A method, device, terminal and storage medium for detecting foreign body intrusion in power transmission area
Through the combination of image acquisition and neural network model, the problem of poor accuracy of transmission line detection in the existing technology is solved, and the rapid and accurate intrusion detection of large mechanical equipment is achieved, and the false alarm rate and communication bandwidth requirements are reduced. It is suitable for remote areas. Real-time risk assessment of the environment.
Patent Information
- Application Number
- CN202210652921.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-10
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2042-06-10
AI Technical Summary
The prior art is difficult to detect transmission lines in a timely and accurate manner when detecting foreign objects such as large machinery approaching, resulting in poor detection accuracy, especially when the machinery moves quickly, it is difficult to effectively identify the risk of intrusion.
Using a combination of image acquisition, sub-graph division, maximum correlation calculation and neural network model, image data is repeatedly collected by adjusting the shooting angle, data sets are constructed and inputted to neural network models to evaluate the risk of intrusion, and rapid detection of foreign objects in the transmission area is achieved. .
It improves the accuracy of intrusion detection of large-scale engineering machinery equipment, reduces the false alarm rate, and reduces the communication bandwidth requirements by not requiring pictures to be transmitted, real-time and accurate intrusion risk judgment in remote environments.
Smart Images

Figure CN115019254B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method, a device, a terminal and a storage medium for detecting foreign body intrusion in a power transmission area, and belongs to the technical field of power transmission network intrusion detection. Background Art
[0002] With the acceleration of my country's urbanization process, large-scale infrastructure construction by local governments has led to the compression of transmission line channels. Illegal construction, sudden construction work and seasonal intrusion in the transmission protection zone pose a great threat to the safe and stable operation of the line. Equipment failure caused by intrusion has become one of the main reasons for the line. Since the intrusion of transmission lines is very random, it is difficult for operating units to prevent it, so it is necessary to study the intelligent identification technology of intrusion. Large process machinery and equipment such as cranes in construction operations are the main cause of threats to transmission lines. It is necessary to design a method to detect the threat of these equipment to transmission lines, so as to prevent and warn risks.
[0003] The detection method of the prior art usually performs photo detection only when a foreign object such as a large machinery approaches. When the foreign object such as a large machinery moves at a fast speed, it is difficult to detect it in time, which easily leads to the problem of poor detection accuracy. Summary of the invention
[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and to provide a method, device, terminal and storage medium for detecting foreign object intrusion in a power transmission area, so as to solve the problem of poor accuracy in detecting foreign object intrusion in the surrounding area of the power transmission area.
[0005] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions:
[0006] In a first aspect, the present invention provides a method for detecting foreign body intrusion in a power transmission area, comprising:
[0007] Step A: Collecting images of the power transmission area and corresponding shooting angles;
[0008] Step B: Divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result;
[0009] Step C: calculating the offset of the sub-image with foreign matter in the power transmission area image, adjusting the shooting angle of the power transmission area image according to the offset, and reshooting the power transmission area image;
[0010] Step D: Repeat steps B and C until n sets of offsets and corresponding shooting angles of the transmission area images are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30;
[0011] Step E: inputting n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign objects intruding into the power transmission area;
[0012] Step F: Perform foreign body intrusion detection in the power transmission area according to the risk assessment value.
[0013] The shooting angles include: the yaw angle and the pitch angle of the camera that shoots the image of the power transmission area;
[0014] The yaw angle refers to the angle between the optical axis of the camera and the zero axis within the preset horizontal rotation range within the horizontal rotation range;
[0015] The pitch angle refers to the angle between the optical axis of the camera and the zero axis within the preset vertical rotation range within the vertical rotation range.
[0016] The horizontal rotation range is 180°, and the vertical rotation range is 90°.
[0017] The method for calculating the maximum correlation between each sub-image and all pre-stored foreign body image templates includes:
[0018] The subgraph P is calculated using the following formula: s With different image templates T i Relevance:
[0019]
[0020] Where: represents the correlation between two images, T i (a,b) is the image template T i The pixel with coordinates (a, b) in the middle, P s (a, b) represents the sub-image P of the target image s The pixel with coordinates (a, b) in E(T i ) indicates T i The mathematical expectation, E(P s ) indicates P s The mathematical expectation of
[0021] Wherein, the size of the sub-image is the same as the size of the foreign body image template;
[0022] Compare the correlation calculation results to obtain the maximum correlation between each sub-image and all pre-stored foreign body image templates
[0023] Methods for determining whether there is a foreign object in a sub-graph based on the maximum correlation calculation result include:
[0024] like It is considered that the subgraph contains foreign matter;
[0025] like It is considered that the subgraph does not contain foreign matter;
[0026] in, is the empirical threshold, and its value range is (0, 1).
