Distributed power grid line disaster damage intelligent statistical system for unmanned aerial vehicle patrol

Abstract

The invention relates to a distributed power grid line disaster damage intelligent statistical system for unmanned aerial vehicle inspection. The system comprises a statistical platform and at least one intelligent terminal. The intelligent terminal communicates with at least one patrol unmanned aerial vehicle, obtains patrol image data uploaded by the patrol unmanned aerial vehicle and generates disaster damage information; a line disaster damage target identification model based on machine learning is embedded in the intelligent terminal, and the disaster damage information comprises an identified disaster damage target and position information of the disaster damage target calculated according to the identified disaster damage target; the statistical platform is used for responding to a network access request of the intelligent terminal, accessing the intelligent terminal, dynamically managing the intelligent terminal and summarizing disaster damage information uploaded by the intelligent terminal; an intelligent terminal information list is configured in the statistical platform, first feature information is stored in the intelligent terminal information list, a network access request initiated by the intelligent terminal comprises second feature information, and the statistical platform performs feature information matching in response to the network access request and accesses the corresponding intelligent terminal after successful matching.

Description

Technical Field

[0001] The invention relates to a distributed power grid line damage intelligent statistical system for unmanned aerial vehicle inspection, belonging to the technical field of power grid damage management. Background Art

[0002] The frequent occurrence of extreme natural disasters has had a serious impact on power grid lines. After a large-scale disaster occurs, the power grid company will send multiple patrol teams to the disaster-stricken area to conduct on-site disaster damage inspections. However, due to the damage to basic conditions such as roads and communications after the disaster, it is difficult to conduct statistics on power grid disaster losses after the disaster. At present, the use of drones for power inspections can collect information remotely, but due to the complex post-disaster on-site environment, the existing deep learning algorithms have low accuracy in identifying power grid equipment disaster damage targets. After completing the on-site inspection, the patrol team still needs to manually review the massive patrol images and count the disaster damage information. At present, deep learning algorithms are used for target recognition, but the post-disaster on-site situation is complex, and the existing intelligent recognition algorithms cannot fully identify the disaster damage targets. The manual statistics method is not only time-consuming but also prone to errors. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a distributed power grid line disaster loss intelligent statistics system for drone inspections. While providing distributed intelligent recognition functions on site, the system can asynchronously transmit information such as patrol original images, disaster loss target identification frames and actual locations, thereby improving the efficiency of power grid disaster loss statistics.

[0004] The technical solution of the present invention is as follows:

[0005] The present invention proposes a distributed power grid line damage intelligent statistical system for drone inspection, comprising: a statistical platform and at least one intelligent terminal;

[0006] The intelligent terminal communicates with at least one patrol drone, obtains patrol image data uploaded by the patrol drone and generates damage information; a line damage target recognition model based on machine learning is embedded in the intelligent terminal, and the damage information includes the damage target identified by the line damage target recognition model and the location information of the damage target calculated according to the identified damage target;

[0007] The statistical platform is used to respond to network access requests from smart terminals, access smart terminals, dynamically manage smart terminals and summarize disaster loss information uploaded by smart terminals;

[0008] Among them, the statistical platform is configured with a smart terminal information list, which stores the first characteristic information of the authorized smart terminal. The network access request initiated by the smart terminal contains the second characteristic information of the local terminal. The statistical platform matches the first characteristic information and the second characteristic information in response to the network access request, and accesses the corresponding smart terminal after the match is successful.

[0009] As a preferred implementation, after the smart terminal is turned on, it self-checks the network operation status of the local machine. When the network operation status is normal, the patrol personnel enter the task characteristic information according to the patrol task, and at the same time, the smart terminal reads the local basic information, combines the local basic information and the task characteristic information into the second characteristic information and initiates a network access request;

[0010] When the intelligent terminal successfully accesses the network, it interacts with the statistical platform to perform network operations. When the intelligent terminal fails to access the network, it performs local stand-alone operations.

[0011] As a preferred implementation, when initiating a network access request, the intelligent terminal sends a status frame by means of a socket, and simultaneously starts a first timer and a counter, wherein the status frame includes the second characteristic information;

[0012] The intelligent terminal sends the status frame at a period set by the first timer, and sends the status frame again after the time set by the first timer has expired. If the number of times the status frame has been continuously sent exceeds the count number set by the counter, the status sending fails, otherwise the status sending succeeds.

