CN114998766A

CN114998766A - Image recognition-based power high-voltage equipment fault detection method

Info

Publication number: CN114998766A
Application number: CN202210536187.1A
Authority: CN
Inventors: 张小陆; 沈伍强; 张金波; 沈桂泉; 崔磊; 梁哲恒; 曾纪钧
Original assignee: Guangdong Power Grid Co Ltd
Current assignee: Guangdong Power Grid Co Ltd
Priority date: 2022-05-17
Filing date: 2022-05-17
Publication date: 2022-09-02

Abstract

The invention provides an electric power high-voltage equipment fault detection method based on image recognition, which is applied to a power distribution network and comprises the following steps: (A) the central control platform issues a shooting instruction to the nearby unmanned aerial vehicle, and the unmanned aerial vehicle goes to the vicinity of the electric power high-voltage equipment and adopts a high-definition camera to shoot images originally and transmits the images back to the central control platform; (B) the central control platform carries out image detection on the original shot image, generates a shooting adjustment instruction and transmits the shooting adjustment instruction back to the unmanned aerial vehicle; (C) the unmanned aerial vehicle adjusts the shooting position of the thermal imaging camera according to the shooting adjustment instruction; (D) the central control platform adopts a filtering algorithm to preprocess the thermal imaging image to remove image noise pixel points; (E) and identifying the specific temperature of the component in the thermal imaging image, forming an analysis report according to the temperature value, and sending the analysis report to a maintenance department for maintenance. The method disclosed by the invention can realize higher efficiency and accuracy in fault detection of the high-voltage power equipment.

Description

Image recognition-based power high-voltage equipment fault detection method

Technical Field

The invention relates to safety detection of power equipment, in particular to a fault detection method of power high-voltage equipment based on image identification.

Background

With the continuous expansion of the national grid scale and the continuous improvement of the voltage grade, the normal operation of a power system, which can not be safely and reliably protected by a high-voltage power device, is one of the important research subjects facing the present. Under the long-term operation condition of the electric high-voltage equipment, a series of fault problems can be caused due to corrosion, oxidation, mechanical faults and the like, the electric high-voltage equipment is often accompanied by the phenomenon that the temperature of a fault part is abnormally increased when the electric high-voltage equipment breaks down, and if the electric high-voltage equipment is not found and eliminated in time, serious economic loss can be caused. Therefore, whether the equipment is in a fault state or not can be observed and analyzed through the change of the temperature information of the high-voltage power equipment, and the method has extremely important significance for the operation and maintenance of the power system.

The high-voltage power equipment has a very important position in a power system, has the main function of controlling and protecting the power system, can complete the quitting and the putting in of a power line and equipment according to an operation instruction, can remove faults in the system and ensures the normal operation of other equipment. If the high-voltage power equipment fails, immeasurable loss is brought to the power system. The familiar high-voltage power equipment generally includes high-voltage circuit breakers, high-voltage load switches, high-voltage fuses, high-voltage isolating switches, high-voltage switch cabinets and the like. The high-voltage circuit breaker is used as important equipment for protecting and controlling a power system, has the function of finishing the command of cutting off or putting into operation, and can quickly cut off the equipment or the line with faults. If the fault occurs in the power system, the power system is difficult to effectively protect, and serious accidents occur. Therefore, the high-voltage circuit breaker is subjected to fault diagnosis periodically. The high-voltage isolating switch is widely used in a power grid, is primary equipment with the most extensive application, has a simple production process, and does not draw sufficient attention of manufacturers. Because the isolating switch is often exposed to the outdoor environment for operation, the problems of long-term repair, untimely overhaul and the like exist, and the isolating switch has a fault defect. The high-voltage isolating switch is a key device for the safe operation of a power grid, and the working condition of the high-voltage isolating switch directly influences the safety and stability of the power grid. The high-voltage load switch is an electric appliance with the function between a high-voltage circuit breaker and a high-voltage isolating switch, is often used in series connection with a high-voltage fuse for controlling a power transformer, has a simple arc-extinguishing device, can break and make certain load current and overload current, but cannot break short-circuit current, so the high-voltage load switch is generally used in series connection with the high-voltage fuse to carry out short-circuit protection by means of the fuse. The high-voltage fuse is a power system protection device with simple structure, and is often used in a power distribution network, if an overload or short-circuit fault occurs in a circuit, when the fault current exceeds the rated current of the melt, the melt can be rapidly heated and fused, so that the current is cut off to prevent the fault from expanding.