[0027] The neural network model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer.
[0028] The first hidden layer and the input layer are defined as follows:
[0029]
[0030]
[0031]
[0032]
[0033] In the formula, Δx u+j Indicates the X-direction offset Δx, Δy in the u+jth group of data in the time series u+j Indicates the Y-direction offset Δy in the u+jth group of data in the time series, α u+j Indicates the yaw angles α, β in the u+jth group of data in the time series u+j represents the pitch angle β in the u+jth group of data in time series; represents the weight, b0 is the linear bias parameter; j represents the range of the pooling operation, and the value range of j is 0, 1, 2, 3, 4; u represents the coordinate in the first hidden layer, Respectively represent the Δx of the input layer u+j , Δy u+j , α u+j , β u+j The corresponding nodes in the first hidden layer;
[0034] σ(x) is a nonlinear function:
[0035]
[0036] Where: e x represents the natural exponential function; the parameter μ is the control parameter, which is used to control the speed at which the output value changes with the input value; the function σ(x) is used to enable the neural network to simulate nonlinear mapping relationships;
[0037] The second hidden layer is defined by the following formula:
[0038]
[0039] In the formula, is the node with coordinates (2u+j,v) in the first hidden layer, j ranges from 0 to 1, max represents the maximum value of the two values corresponding to j=0 and j=1, and b1 is the linear bias parameter;
[0040] The third hidden layer and the second hidden layer are defined as follows:
[0041]
[0042] In the formula, represents the node with coordinates (u+j,v) in the second hidden layer, For the node The corresponding weight, j ranges from 0, 1, 2, 3, 4, corresponding to Five nodes, with corresponding weights When v is the same, the five nodes in each group share the same five weights; b2 is the linear bias parameter;
[0043] The relationship between the output layer and the third hidden layer is defined as follows:
[0044]
[0045] In the formula, represents the node with coordinates (l,k) in the third hidden layer, w lk is the corresponding weight, b3 is the linear bias parameter; the output z represents the risk assessment value of foreign matter intrusion into the transmission area.
[0046] Methods for training neural network models include:
[0047] Collecting multiple sets of training data sets and inputting them into the neural network model;
[0048] The risk assessment value output by the neural network model and the manually annotated risk assessment value in the training data set are input into formula (10):
[0049]
[0050] In the formula, represents the manually annotated risk assessment value, z represents the risk assessment value output by the neural network model; λ is a control parameter, and its value range is (0.01, 0.1);
[0051] Solve the function according to the back propagation BP algorithm , and obtain all the parameters of the neural network model.
[0052] In a second aspect, the present invention provides a device for detecting foreign body intrusion in a power transmission area, the device comprising:
[0053] Acquisition module: used to collect images of the power transmission area and the corresponding shooting angles;
[0054] The first calculation module is used to divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result;
[0055] A second calculation module is used to calculate the offset of the sub-image with foreign objects in the power transmission area image, adjust the shooting angle of the power transmission area image according to the offset, and then re-shoot the power transmission area image;
[0056] The third calculation module is used to repeat the first calculation module and the second calculation module until n sets of offsets and corresponding shooting angles of the transmission area image are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30;
[0057] The fourth calculation module is used to input n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign matter intrusion into the power transmission area;
[0058] Detection module: used to detect foreign body intrusion in the power transmission area according to the risk assessment value.
[0059] In a third aspect, the present invention provides a terminal, including a processor and a storage medium;
[0060] The storage medium is used to store instructions;
[0061] The processor is used to operate according to the instructions to execute the steps of the method according to any one of the first aspects.
[0062] In a fourth aspect, a computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.
[0063] Compared with the prior art, the present invention has the following beneficial effects:
[0064] 1. The method for detecting foreign body intrusion in the power transmission area calculates the maximum correlation between the power transmission area image sub-image and all pre-stored foreign body image templates to quickly detect whether the target image contains foreign bodies. It also cooperates with a dedicated neural network model to improve the accuracy of intrusion detection of large-scale engineering machinery and equipment and reduce the false alarm rate.
