting the severity of the oil leakage hidden danger of the patrol image of the transformer substation.
The step A comprises the following steps:
acquiring a patrol image and a video of the power transformation equipment, storing a patrol image sample with the oil leakage condition of the equipment, extracting an oil leakage image frame from the video containing the oil leakage condition of the equipment, storing the image sample, and removing a data sample with poor quality; the method comprises the following steps: (1) a sample in which the pixel area occupied by the oil leakage area in the scene image is less than 5%; (2) only less than 10% of the leakage oil area is displayed in the scene image, thereby constituting a preliminary leakage oil image sample set.
The step B comprises the following steps:
step B-1: simulating oil leakage scene images of oil filling equipment, simulating slight, medium and severe oil leakage conditions of the equipment by smearing insulating oil on potential oil leakage parts of the oil filling equipment during safety shutdown, shooting and collecting the oil leakage, ensuring that the oil leakage falls in a central area of a visual field of the camera equipment when shooting the oil leakage, respectively shooting the oil leakage from different distances and angles, and obtaining images with different brightness by adjusting an aperture of the equipment to improve the number and diversity of oil leakage sample image samples;
step B-2: increasing the number of the obtained oil leakage samples of the oil-filled equipment by using a turning, cutting and scaling method; the method for processing the oil leakage trace image through overturning comprises the following steps: turning the image every 30 degrees to obtain a new image; the method for processing the oil leakage body image by cutting comprises the following steps: cutting the obtained oil leakage body image randomly, cutting 20%, 40% and 60% of the image from the upper, lower, left and right sides of the image respectively, and finally obtaining 12 new cut images from each oil leakage image; the method of performing image processing of the oil leakage sample by scaling is as follows: respectively changing the leakage oil image into 25%, 50% and 200% of the original size to obtain new images;
step B-3: for the obtained oil leakage sample image, image noise is introduced to increase the number of image samples, and the specific method is as follows: the image is processed by additive noise, the formula is as follows: f (x, y) ═ g (x, y) + q (), where x and y represent the horizontal and vertical coordinates of the pixels in the image, g (x, y) represents the true value of each pixel, g (x, y) represents the vector corresponding to each pixel for the multichannel image, q () represents the noise function, and f (x, y) represents the value of each pixel of the image after noise introduction, where the noise is introduced by the probability density function of gaussian random variables, represented by the following formula:
wherein: z is a variable, P (z) is the probability density function of z, μ is the expectation of z, and σ is the variance of z.
The step C comprises the following steps:
for each image, marking out all oil leakage areas by using a minimum rectangle under the condition that the target of the oil leakage area of the equipment is completely displayed in the image; and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that only one part of the oil leakage area of the equipment is displayed in the image, and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that the oil leakage area of the equipment is partially shielded, wherein each oil leakage area in the image is marked with a slight label, a moderate label and a severe label according to the severity of the oil leakage.
The step D comprises the following steps:
the model training is an iterative loop process, and each iteration comprises the main steps of data splitting, algorithm parameter adjustment, model training, model verification and the like;
step D-1: data splitting: dividing all images into three parts, namely a training set, a verification set and a test set respectively, and enabling the number of image samples of the three subsets to be 70%, 20% and 10% respectively;
step D-2: adjusting parameters in an algorithm: selecting a target detection algorithm based on a deep learning algorithm, and carrying out preliminary adjustment on parameters of the algorithm;
step D-3: training a model: training the algorithm on the basis of the training set images to obtain a target detection model of the oil leakage of the substation equipment;
step D-4: and (3) model verification: on the basis of the verification set, the accuracy of the target detection model is evaluated, if the accuracy does not meet the requirement, the parameters of the target detection algorithm are adjusted, the training set is used for model training, and the verification set is used for evaluating the accuracy of the model; if the accuracy meets the requirement, the next step is carried out, if the accuracy does not meet the requirement, the algorithm parameter adjusting step is returned again, and the operation in the previous step is repeated until the accuracy meets the condition; and B, evaluating the accuracy of the model again on the basis of the test set, if the accuracy meets the requirement, keeping the model, if the accuracy cannot meet the requirement, re-expanding and optimizing the quality of the sample through the step B, and finishing the training of the model according to the flow in sequence until the model finally meets the accuracy requirement.
