Hot-line work monitoring method based on real-time space positioning

Abstract

The invention discloses a live-line work monitoring method based on real-time space positioning. The method comprises the following steps that 1, an unmanned aerial vehicle collects real-time video image information of high-altitude work; 2, feature information in the video image information is extracted, and the actual spatial position of the operator is determined; step 3, selecting video image information parallel to a vertical plane where the electric power defect elimination object is located as a monitoring plane for measuring and calculating a safe distance; 4, calculating a combined gap value according to the real distance between the live-line worker and the grounding body and the real distance between the live-line worker and the electrified body; and 5, comparing the calculated combined gap value with an operation standard value, if the combined gap value is greater than or equal to the operation standard value, judging that the current hot-line work is safe, and if the combined gap value is smaller than the operation standard value, judging that the hot-line work is dangerous. The three-dimensional monitoring of the safe distance and the spatial relative position of the live working personnel in a high-voltage power transmission scene is realized, and the hidden danger of breaking through a standard distance and entering a dangerous area is avoided.

Description

Technical Field

[0001] The invention relates to the technical field of power transmission line operation devices, and in particular to a live operation monitoring method based on real-time spatial positioning. Background Art

[0002] The equipotential working method in live working means that the operator is at the same potential as the charged body and wears shielding equipment to perform live working. However, when the human body enters the equipotential state, a short transient process will occur, and a large transient current will appear in this process. Therefore, when working at the equipotential state, the operator should move quickly to enter and leave the equipotential state, and the human body should always maintain the equipotential state with the charged body during the working process.

[0003] Due to the high risk of live working, multiple people are required to cooperate to eliminate power grid faults during the operation. Generally, there are safety supervisors on the ground and on the tower who are responsible for the safety monitoring of the high-altitude operation process. The monitoring process is mainly based on visual observation. Due to objective factors such as angle, distance, and weather conditions, the line of sight is prone to deviation, blind spots, or obstruction. In recent years, real-time monitoring through surveillance cameras fixed on towers has been widely used. Compared with manual monitoring, camera monitoring is more accurate and can last longer. However, due to the installation conditions and location restrictions of the camera, the problem of screen obstruction still exists, and the quality and delay of the video return often fluctuate due to the influence of the network environment, and the timeliness and effectiveness of live working monitoring cannot be guaranteed. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art that the monitoring of live working cannot guarantee timeliness and effectiveness, resulting in reduced working safety of workers, and to provide a live working monitoring method based on real-time spatial positioning.

[0005] The purpose of the present invention is achieved through the following technical solutions: A live working monitoring method based on real-time spatial positioning comprises the following steps: Step 1: The drone collects real-time video image information of the high-altitude operation; Step 2: extracting feature information from the video image information, the feature information including the location information of the operator and the power equipment, and using the location of the power equipment as a reference to determine the actual spatial location of the operator; Step 3, select the video image information parallel to the vertical plane where the power fault elimination object is located as the monitoring surface for measuring the safety distance; Step 4, according to the ratio between the real size of the power equipment and the pixel size of the power equipment in the monitoring surface, calculate the real distance between the live working personnel and the grounding body and the live body in the monitoring surface, and calculate the combined gap value according to the real distance between the live working personnel and the grounding body and the live body; Step 5, compare the calculated combined gap value with the operation specification value. If the combined gap value is greater than or equal to the operation specification value, the current live operation is judged to be safe. If the combined gap value is less than the operation specification value, the live operation is judged to be dangerous.

[0006] Preferably, in step 2, the feature information in the video image information is extracted by a trained deep learning model to automatically identify and track the operating personnel and power equipment.

[0007] Preferably, the deep learning model is a convolutional neural network image recognition model, and the process of training the convolutional neural network image recognition model is: Step 2a, initialize the convolution kernel, weight matrix and bias vector; Step 2b, selecting a cross entropy loss function or a mean square error loss function and minimizing the loss function by using a stochastic gradient descent method; Step 2c, inputting historical video images into a convolutional neural network image recognition model for training, and using a back propagation algorithm to update the model parameters; In step 2d, the recognition accuracy and generalization ability of the model are evaluated using the validation set.