[0065] 2. The present invention captures multiple groups of transmission area images, collects multiple groups of training data sets and inputs them into the neural network model, and performs intrusion detection based on the acquired risk assessment values. This process does not require the transmission of images or the uploading of a large number of images of intrusion devices, thereby reducing the requirements for communication bandwidth and improving detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 The present invention is a schematic diagram of camera deployment for a method for detecting foreign body intrusion in a power transmission area provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION
[0067] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.
[0068] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.
[0069] Embodiment 1:
[0070] A method for detecting foreign object intrusion in a power transmission area provided in this embodiment can be applied to a terminal and can be executed by a foreign object intrusion detection device in a power transmission area. The device can be implemented by software and / or hardware, and the device can be integrated in a terminal, for example: any smart phone, tablet computer or computer device with communication function.
[0071] An embodiment of the present invention provides a method for detecting foreign body intrusion in a power transmission area, which can be executed by an entity end. The method includes:
[0072] Step A: Collecting images of the power transmission area and corresponding shooting angles;
[0073] Step B: Divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result;
[0074] Step C: calculating the offset of the sub-image with foreign matter in the power transmission area image, adjusting the shooting angle of the power transmission area image according to the offset, and reshooting the power transmission area image;
[0075] Step D: Repeat steps B and C until n sets of offsets and corresponding shooting angles of the transmission area images are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30;
[0076] Step E: inputting n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign objects intruding into the power transmission area;
[0077] Step F: Perform foreign body intrusion detection in the power transmission area according to the risk assessment value.
[0078] It should be noted that in the embodiment of the present invention, steps A, B, C, and D can be executed by the entity end, and steps E and F can be executed by the service end. When foreign object intrusion is found inside or around the power transmission area, detection and judgment are achieved by sending the collected data set at the entity end to the service end. This process does not require the transmission of images, thereby reducing the requirements for communication bandwidth. Even in remote field environments, the intrusion risk can be judged in real time and accurately.
[0079] If no sub-image containing foreign matter is detected according to step B, the image acquisition continues until a sub-image containing foreign matter is detected.
[0080] In this embodiment, when collecting images of the transmission area, the camera can be installed on a rotatable pan-tilt head, and the shooting range of the camera can be controlled by controlling the rotation of the pan-tilt head. The image can be captured at a specified time by controlling the camera shutter at a specified frequency. While the camera is shooting, the corresponding shooting angle is recorded. It should be noted that the pan-tilt head can be installed on a transmission line tower, or a pole can be set in a high-risk area around the transmission line to support the pan-tilt head. By setting up the pan-tilt head and allowing the camera to rotate with the pan-tilt head, the shooting range of the camera can be expanded, thereby completely covering the inside and surroundings of the transmission area, achieving an expansion of the detection field of view, and cooperating with the neural network model to complete intrusion detection, thereby improving the accuracy of intrusion detection of large-scale engineering machinery and equipment and reducing the false alarm rate.
[0081] In an embodiment of the present invention, the shooting angle includes: the yaw angle and the pitch angle of the camera for shooting the image of the power transmission area; the yaw angle refers to the angle between the optical axis of the camera and the zero axis within the preset horizontal rotation range within the horizontal rotation range; the pitch angle refers to the angle between the optical axis of the camera and the zero axis within the preset vertical rotation range within the vertical rotation range; the horizontal rotation range is 180°, and the vertical rotation range is 90°.
[0082] Among them, optionally, the zero axis of the camera's horizontal range is selected in the direction where the camera completely captures the transmission area and the transmission area is located at the center of the shooting; the camera is installed at a height of about 3 meters from the ground, and the optical axis on the horizontal plane is selected as the zero axis of the vertical rotation range, and the corresponding vertical rotation range is -45° to 45°.
[0083] The method for calculating the maximum correlation between each sub-image and all pre-stored foreign body image templates includes:
[0084] The subgraph P is calculated using the following formula: s With different image templates T i Relevance:
[0085]
[0086] Where: represents the correlation between two images, T i (a,b) is the image template T i The pixel with coordinates (a, b) in the middle, P s (a, b) represents the sub-image P of the target image s The pixel with coordinates (a, b) in E(T i ) indicates T i The mathematical expectation, E(P s ) indicates P s The mathematical expectation of
[0087] Wherein, the size of the sub-image is the same as the size of the foreign body image template;
[0088] Compare the correlation calculation results to obtain the maximum correlation between each sub-image and all pre-stored foreign body image templates
[0089] The foreign body image template refers to an image containing a certain type of foreign body, which may include a complete foreign body or a characteristic part of the foreign body.