The high-voltage fuse can be divided into an indoor type and an outdoor type according to the using place. The indoor type is generally made into a regular type, while the outdoor type is made into a drop-out type (the fuse tube is automatically disconnected after the fuse is fused). The high-voltage switch cabinet is important equipment in a power transmission and distribution system, along with the aggravation of loads and the miniaturization of a box body, the high-voltage switch cabinet is easy to generate heat, and short-circuit accidents can be caused by long-term heating. As known from experience accumulated by engineers and experimental work performed by researchers, if a thermal anomaly occurs in an electric high-voltage device, each type of electric high-voltage device may display different temperature information in a thermal imaging image according to different structures of the electric high-voltage device. When the thermal imaging image of the high-voltage power equipment acquired by the thermal imaging instrument is analyzed by the operation and maintenance department of the power equipment, an experienced engineer is required to check and judge the thermal imaging image of the target equipment one by one, and although the method can diagnose the thermal abnormal condition of the high-voltage power equipment, the unnecessary burden and workload of workers are increased, and on the other hand, the fault information in the thermal imaging image cannot be checked quickly and in batches.

The existing fault detection method for the electric power high-voltage equipment comprises traditional contact measurement and thermal imaging non-contact measurement, and the comparison of the two methods is shown in the following table 1:

TABLE 1

The application of thermal imaging in the power industry is more and more extensive, and the combination of the acquired thermal imaging image and the existing algorithm is used for quickly and accurately diagnosing the fault state of the target equipment, which is a research hotspot in the fault diagnosis direction of the high-voltage switch equipment at present, mainly because of the following factors: firstly, the voltage grade and scale of the power grid in China are continuously improved, the types and the number of high-voltage switch equipment are also sharply increased, the number of the collected thermal imaging images is also sharply increased, and the workload of operation and maintenance personnel is huge; secondly, the intelligent diagnosis method can help engineers and operators to maintain the health of the equipment, predict any potential fault, close the equipment before the fault occurs, avoid being damaged by high-voltage switch equipment and protective equipment thereof, reduce capital loss and ensure the safety of the operators and the equipment; finally, the power grid is developed towards a more intelligent direction, and the thermal anomaly fine management of the high-voltage switch equipment is used for demonstrating the thermal imaging image standardization processing of other equipment.

The intelligent degree of the power industry can be improved by accurately analyzing and identifying the acquired thermal imaging image of the target equipment. From the fault diagnosis method for the thermal imaging image of the high-voltage switch equipment at home and abroad, the method mainly comprises the following aspects:

(1) based on the traditional thermal imaging image processing method;

(2) fault diagnosis based on an expert system;

(3) a method based on deep learning.

Although the thermal imaging technology has more advantages, in an actual application scene, different types and degrees of noise often exist in the acquired thermal imaging image, gaussian noise, rayleigh noise, salt and pepper noise and the like are common, and due to the existence of the noise, most of the thermal imaging image has the problems of low contrast, low signal-to-noise ratio, unclear image and the like, so that serious interference is brought to subsequent segmentation identification and fault diagnosis of high-voltage switch equipment.

With the continuous development of thermal imaging image processing technology, researchers related to the world are also exploring faster and better image processing technology to ensure that the obtained thermal imaging image does not interfere with the influence on the actual work requirement as much as possible. The combination of intelligent computer vision technology and target detection technology also brings the processing technology of thermal imaging images and the target equipment identification technology into a more favorable development stage.

Disclosure of Invention

In order to solve the existing problems, ensure the reliability and the high efficiency of the power supply quality and provide a basis for fault detection and prevention of the thermal abnormality of the high-voltage power equipment, the invention provides a fault detection method of the high-voltage power equipment based on image recognition.

In order to achieve the above object, the technical solution of the present invention is implemented as follows:

a method for detecting faults of electric high-voltage equipment based on image recognition is applied to a power distribution network and is characterized in that: including a plurality of self-align unmanned aerial vehicle, central control platform, be provided with each electric power high-voltage apparatus's in the distribution network GPS location data among the unmanned aerial vehicle, two cameras of at least configuration on the unmanned aerial vehicle: high definition camera and thermal imaging camera, the method includes the following steps:

(A) the central control platform issues a shooting instruction to the nearby unmanned aerial vehicle, and the unmanned aerial vehicle goes to the vicinity of the power high-voltage equipment and adopts a high-definition camera to shoot images originally and transmits the images back to the central control platform;