[0008] Preferably, in step 2c, the historical video image is a preprocessed image, and the preprocessing process is: Obtain the shooting time of historical video images, and classify the historical video images according to the shooting time, where the classification conditions are the shooting date and shooting time; classify similar shooting dates or similar shooting times into one category, and then perform denoising on the historical video images of this category; After the historical video images are denoised, a convolutional neural network image recognition model is trained on this type of historical video images, and the trained model is used to process this type of historical video images.

[0009] Preferably, in step 1, the real-time video image is also compared with the historical video image, and the historical video image that is most similar to the real-time video image is found in the historical video images, and the classification category corresponding to the historical video image is obtained, and the real-time video image is recognized using the convolutional neural network image recognition model corresponding to the classification category.

[0010] Preferably, the combined gap value is the sum of the shortest straight-line distance between a person and a charged object and the shortest straight-line distance between a person and a grounded object.

[0011] Preferably, the step 4 is specifically as follows: Assume that the pixel length of the power equipment is N pixels, and the minimum pixel distance between the live working personnel and the live body is N d , corresponding to the actual distance Ld ; The minimum pixel distance between live working personnel and grounding body is N j , corresponding to the actual distance L j , According to It can be concluded that Therefore, the combined gap value D is: D = L d +L j .

[0012] Preferably, the live working monitoring method based on real-time spatial positioning also collects meteorological condition information of high-altitude working, and the meteorological condition information includes wind speed, temperature and humidity.

[0013] Preferably, when classifying historical video images, they are also classified according to meteorological conditions, and historical video images with similar meteorological conditions and similar shooting dates are grouped together, or historical video images with similar meteorological conditions and similar shooting times are grouped together.

[0014] Preferably, when live working is determined to be dangerous, an alarm is issued to alert the workers, and the alarm signal includes one or more of an audible alarm, a visual alarm or a vibration alarm.

[0015] The beneficial effects of the present invention are as follows: the present invention utilizes drones to surround personnel who are performing live operations at high altitudes, and shoots and pushes out video image information in real time at a specific angle and distance, so as to obtain the position and posture information of the personnel in real time. Based on the existing three-dimensional model, the spatial positioning of the personnel can be realized, and their three-dimensional coordinates can be obtained. According to the corresponding relationship between the actual size of the power equipment and the pixel size in the picture, the distance between the live operator and the grounding body and the live body during the live fault elimination process is calculated. Thereby, three-dimensional monitoring of the safe distance and relative spatial position of the live operator in the high-voltage power transmission scenario is realized, and the hidden danger of breaking through the standard distance and entering the dangerous area is avoided, thereby monitoring the construction safety of the operators in real time and improving the safety of the operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0017] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.

[0018] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0019] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0020] Example: A live working monitoring method based on real-time spatial positioning, such as Figure 1 As shown, the following steps are included: Step 1: The drone collects real-time video image information of the high-altitude operation; Step 2: extracting feature information from the video image information, the feature information including the location information of the operator and the power equipment, and using the location of the power equipment as a reference to determine the actual spatial location of the operator; Step 3, select the video image information parallel to the vertical plane where the power fault elimination object is located as the monitoring surface for measuring the safety distance; Step 4, according to the ratio between the real size of the power equipment and the pixel size of the power equipment in the monitoring surface, calculate the real distance between the live working personnel and the grounding body and the live body in the monitoring surface, and calculate the combined gap value according to the real distance between the live working personnel and the grounding body and the live body; Step 5, compare the calculated combined gap value with the operation specification value. If the combined gap value is greater than or equal to the operation specification value, the current live operation is judged to be safe. If the combined gap value is less than the operation specification value, the live operation is judged to be dangerous.

[0021] In the step 2, the feature information in the video image information is extracted by using a trained deep learning model to automatically identify and track the operating personnel and power equipment.

[0022] The deep learning model is a convolutional neural network image recognition model, and the process of training the convolutional neural network image recognition model is as follows: Step 2a, initialize the convolution kernel, weight matrix and bias vector; Step 2b, selecting a cross entropy loss function or a mean square error loss function and minimizing the loss function by using a stochastic gradient descent method; Step 2c, inputting historical video images into a convolutional neural network image recognition model for training, and using a back propagation algorithm to update the model parameters; In step 2d, the recognition accuracy and generalization ability of the model are evaluated using the validation set.