[0090] Methods for determining whether there is a foreign object in a sub-graph based on the maximum correlation calculation result include:
[0091] like It is considered that the subgraph contains foreign matter;
[0092] like It is considered that the subgraph does not contain foreign matter;
[0093] in, is the empirical threshold, whose value range is (0, 1); optional, When it is 0.73, the accuracy is high and the false alarm rate is low; this method is used to quickly detect whether the target image contains foreign objects, and the detection speed is fast.
[0094] The neural network model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer.
[0095] The first hidden layer and the input layer are defined as follows:
[0096]
[0097]
[0098]
[0099]
[0100] In the formula, Δx u+j Indicates the X-direction offset Δx, Δy in the u+jth group of data in the time series u+j Indicates the Y-direction offset Δy in the u+jth group of data in the time series, α u+j Indicates the yaw angles α, β in the u+jth group of data in the time series u+j represents the pitch angle β in the u+jth group of data in time series; represents the weight, b0 is the linear bias parameter; j represents the range of the pooling operation, and the value range of j is 0, 1, 2, 3, 4; u represents the coordinate in the first hidden layer, Respectively represent the Δx of the input layer u+j , Δy u+j , α u+j , β u+j The corresponding nodes in the first hidden layer;
[0101] Each node in the first hidden layer is connected to the five nodes in the input layer. The pooling operation is to map the nodes within a certain range of the previous layer of the neural network to a certain node in the next layer.
[0102] σ(x) is a nonlinear function:
[0103]
[0104] Where: e x represents the natural exponential function; the parameter μ is the control parameter, which is used to control the speed at which the output value changes with the input value; the function σ(x) is used to enable the neural network to simulate nonlinear mapping relationships;
[0105] Among them, compared with the classical function, the nonlinear function of the present invention can better resist data noise interference and help improve the risk identification effect.
[0106] The second hidden layer is defined by the following formula:
[0107]
[0108] In the formula, is the node with coordinates (2u+j,v) in the first hidden layer, j ranges from 0 to 1, max represents the maximum value of the two values corresponding to j=0 and j=1, and b1 is the linear bias parameter;
[0109] The third hidden layer and the second hidden layer are defined as follows:
[0110]
[0111] In the formula, represents the node with coordinates (u+j,v) in the second hidden layer, For the node The corresponding weight, j ranges from 0, 1, 2, 3, 4, corresponding to Five nodes, with corresponding weights When v is the same, the five nodes in each group share the same five weights; b2 is the linear bias parameter;
[0112] The relationship between the output layer and the third hidden layer is defined as follows:
[0113]
[0114] In the formula, represents the node with coordinates (l,k) in the third hidden layer, w lk is the corresponding weight, b3 is the linear bias parameter; the output z represents the risk assessment value of foreign matter intrusion into the transmission area.
[0115] Among them, by calculating the maximum correlation between the transmission area image sub-image and all pre-stored foreign object image templates, it is possible to quickly detect whether the target image contains foreign objects, and in conjunction with the neural network model, it improves the accuracy of intrusion detection of large-scale engineering machinery and equipment and reduces the false alarm rate.
[0116] Methods for training neural network models include:
[0117] Collecting multiple sets of training data sets and inputting them into the neural network model;
[0118] The risk assessment value output by the neural network model and the manually annotated risk assessment value in the training data set are input into formula (10):
[0119]
[0120] In the formula, represents the manually labeled risk assessment value, z represents the risk assessment value output by the neural network model; λ is a control parameter with a value range of (0.01, 0.1), which helps to reduce the impact of noise in the learning process; optionally, λ is 0.035.
[0121] Solve the function according to the back propagation BP algorithm , and obtain all the parameters of the neural network model.
[0122] Among them, the present invention captures multiple groups of images of the power transmission area, collects multiple groups of training data sets and inputs them into the neural network model, and performs intrusion detection through the obtained risk assessment values, without the need to upload a large number of images of intrusion devices, thereby reducing the requirements for communication bandwidth and improving detection accuracy. It can not only realize foreign body intrusion monitoring during power transmission, but also detect the risk of foreign body intrusion in advance at a long distance from the power transmission area, and can achieve higher detection accuracy, thereby improving the ability to prevent the risk of intrusion of large-scale engineering machinery and equipment in the power transmission area.