(B) the central control platform carries out image detection on the original shot image, generates a shooting adjustment instruction and transmits the shooting adjustment instruction back to the unmanned aerial vehicle;

the display of the central control platform displays the original shot image, an image recognition algorithm is adopted to recognize whether a target component needing to be detected exists in the original shot image, a target image of the target component is obtained by cutting from the original shot image according to a monitoring display specification and the image recognition algorithm, the image offset between the center of the target image and the center of the original shot image is calculated, the target image and the image offset are used as the shooting adjustment instruction and sent to the unmanned aerial vehicle, and the original shot image is an image shot by taking the high-definition camera as the center;

(C) the unmanned aerial vehicle adjusts the shooting position of the thermal imaging camera according to the shooting adjusting instruction, and shoots a thermal imaging image by the thermal imaging camera and transmits the thermal imaging image back to the central control platform; further, the unmanned aerial vehicle is provided with an image processing module, and the image processing module is used for cutting partial images of corresponding areas in the thermal imaging image according to the areas of the target images in the original shooting image and sending the partial images back to the central control platform;

(D) the central control platform adopts a filtering algorithm to preprocess the thermal imaging image to remove image noise pixel points;

the pretreatment comprises the following specific operation steps:

(d1) selecting K adjacent pixels around the window mxm by taking the current pixel point as the center, and simultaneously finding out the maximum value Zmax and the minimum value Zmin in the window mxm; wherein when m is 3, K is 5; when m is 5, K is 9; when m is 7, K is 25;

(d2) solving the mean value Zmean of K pixels;

(d3)A1＝Zmean－Zmin；A2＝Zmean－Zmax；

(d4) judging whether the mean value is a noise point, namely if A1 is more than 0 and A2 is less than or equal to 0, outputting Zmean as the current pixel point value, otherwise, returning the size of the enhancement window to (d 1);

(E) and identifying the specific temperature of the component in the thermal imaging image, forming an analysis report according to the temperature value, and sending the analysis report to a maintenance department for maintenance.

The image offset amount comprises an offset component in a first direction and an offset component in a second direction, and the first direction and the second direction are perpendicular to each other; by setting the offset components in the two mutually perpendicular directions, the center of the target image can be ensured to be accurately offset, and the offset precision of the target image is improved.

The image shift amount includes a straight-line distance and a shift direction.

The specific steps of cutting out the original shot image according to the monitoring display specification and the image recognition algorithm to obtain the target image are as follows: identifying the position of a target object in a component of an original shot image by adopting a target object image identification algorithm, and identifying the central position of the original image; and the central control platform obtains a target image of a target object cut from the original shot image according to the central position, the position of the component and the monitoring display specification.

Before displaying the target image in the display according to the monitoring display specification, the method further comprises: acquiring an inclination angle of the unmanned aerial vehicle; correcting the image offset according to the inclination angle;

the offset between the center of the target image and the center of the original shot image is the corrected offset; under the condition that there is the slope at unmanned aerial vehicle shooting angle, shoot the detection of the angle of the display in-process of image and obtain inclination through central control platform display in real time, and then revise the offset to the image difference that will shoot the angular deviation and lead to further improves and shoots the accuracy, improves final detection accuracy.

The invention has the beneficial effects that: the method provided by the invention can realize higher efficiency and accuracy of fault detection of the high-voltage power equipment, can truly reflect whether the equipment has thermal abnormity and hidden danger in the running state or not in the image of the high-voltage power equipment through identification of target equipment and detection of a hot spot region, and can reduce the influence of factors such as observation distance, visual angle and atmospheric transmittance compared with the existing thermal imaging detection method. The acquired target equipment image has low spatial resolution, low contrast, fuzzy edge, missing details, easy noise interference and the like, and provides more rapid and accurate information for identifying and diagnosing equipment faults.

Drawings

FIG. 1 is a schematic diagram of a mobile network security guard with feature detection capability of the present invention;

fig. 2 is a flow chart of the filtering preprocessing.

Detailed Description

The invention provides an electric power high-voltage equipment fault detection method based on image recognition, which is applied to a whole power distribution network, as shown in figure 1, the power distribution network comprises a plurality of self-positioning unmanned aerial vehicles and a central control platform, the unmanned aerial vehicles are provided with GPS positioning data of each electric power high-voltage equipment in the power distribution network, and at least two cameras are configured on the unmanned aerial vehicles: high definition digtal camera and thermal imaging camera. The method comprises the following steps:

when the device is detected under the operation condition, shielding objects such as power transmission lines, branches and the like are avoided to the greatest extent in the acquisition process; overload and short-circuit protection of the device. High-voltage fuses are the simplest protective electrical appliances, which are used to protect electrical equipment from overload and short-circuit currents; different types of high-voltage fuses such as outdoor drop-out type and indoor type are selected according to installation conditions and purposes, and special series should be selected for the high-voltage fuses of special equipment, so that the fuses are fuses.