[0023] In the step 2c, the historical video image is a pre-processed image, and the pre-processing process is: Obtain the shooting time of historical video images, and classify the historical video images according to the shooting time, where the classification conditions are the shooting date and shooting time; classify similar shooting dates or similar shooting times into one category, and then perform denoising on the historical video images of this category; After the historical video images are denoised, a convolutional neural network image recognition model is trained on this type of historical video images, and the trained model is used to process this type of historical video images.

[0024] The design of this solution classifies similar shooting dates or similar shooting times into one category, that is, video images with similar lighting conditions, weather conditions and other environmental factors are classified into one category. In this way, when training the convolutional neural network image recognition model, the model can focus more on learning the features under these specific conditions, thereby improving the recognition accuracy of operators and power equipment. At the same time, classifying video images of different time periods and training the model separately can make the model better adapt to the changes in lighting and weather in different time periods. In this way, when the model processes new real-time video images, no matter when these images were taken, the model can more accurately identify operators and power equipment, enhancing the generalization ability of the model.

[0025] In step 1, the real-time video image is also compared with the historical video image, and the historical video image that is most similar to the real-time video image is found in the historical video images, and the classification category corresponding to the historical video image is obtained, and the convolutional neural network image recognition model corresponding to the classification category is used to identify the real-time video image.

[0026] The combined gap value is the sum of the shortest straight-line distance between a person and a charged object and the shortest straight-line distance between a person and a grounded object.

[0027] The step 4 is specifically as follows: Assume that the pixel length of the power equipment is N pixels, and the minimum pixel distance between the live working personnel and the live body is N d , corresponding to the actual distance L d ; The minimum pixel distance between live working personnel and grounding body is N j , corresponding to the actual distance Lj , According to It can be concluded that Therefore, the combined gap value D is: D = L d +L j .

[0028] The live working monitoring method based on real-time spatial positioning also collects meteorological condition information of high-altitude working, which includes wind speed, temperature and humidity.

[0029] When classifying historical video images, they are also classified according to meteorological conditions, and historical video images with similar meteorological conditions and similar shooting dates are grouped together, or historical video images with similar meteorological conditions and similar shooting times are grouped together.

[0030] When it is determined that live working is dangerous, an alarm is issued to alert the workers. The alarm signal includes one or more of an audible alarm, a visual alarm or a vibration alarm.

[0031] In this embodiment, the meteorological condition information of the working environment is collected, which can not only evaluate the safety of live working and avoid the risk of live working caused by severe meteorological conditions, but also cleverly use meteorological conditions to classify historical video images, further increasing the accuracy and generalization ability of model recognition.

[0032] The present embodiment is further described below by taking a specific application scenario of a certain power company. The certain power company needs to perform live work on a 110kV transmission line, and the work content includes replacing insulators.

[0033] The drone is equipped with a high-definition camera and flies over the work site to collect real-time video image information of the high-altitude work. The drone flies at an altitude of 30 meters and a speed of 5 meters per second, ensuring that the details of the workers and power equipment can be clearly captured.

[0034] The trained convolutional neural network image recognition model is used to extract features from the collected video images. The model can automatically identify and track the location information of operators and power equipment, and use the location of power equipment as a reference to determine the actual spatial location of the operators.

[0035] The video image information parallel to the vertical plane where the power fault elimination object (i.e. the insulator that needs to be replaced) is located is selected as the monitoring surface for calculating the safe distance. According to the ratio between the real size of the power equipment (such as the length and width of the insulator) and the pixel size of the power equipment in the monitoring surface, the real distance between the live working personnel and the grounding body and the live body in the monitoring surface is calculated. The real length of the insulator is 1 meter, the pixel length in the monitoring surface is 100 pixels, and the minimum pixel distance between the live working personnel and the insulator is 250 pixels, then the corresponding actual distance is 2.5 meters.

[0036] The combined clearance value is calculated based on the actual distance between the live working personnel and the grounding body and the live body. The combined clearance value is the sum of the shortest straight-line distance between the person and the live body and the shortest straight-line distance between the person and the grounding body. After calculation, the shortest straight-line distance between the live working personnel and the grounding body is 1.9 meters, and the shortest straight-line distance between the live working personnel and the live body is 3.5 meters, so the combined clearance value is 5.4 meters.