[0123] Embodiment 2:
[0124] The embodiment of the present invention further provides a device for detecting foreign body intrusion in a power transmission area, the device comprising:
[0125] Acquisition module: used to collect images of the power transmission area and the corresponding shooting angles;
[0126] The first calculation module is used to divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result;
[0127] A second calculation module is used to calculate the offset of the sub-image with foreign objects in the power transmission area image, adjust the shooting angle of the power transmission area image according to the offset, and then re-shoot the power transmission area image;
[0128] The third calculation module is used to repeat the first calculation module and the second calculation module until n sets of offsets and corresponding shooting angles of the transmission area image are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30;
[0129] The fourth calculation module is used to input n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign matter intrusion into the power transmission area;
[0130] Detection module: used to detect foreign body intrusion in the power transmission area according to the risk assessment value.
[0131] Embodiment three:
[0132] An embodiment of the present invention further provides a terminal, including a processor and a storage medium;
[0133] The storage medium is used to store instructions;
[0134] The processor is used to operate according to the instructions to execute the steps of the method described in embodiment 1.
[0135] The processor may include an entity end and / or a service end.
[0136] When executing the steps of the method described in Example 1, by setting a camera at a designated location in the power transmission area, the work of collecting the image of the power transmission area can be performed by the entity end, so as to detect and monitor the content of the image of the power transmission area; when foreign objects invade the inside or around the power transmission area, the entity end generates an alarm and transmits it to the service end for the duty personnel to view and take corresponding measures. The intrusion detection is realized by using a visual method, and the detection task can be completed in a non-contact situation, which is efficient, low-cost and low in potential risk, and therefore has broad application prospects.
[0137] Embodiment 4:
[0138] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the method described in the first embodiment are implemented.
[0139] Since the storage medium provided in the embodiment of the present invention can execute the method provided in the first embodiment of the present invention, it has the corresponding functional modules and beneficial effects of the execution method.
[0140] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0141] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0142] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0144] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for detecting foreign body intrusion in a power transmission area. It is characterized in that include, Step A: Collecting images of the power transmission area and corresponding shooting angles; Step B: Divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result; Step C: calculating the offset of the sub-image with foreign matter in the power transmission area image, adjusting the shooting angle of the power transmission area image according to the offset, and reshooting the power transmission area image; Step D: Repeat steps B and C until n sets of offsets and corresponding shooting angles of the transmission area images are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30; Step E: inputting n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign objects intruding into the power transmission area; The neural network model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer. The first hidden layer and the input layer are defined as follows: In the formula, Indicates the X-direction offset of the u+jth group of data in the time series , Indicates the Y-direction offset of the u+jth group of data in the time series , Indicates the yaw angle in the u+jth group of data in the time series , Indicates the pitch angle in the u+jth group of data in the time series ; , , , represents the weight, is the linear bias parameter; j represents the range of the pooling operation, and the value range of j is 0, 1, 2, 3, 4; u represents the coordinate in the first hidden layer, , , , Respectively represent the input layer , , , The corresponding nodes in the first hidden layer; is a nonlinear function: Where: represents the natural exponential function; parameter is a control parameter used to control the speed at which the output value changes with the input value; function The role of is to enable the neural network to simulate nonlinear mapping relationships; The second hidden layer is defined by the following formula: In the formula, is the coordinate in the first hidden layer The node of j has a value range of 0 and 1. max represents the maximum value of the two values corresponding to j=0 and j=1. is the linear bias parameter; The third hidden layer and the second hidden layer are defined as follows: In the formula, Indicates that the coordinates in the second hidden layer are Nodes, For the node The corresponding weight, j ranges from 0, 1, 2, 3, 4, corresponding to , , …, Five nodes, with corresponding weights , , …, , when v is the same, the five nodes in each group share the same five weights; is the linear bias parameter; The relationship between the output layer and the third hidden layer is defined as follows: In the formula, Indicates that the coordinates in the third hidden layer are Nodes, is the corresponding weight, is a linear bias parameter; the output z represents the risk assessment value of foreign matter intrusion into the transmission area; Step F: Perform foreign body intrusion detection in the power transmission area according to the risk assessment value.
2. A method for detecting foreign body intrusion in a power transmission area according to claim 1, It is characterized in that The shooting angles include: the yaw angle and the pitch angle of the camera that shoots the image of the power transmission area; The yaw angle refers to the angle between the optical axis of the camera and the zero axis within the preset horizontal rotation range within the horizontal rotation range; The pitch angle refers to the angle between the optical axis of the camera and the zero axis within the preset vertical rotation range within the vertical rotation range.