The step E comprises the following steps:
the trained model is deployed on hardware equipment to provide service for oil leakage oil recognition of the transformer substation equipment in the later period, and the model is deployed mainly on a transformer substation inspection robot body and a transformer substation fixed camera body.
The step F comprises the following steps:
the method mainly comprises the steps of acquiring a robot patrol video and a fixed camera video in real time, extracting frames from the videos, processing the images and detecting the images;
step F-1: real-time collection of a robot tour video and a fixed camera video; in the actual process of tour and monitoring, no matter the robot or the fixed camera shoots real-time videos, the real-time videos need to be transmitted to the intelligent identification module;
step F-2: video frame extraction: the module then frames the video at fixed intervals, the specific length of the intervals depends on the performance of the module and the computing power of the hardware carrier;
step F-3: image processing: the frame image needs image enhancement, and the specific method comprises linear gray level enhancement and logarithmic function nonlinear transformation; due to the fact that the quality of the frame image is poor and the identification accuracy of the image intelligent module is high in the case of the image under the shooting condition, image enhancement needs to be performed on each frame image, and the specific method comprises linear gray scale enhancement and logarithmic function nonlinear transformation;
the method of linear gray scale enhancement is as follows,
wherein g (x, y) is the true value of each pixel, [ a, b ] is the gray scale range of the original image, [ c, d ] is the expected range of the gray scale of the new image, and f (x, y) is the pixel value of the changed image;
the logarithmic function non-linear transformation formula is as follows,
a, b and c are adjustable parameters, and the gray scale range of a newly generated image can be within an ideal range after iterative correction; if the processed image is a color image, the conversion from color to gray needs to be carried out before the image enhancement, and the conversion can be realized by using a built-in method of opencv, and the enhanced gray image can be converted back to the color image by using a method provided by opencv;
step F-4: and transmitting the corrected picture to an image identification module, detecting whether the image contains the oil leakage condition of the equipment, the specific pixel position of the oil leakage area and the severity level of the oil leakage condition, and storing an oil leakage detection result image.
Claims (7)
1. A method for identifying hidden oil leakage hazards of transformer substation oil filling equipment based on deep learning is characterized by comprising the following steps: the method comprises the following steps:
step A, collecting potential oil leakage hazard image samples of the power transformation equipment;
b, expanding an image sample of the potential oil leakage hazard of the power transformation equipment;
c, marking the severity of the sample with the oil leakage hidden danger of the power transformation equipment;
step D, training a detection model for the potential oil leakage hazard of the power transformation equipment;
e, deploying a detection model for the potential oil leakage hazard of the power transformation equipment;
and F, detecting the severity of the oil leakage hidden danger of the patrol image of the transformer substation.
2. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step A comprises the following steps:
acquiring a patrol image and a video of the power transformation equipment, storing a patrol image sample with the oil leakage condition of the equipment, extracting an oil leakage image frame from the video containing the oil leakage condition of the equipment, storing the image sample, and removing a data sample with poor quality; the method comprises the following steps: (1) a sample in which the pixel area occupied by the oil leakage area in the scene image is less than 5%; (2) only less than 10% of the leakage oil area is displayed in the scene image, thereby constituting a preliminary leakage oil image sample set.
3. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step B comprises the following steps:
step B-1: simulating oil leakage scene images of oil filling equipment, simulating slight, medium and severe oil leakage conditions of the equipment by smearing insulating oil on potential oil leakage parts of the oil filling equipment during safety shutdown, shooting and collecting the oil leakage, ensuring that the oil leakage falls in a central area of a visual field of the camera equipment when shooting the oil leakage, respectively shooting the oil leakage from different distances and angles, and obtaining images with different brightness by adjusting an aperture of the equipment to improve the number and diversity of oil leakage sample image samples;
step B-2: increasing the number of the obtained oil leakage samples of the oil-filled equipment by using a turning, cutting and scaling method; the method for processing the oil leakage trace image through overturning comprises the following steps: turning the image every 30 degrees to obtain a new image; the method for processing the oil leakage body image by cutting comprises the following steps: cutting the obtained oil leakage body image randomly, cutting 20%, 40% and 60% of the image from the upper, lower, left and right sides of the image respectively, and finally obtaining 12 new cut images from each oil leakage image; the method of performing image processing of the oil leakage sample by scaling is as follows: respectively changing the leakage oil image into 25%, 50% and 200% of the original size to obtain new images;
step B-3: for the obtained oil leakage sample image, image noise is introduced to increase the number of image samples, and the specific method is as follows: the image is processed by additive noise, the formula is as follows: f (x, y) ═ g (x, y) + q (), where x and y represent the horizontal and vertical coordinates of the pixels in the image, g (x, y) represents the true value of each pixel, g (x, y) represents the vector corresponding to each pixel for the multichannel image, q () represents the noise function, and f (x, y) represents the value of each pixel of the image after noise introduction, where the noise is introduced by the probability density function of gaussian random variables, represented by the following formula:
wherein: z is a variable, P (z) is the probability density function of z, μ is the expectation of z, and σ is the variance of z.
4. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step C comprises the following steps:
for each image, marking out all oil leakage areas by using a minimum rectangle under the condition that the target of the oil leakage area of the equipment is completely displayed in the image; and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that only one part of the oil leakage area of the equipment is displayed in the image, and marking the visible part of the oil leakage area of the equipment by using the minimum rectangle under the condition that the oil leakage area of the equipment is partially shielded, wherein each oil leakage area in the image is marked with a slight label, a moderate label and a severe label according to the severity of the oil leakage.
5. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step D comprises the following steps:
the model training is an iterative loop process, and each iteration comprises the main steps of data splitting, algorithm parameter adjustment, model training, model verification and the like;
step D-1: data splitting: dividing all images into three parts, namely a training set, a verification set and a test set respectively, and enabling the number of image samples of the three subsets to be 70%, 20% and 10% respectively;
step D-2: adjusting parameters in an algorithm: selecting a target detection algorithm based on a deep learning algorithm, and carrying out preliminary adjustment on parameters of the algorithm;
step D-3: training a model: training the algorithm on the basis of the training set images to obtain a target detection model of the oil leakage of the substation equipment;
step D-4: and (3) model verification: on the basis of the verification set, the accuracy of the target detection model is evaluated, if the accuracy does not meet the requirement, the parameters of the target detection algorithm are adjusted, the training set is used for model training, and the verification set is used for evaluating the accuracy of the model; if the accuracy meets the requirement, the next step is carried out, if the accuracy does not meet the requirement, the algorithm parameter adjusting step is returned again, and the operation in the previous step is repeated until the accuracy meets the condition; and B, evaluating the accuracy of the model again on the basis of the test set, if the accuracy meets the requirement, keeping the model, if the accuracy cannot meet the requirement, re-expanding and optimizing the quality of the sample through the step B, and finishing the training of the model according to the flow in sequence until the model finally meets the accuracy requirement.
6. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step E comprises the following steps:
the trained model is deployed on hardware equipment to provide service for oil leakage oil recognition of the transformer substation equipment in the later period, and the model is deployed mainly on a transformer substation inspection robot body and a transformer substation fixed camera body.