The method has many image shooting requirements, and specifically comprises the steps of avoiding other surrounding electric equipment to the maximum extent, searching a position with balanced background radiation as much as possible, ensuring that the working time of target equipment is preferably more than 6 hours, and accurately meeting the requirements on the accuracy of temperature information reflected by an acquired thermal imaging image; and ensuring that the shooting lens is aligned with the target high-voltage switch equipment. If a fault exists, an operation and maintenance engineer is informed to maintain and replace the high-voltage switch equipment in time, and the safety of the power system is ensured. Still other environmental requirements are as follows: the relative humidity at the time of detection should be less than 85%; during detection, the wind speed is less than 5m/s, and the weather conditions of thunder and rain are avoided; the quality of the thermal imaging image acquired at night is the best, and the thermal imaging image acquired in cloudy days and cloudy days can also be acquired; preferably in cloudy days, at night or in sunny days after 2 hours; when the device is used for outdoor measurement, the acquisition end is ensured not to be directly irradiated or reflected by the sun; the weather condition is preferably cloudy or in the nighttime period when the outdoor measurement is carried out; besides meeting general environmental requirements, the following requirements are also met: the wind speed is generally not more than 0.5 m/s;

the target object of interest in the infrared pattern of the high-voltage switchgear refers to a specific component of the high-voltage switchgear, which is most concerned by a detector in a thermal imaging image, and generally, the data volume of the thermal imaging image is large, but data required by the detector may be only a part of the thermal imaging image, namely the specific component with a fault, so that other unnecessary data are removed in the thermal imaging process, and therefore shooting adjustment is needed, and if the data are not removed, the data volume is large, and subsequent processing and analysis workload is affected.

The display of the central control platform displays the original shot image, an image recognition algorithm is adopted to recognize whether a target component needing to be detected exists in the original shot image, a target image of the target component is obtained by cutting the original shot image according to a monitoring display specification and the image recognition algorithm, the image offset between the center of the target image and the center of the original shot image is calculated, the target image and the image offset are used as the shooting adjustment instruction and sent to the unmanned aerial vehicle, and the original shot image is an image shot by taking the high-definition camera as the center;

the image offset amount comprises an offset component in a first direction and an offset component in a second direction, and the first direction and the second direction are perpendicular to each other; by setting the offset components in two mutually perpendicular directions, the center of the target image can be ensured to be accurately offset, and the offset precision of the target image is improved;

the image offset comprises a linear distance and an offset direction;

the specific steps of cutting out the target image from the original shot image according to the monitoring display specification and the image recognition algorithm are as follows: identifying the position of a target object in a component of an original shot image by adopting a target object image identification algorithm, and identifying the central position of the original image; and the central control platform obtains a target image of a target object cut from the original shot image according to the central position, the position of the component and the monitoring display specification.

The monitoring display specification is the minimum image display specification required by the thermal imaging image fault detection algorithm;

by setting the linear distance and the offset direction, how to accurately adjust the shooting angle can be ensured, and the offset precision of the center of the target image is improved.

the pretreatment comprises the following specific operation steps:

(d2) solving the mean value Zmean of K pixels;

(d3)A1＝Zmean－Zmin；A2＝Zmean－Zmax；

The method can remove redundant data, improve the speed of image processing and analysis, eliminate the interference of other data and obtain the fault object in the thermal imaging image.

The method can further segment the target object in the thermal imaging image by adopting a segmentation algorithm, and can also keep other color information in the original thermal imaging image after the target object is segmented, so that the detection speed is higher. The method is used for directly reading the color information of the mask part in the divided image in the infrared diagnosis work so as to detect the abnormal heating part, and further carrying out the diagnosis work according to the temperature information.

The method of the invention can be applied to fault diagnosis of various electric high-voltage equipment, such as detection of high-voltage isolating switch components, if the temperature reaches 80 ℃, buses and leads in a power grid connected with the high-voltage isolating switch components can be burnt out, and the porcelain piece of the isolating switch can be burst due to the overheat temperature. If the temperature of the high-voltage isolating switch is detected to be abnormal, the high-voltage isolating switch can timely pass through related workers, so that the load borne by the high-voltage isolating switch can be reduced, and the heat emitted by the high-voltage isolating switch can be reduced. If the temperature is still high, the repair should be taken as soon as possible by a power outage.