3. A method for detecting foreign body intrusion in a power transmission area according to claim 2, It is characterized in that The horizontal rotation range is 180°, and the vertical rotation range is 90°.
4. A method for detecting foreign body intrusion in a power transmission area according to claim 1, It is characterized in that The method for calculating the maximum correlation between each sub-image and all pre-stored foreign body image templates includes: The subgraph is calculated using the following formula Templates with various images Relevance: Where: represents the correlation between two images, Is an image template The median coordinate is Pixels, A sub-image representing the target image The median coordinate is Pixels, express The mathematical expectation of express The mathematical expectation of Wherein, the size of the sub-image is the same as the size of the foreign body image template; Compare the correlation calculation results to obtain the maximum correlation between each sub-image and all pre-stored foreign body image templates .
5. A method for detecting foreign body intrusion in a power transmission area according to claim 4, It is characterized in that Methods for determining whether there is a foreign object in a sub-graph based on the maximum correlation calculation result include: like , then the subgraph is considered to contain foreign matter; like , then it is considered that the subgraph does not contain foreign objects; in, is the empirical threshold, and its value range is (0, 1).
6. A method for detecting foreign body intrusion in a power transmission area according to claim 1, It is characterized in that Methods for training neural network models include: Collecting multiple sets of training data sets and inputting them into the neural network model; The risk assessment value output by the neural network model and the manually annotated risk assessment value in the training data set are input into formula (10): In the formula, represents the manually labeled risk assessment value, Represents the risk assessment value output by the neural network model; , whose value range is (0.01, 0.1); Solve the function according to the back propagation BP algorithm , and obtain all the parameters of the neural network model.
7. A foreign body intrusion detection device for power transmission area, It is characterized in that The device comprises: Acquisition module: used to collect images of the power transmission area and the corresponding shooting angles; The first calculation module is used to divide the transmission area image into several sub-images, and calculate the maximum correlation between each sub-image and all pre-stored foreign body image templates, and determine whether there is a foreign body in the sub-image according to the maximum correlation calculation result; A second calculation module is used to calculate the offset of the sub-image with foreign objects in the power transmission area image, adjust the shooting angle of the power transmission area image according to the offset, and then re-shoot the power transmission area image; The third calculation module is used to repeat the first calculation module and the second calculation module until n sets of offsets and corresponding shooting angles of the transmission area image are obtained to form n sets of data sets; wherein n is a positive integer set to be no less than 30; The fourth calculation module is used to input n sets of data sets into a pre-built and trained neural network model to obtain a risk assessment value of foreign matter intrusion into the power transmission area; The neural network model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer. The first hidden layer and the input layer are defined as follows: In the formula, Indicates the X-direction offset of the u+jth group of data in the time series , Indicates the Y-direction offset of the u+jth group of data in the time series , Indicates the yaw angle in the u+jth group of data in the time series , Indicates the pitch angle in the u+jth group of data in the time series ; , , , represents the weight, is the linear bias parameter; j represents the range of the pooling operation, and the value range of j is 0, 1, 2, 3, 4; u represents the coordinate in the first hidden layer, , , , Respectively represent the input layer , , , The corresponding nodes in the first hidden layer; is a nonlinear function: Where: represents the natural exponential function; parameter is a control parameter used to control the speed at which the output value changes with the input value; function The role of is to enable the neural network to simulate nonlinear mapping relationships; The second hidden layer is defined by the following formula: In the formula, is the coordinate in the first hidden layer The node of j has a value range of 0 and 1. max represents the maximum value of the two values corresponding to j=0 and j=1. is the linear bias parameter; The third hidden layer and the second hidden layer are defined as follows: In the formula, Indicates that the coordinates in the second hidden layer are Nodes, For the node The corresponding weight, j ranges from 0, 1, 2, 3, 4, corresponding to , , …, Five nodes, with corresponding weights , , …, , when v is the same, the five nodes in each group share the same five weights; is the linear bias parameter; The relationship between the output layer and the third hidden layer is defined as follows: In the formula, Indicates that the coordinates in the third hidden layer are Nodes, is the corresponding weight, is a linear bias parameter; the output z represents the risk assessment value of foreign matter intrusion into the transmission area; Detection module: used to detect foreign body intrusion in the power transmission area according to the risk assessment value.
8. A terminal, It is characterized in that including processor and storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are implemented.