7. The deep learning-based substation oil filling equipment oil leakage hidden danger identification method according to claim 1 is characterized in that: the step F comprises the following steps:
the method mainly comprises the steps of acquiring a robot patrol video and a fixed camera video in real time, extracting frames from the videos, processing the images and detecting the images;
step F-1: real-time collection of a robot tour video and a fixed camera video; in the actual process of tour and monitoring, no matter the robot or the fixed camera shoots real-time videos, the real-time videos need to be transmitted to the intelligent identification module;
step F-2: video frame extraction: the module then frames the video at fixed intervals, the specific length of the intervals depends on the performance of the module and the computing power of the hardware carrier;
step F-3: image processing: the frame image needs image enhancement, and the specific method comprises linear gray level enhancement and logarithmic function nonlinear transformation; due to the fact that the quality of the frame image is poor and the identification accuracy of the image intelligent module is high in the case of the image under the shooting condition, image enhancement needs to be performed on each frame image, and the specific method comprises linear gray scale enhancement and logarithmic function nonlinear transformation;
the method of linear gray scale enhancement is as follows,
wherein g (x, y) is the true value of each pixel, [ a, b ] is the gray scale range of the original image, [ c, d ] is the expected range of the gray scale of the new image, and f (x, y) is the pixel value of the changed image;
the logarithmic function non-linear transformation formula is as follows,
a, b and c are adjustable parameters, and the gray scale range of a newly generated image can be within an ideal range after iterative correction; if the processed image is a color image, the conversion from color to gray needs to be carried out before the image enhancement, and the conversion can be realized by using a built-in method of opencv, and the enhanced gray image can be converted back to the color image by using a method provided by opencv;
step F-4: and transmitting the corrected picture to an image identification module, detecting whether the image contains the oil leakage condition of the equipment, the specific pixel position of the oil leakage area and the severity level of the oil leakage condition, and storing an oil leakage detection result image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110233358.9A CN113033322A (en) | 2021-03-02 | 2021-03-02 | Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110233358.9A CN113033322A (en) | 2021-03-02 | 2021-03-02 | Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113033322A true CN113033322A (en) | 2021-06-25 |
Family
ID=76465965
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110233358.9A Pending CN113033322A (en) | 2021-03-02 | 2021-03-02 | Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113033322A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743454A (en) * | 2021-07-22 | 2021-12-03 | 南方电网深圳数字电网研究院有限公司 | Detection method, device and equipment of oil-immersed transformer and storage medium |
| CN113919395A (en) * | 2021-10-12 | 2022-01-11 | 大连理工大学 | Water supply pipe network leakage accident diagnosis method based on one-dimensional convolutional neural network |
| CN114062224A (en) * | 2021-11-19 | 2022-02-18 | 广东电网有限责任公司广州供电局 | Oil leakage oil detection system and method for oil-immersed power equipment |
| CN114937247A (en) * | 2022-07-21 | 2022-08-23 | 四川金信石信息技术有限公司 | Transformer substation monitoring method and system based on deep learning and electronic equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110197259A (en) * | 2019-05-30 | 2019-09-03 | 浙江大学 | Wafer defect detection method based on small lot data set Yu deep learning algorithm |
| CN110705601A (en) * | 2019-09-09 | 2020-01-17 | 安徽继远软件有限公司 | Transformer substation equipment oil leakage image identification method based on single-stage target detection |
| CN110807353A (en) * | 2019-09-03 | 2020-02-18 | 国网辽宁省电力有限公司电力科学研究院 | Transformer substation foreign matter identification method, device and system based on deep learning |
| CN111191546A (en) * | 2019-12-20 | 2020-05-22 | 广西柳州联耕科技有限公司 | Intelligent product assembling method based on machine vision recognition |
| CN111830051A (en) * | 2020-08-14 | 2020-10-27 | 成都理工大学 | Transformer substation equipment oil leakage detection method and detection system based on deep learning |
| CN112036463A (en) * | 2020-08-26 | 2020-12-04 | 国家电网有限公司 | Power equipment defect detection and identification method based on deep learning |
| CN112101181A (en) * | 2020-09-10 | 2020-12-18 | 湖北烽火平安智能消防科技有限公司 | Automatic hidden danger scene recognition method and system based on deep learning |