In normal operation, the equipment operation and maintenance personnel need to carry out standardized inspection on the high-voltage load switch according to a specified time period. When the temperature of the diversion part of the knife switch is checked to be higher than a specified value, abnormal heating is found, and reduction or transfer of the load is carried out immediately. If the disconnecting link close to the side of the line is in an abnormal high-temperature state, immediate power failure maintenance is required, and monitoring of the part is strengthened for a period of time after maintenance is completed. It should be noted that the knife switch generates heat in the closed high-pressure chamber, and the observation period after the overhaul is finished not only needs to reduce the load of the equipment, but also can ensure that the surrounding environment is in a ventilation state.

When the load switch is overhauled, attention is paid to checking external conditions, such as: the actuator is not sensitive; the burning trace of the contact and the stain are cleaned, and the simultaneous contact of the three phases is also noticed; if the arc extinguishing device is not replaced in time, whether a gap exists between the insulation of the arc extinguishing cover or not is judged; after comprehensive maintenance, the load switch can be put into use after a series of professional detections.

In daily maintenance, the overhaul and maintenance of the high-voltage fuse are enhanced, the high-voltage fuse is inspected regularly, the defective high-voltage fuse is replaced as early as possible, and meanwhile, the replaced fuse body is reasonable. The installation and the disassembly are completed according to the rules and procedures strictly, so that the problem of damage caused by construction is avoided. When installed in a place where the use environment is not ideal, the replacement should be performed as soon as possible according to the actual situation.

The technical solutions described above only represent the preferred technical solutions of the present invention, and some possible modifications to some parts of the technical solutions by those skilled in the art all represent the principles of the present invention, and fall within the protection scope of the present invention.

Claims

1. A fault detection method of electric power high-voltage equipment based on image recognition is applied to a power distribution network and is characterized in that: including a plurality of self-align unmanned aerial vehicle, central control platform, be provided with each electric power high-voltage apparatus's in the distribution network GPS location data among the unmanned aerial vehicle, two cameras of at least configuration on the unmanned aerial vehicle: high definition camera and thermal imaging camera, the method includes the following step:

(A) the central control platform issues a shooting instruction to the nearby unmanned aerial vehicle, and the unmanned aerial vehicle goes to the vicinity of the electric power high-voltage equipment, adopts a high-definition camera to shoot images originally and transmits the images back to the central control platform;

(C) the unmanned aerial vehicle adjusts the shooting position of the thermal imaging camera according to the shooting adjusting instruction, and shoots a thermal imaging image by the thermal imaging camera and transmits the thermal imaging image back to the central control platform; furthermore, the unmanned aerial vehicle is provided with an image processing module, and the image processing module is used for cutting partial images of corresponding areas in the thermal imaging image according to the areas of the target image in the original shot image and sending the partial images back to the central control platform;

the pretreatment comprises the following specific operation steps:

(d2) solving the mean value Zmean of K pixels;

(d3)A1＝Zmean－Zmin；A2＝Zmean－Zmax；

2. The image recognition-based power high voltage equipment fault detection method according to claim 1, characterized in that: the image offset amount comprises an offset component in a first direction and an offset component in a second direction, and the first direction and the second direction are perpendicular to each other; by setting the offset components in the two mutually perpendicular directions, the center of the target image can be ensured to be accurately offset, and the offset precision of the target image is improved.

3. The method for detecting a fault in an electric power high voltage device according to claim 2, characterized in that: the image shift amount includes a straight-line distance and a shift direction.

4. The image recognition-based power high voltage equipment fault detection method according to claim 3, characterized in that: the specific steps of cutting out the target image from the original shot image according to the monitoring display specification and the image recognition algorithm are as follows: identifying the position of a target object in a component of an original shot image by adopting a target object image identification algorithm, and identifying the central position of the original image; and the central control platform obtains a target image of a target object cut from the original shot image according to the central position, the position of the component and the monitoring display specification.

5. An electric high voltage equipment fault detection method based on image recognition according to any one of claims 1-4, characterized in that: before displaying the target image in accordance with the monitored display specification in the display, the method further comprises: acquiring an inclination angle of the unmanned aerial vehicle; correcting the image offset according to the inclination angle;