[0037] The calculated combined clearance value is compared with the work specification value (such as 3.9 meters). Since 5.4 meters is greater than 3.9 meters, it is determined that the current live work is safe.

[0038] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.

[0039] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A live working monitoring method based on real-time spatial positioning, characterized in that: The following steps are involved: Step 1: The drone collects real-time video image information of the high-altitude operation; Step 2: extracting feature information from the video image information, the feature information including the location information of the operator and the power equipment, and using the location of the power equipment as a reference to determine the actual spatial location of the operator; Step 3, selecting video image information parallel to the vertical plane where the power failure elimination object is located as the monitoring plane for measuring the safety distance; Step 4: According to the ratio between the real size of the power equipment and the pixel size of the power equipment in the monitoring surface, the real distance between the live working personnel and the grounding body and the live body in the monitoring surface is calculated, and the combined gap value is calculated according to the real distance between the live working personnel and the grounding body and the live body; Step 5, compare the calculated combined gap value with the operation specification value. If the combined gap value is greater than or equal to the operation specification value, the current live operation is judged to be safe. If the combined gap value is less than the operation specification value, the live operation is judged to be dangerous.

2. A method for monitoring live working based on real-time spatial positioning according to claim 1, characterized in that: In the step 2, the feature information in the video image information is extracted by using a trained deep learning model to automatically identify and track the operating personnel and power equipment.

3. A method for monitoring live working based on real-time spatial positioning according to claim 2, characterized in that: The deep learning model is a convolutional neural network image recognition model, and the process of training the convolutional neural network image recognition model is as follows: Step 2a, initialize the convolution kernel, weight matrix and bias vector; Step 2b, select the cross entropy loss function or the mean square error loss function and minimize the loss function by stochastic gradient descent method; Step 2c, inputting historical video images into the convolutional neural network image recognition model for training, and using the back propagation algorithm to update the model parameters; In step 2d, the recognition accuracy and generalization ability of the model are evaluated using the validation set.

4. A method for monitoring live working based on real-time spatial positioning according to claim 3, characterized in that: In the step 2c, the historical video image is a pre-processed image, and the pre-processing process is: Obtain the shooting time of historical video images, and classify the historical video images according to the shooting time, where the classification conditions are the shooting date and shooting time; classify similar shooting dates or similar shooting times into one category, and then perform denoising on the historical video images of this category; After the historical video images are denoised, a convolutional neural network image recognition model is trained on this type of historical video images, and the trained model is used to process this type of historical video images.

5. A method for monitoring live working based on real-time spatial positioning according to claim 4, characterized in that: In step 1, the real-time video image is also compared with the historical video image, and the historical video image that is most similar to the real-time video image is found in the historical video images, and the classification category corresponding to the historical video image is obtained, and the convolutional neural network image recognition model corresponding to the classification category is used to identify the real-time video image.

6. The method for monitoring live working based on real-time spatial positioning according to claim 1 is characterized in that: The combined gap value is the sum of the shortest straight-line distance between a person and a charged object and the shortest straight-line distance between a person and a grounded object.

7. A method for monitoring live working based on real-time spatial positioning according to claim 6, characterized in that: The step 4 is specifically as follows: Assume that the pixel length of the power equipment is N pixels, and the minimum pixel distance between the live working personnel and the live body is N d , the corresponding actual distance is Ld; the minimum pixel distance between the live working personnel and the grounding body is N j , corresponding to the actual distance L j , According to It can be concluded that Therefore, the combined gap value D is: D = L d +L j 8. The method for monitoring live working based on real-time spatial positioning according to claim 4 is characterized in that: It also collects meteorological condition information for high-altitude operations, including wind speed, temperature and humidity.

9. A method for monitoring live working based on real-time spatial positioning according to claim 8, characterized in that: When classifying historical video images, they are also classified according to meteorological conditions, and historical video images with similar meteorological conditions and similar shooting dates are grouped together, or historical video images with similar meteorological conditions and similar shooting times are grouped together.

10. The method for monitoring live working based on real-time spatial positioning according to claim 1 is characterized in that: When it is determined that live working is dangerous, an alarm is issued to alert the workers. The alarm signal includes one or more of an audible alarm, a visual alarm or a vibration alarm.

Images