-
2021
- 2021-03-02 CN CN202110233358.9A patent/CN113033322A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110197259A (en) * | 2019-05-30 | 2019-09-03 | 浙江大学 | Wafer defect detection method based on small lot data set Yu deep learning algorithm |
| CN110807353A (en) * | 2019-09-03 | 2020-02-18 | 国网辽宁省电力有限公司电力科学研究院 | Transformer substation foreign matter identification method, device and system based on deep learning |
| CN110705601A (en) * | 2019-09-09 | 2020-01-17 | 安徽继远软件有限公司 | Transformer substation equipment oil leakage image identification method based on single-stage target detection |
| CN111191546A (en) * | 2019-12-20 | 2020-05-22 | 广西柳州联耕科技有限公司 | Intelligent product assembling method based on machine vision recognition |
| CN111830051A (en) * | 2020-08-14 | 2020-10-27 | 成都理工大学 | Transformer substation equipment oil leakage detection method and detection system based on deep learning |
| CN112036463A (en) * | 2020-08-26 | 2020-12-04 | 国家电网有限公司 | Power equipment defect detection and identification method based on deep learning |
| CN112101181A (en) * | 2020-09-10 | 2020-12-18 | 湖北烽火平安智能消防科技有限公司 | Automatic hidden danger scene recognition method and system based on deep learning |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743454A (en) * | 2021-07-22 | 2021-12-03 | 南方电网深圳数字电网研究院有限公司 | Detection method, device and equipment of oil-immersed transformer and storage medium |
| CN113919395A (en) * | 2021-10-12 | 2022-01-11 | 大连理工大学 | Water supply pipe network leakage accident diagnosis method based on one-dimensional convolutional neural network |
| CN114062224A (en) * | 2021-11-19 | 2022-02-18 | 广东电网有限责任公司广州供电局 | Oil leakage oil detection system and method for oil-immersed power equipment |
| CN114937247A (en) * | 2022-07-21 | 2022-08-23 | 四川金信石信息技术有限公司 | Transformer substation monitoring method and system based on deep learning and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113033322A (en) | Method for identifying hidden danger of oil leakage of transformer substation oil filling equipment based on deep learning | |
| CN109753929B (en) | High-speed rail insulator inspection image recognition method | |
| CN111339882B (en) | Power transmission line hidden danger detection method based on example segmentation | |
| CN105447471A (en) | Infrared detection based device gas leakage identification method and apparatus | |
| CN106407928A (en) | Transformer composite insulator bushing monitoring method and transformer composite insulator bushing monitoring system based on raindrop identification | |
| CN104268505A (en) | Automatic cloth defect point detection and recognition device and method based on machine vision | |
| CN112288711A (en) | Unmanned aerial vehicle inspection image defect image identification method, device, equipment and medium | |
| CN106845434B (en) | Image type machine room water leakage monitoring method based on support vector machine | |
| CN111639530B (en) | Method and system for detecting and identifying power transmission tower and insulator of power transmission line | |
| CN114241364A (en) | Method for quickly calibrating foreign object target of overhead transmission line | |
| CN106951863B (en) | Method for detecting change of infrared image of substation equipment based on random forest | |
| CN109509188B (en) | Power transmission line typical defect identification method based on HOG characteristics | |
| CN112561899A (en) | Electric power inspection image identification method | |
| CN114648714A (en) | YOLO-based workshop normative behavior monitoring method | |
| CN108665468B (en) | Device and method for extracting tangent tower insulator string | |
| CN113298077A (en) | Transformer substation foreign matter identification and positioning method and device based on deep learning | |
| CN113421223B (en) | Industrial product surface defect detection method based on deep learning and Gaussian mixture | |
| CN107977959B (en) | Respirator state identification method suitable for electric power robot | |
| CN109272484B (en) | Rainfall detection method based on video image | |
| CN112634179B (en) | Camera shake prevention power transformation equipment image change detection method and system | |
| CN111830051A (en) | Transformer substation equipment oil leakage detection method and detection system based on deep learning | |
| CN108693441B (en) | Method and system for identifying insulator of power transmission line | |
| CN116612065A (en) | YOLO v 5-based intelligent identification method for transmission line inspection image defects | |
| CN109325441B (en) | Method for identifying insulator object of power transmission line | |
| CN112241707A (en) | Wind-powered electricity generation field intelligence video identification device |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210625 |
|
| RJ01 | Rejection of invention patent application after publication |