[0013] After the status is sent successfully, the intelligent terminal turns on the listening function and starts the second timer to wait for the network access indication fed back by the statistics platform;

[0014] The statistical platform dynamically listens to the status frame sent by the intelligent terminal, parses the received status frame to obtain the second characteristic information, and matches the second characteristic information with the first characteristic information. If the match is successful, the statistical platform feeds back the network access instruction to the intelligent terminal, and does not respond if the match fails.

[0015] If the intelligent terminal receives the network access indication fed back by the statistical platform within the set time of the second timer, the network access is successful, otherwise the network access fails.

[0016] As a preferred implementation, the intelligent terminal reads the attribute information of the patrol image, the attribute information including the shooting height and the image size, and cuts the image according to the shooting height and the image size to obtain a plurality of sub-images;

[0017] Using the line damage target recognition model, damage targets are identified on multiple sub-graphs, and the identified damage targets are output;

[0018] The multiple sub-images are restored and spliced, and the identified damaged targets are spliced ​​at the same time to obtain the coordinate set of the damaged targets corresponding to the patrol image.

[0019] As a preferred implementation, the attribute information further includes longitude and latitude, and the step of calculating the location information of the disaster damage target according to the identified disaster damage target includes:

[0020] Extract the coordinate set of the damaged target corresponding to the patrol image, and extract the longitude, latitude and shooting height accordingly;

[0021] Convert the corresponding patrol drone from the geodetic coordinate system to the world coordinate system, read the heading angle information, and establish the coordinate conversion relationship;

[0022] Taking the coordinates of the corresponding patrol drone as the reference point, the location information of the disaster target is calculated according to the coordinate set, longitude and latitude of the disaster target and the shooting height of the corresponding patrol image.

[0023] As a preferred implementation, after the statistical platform is connected to the intelligent terminal, the intelligent terminal uploads the disaster loss information to the statistical platform. The specific steps are as follows:

[0024] The intelligent terminal reads the patrol image and generates a unique UUID;

[0025] The intelligent terminal encapsulates the UUID, the target identification frame coordinates of all damaged targets in the patrol image, the shooting time, and the location information of the damaged targets into a location data frame and uploads it to the statistical platform using a socket;

[0026] After completing the location data frame transmission, the intelligent terminal encapsulates the corresponding patrol image and UUID together and uploads them to the statistical platform using a socket;

[0027] The intelligent terminal has a batch uploading function;

[0028] When uploading in batches, the intelligent terminal uniformly transmits all the position data frames of the patrol images to be transmitted at one time;

[0029] When uploading in batches, the intelligent terminal uploads the patrol images to be transmitted one by one, and listens to the feedback from the statistical platform at the same time. If the statistical platform feedbacks the UUID information of the currently transmitted patrol image, the transmission of the current patrol image is terminated and the next patrol image is transmitted until all images are transmitted.

[0030] As a preferred implementation, the steps of dynamically managing smart terminals and aggregating disaster information uploaded by smart terminals by the statistical platform include:

[0031] The statistical platform receives the location data frame uploaded by the smart terminal and parses it. If the same UUID information already exists on the statistical platform, the corresponding location data frame will be deleted. If the same UUID information does not exist on the statistical platform, new disaster loss information will be recorded and it will be marked in the Xinjiang disaster loss information whether there is an original patrol image.

[0032] As a preferred implementation, after creating new disaster damage information, the statistical platform extracts the disaster damage location information in the new disaster damage information, and performs distance calculation with the disaster damage location information corresponding to other UUIDs. When the calculated distance is less than the preset distance, the disaster damage information is merged according to the number of disaster damage targets.

[0033] As a preferred implementation, the statistical platform responds to the image upload request of the smart terminal and obtains the UUID of the currently uploaded inspection image. If the disaster damage information corresponding to the UUID in the statistical platform does not have the original image, the existing inspection image is fed back to the smart terminal; if the disaster damage information corresponding to the UUID in the statistical platform already has the original image, the currently uploaded inspection image is received.

[0034] As a preferred implementation, the steps of dynamically managing smart terminals and aggregating disaster information uploaded by smart terminals by the statistical platform include:

[0035] The statistical platform maps the received disaster location information to a specific area and identifies the disaster target in the patrol image according to the target identification frame of the disaster target;

[0036] The statistical platform filters and summarizes the disaster loss information according to time and specific area, and displays the disaster loss distribution points on the map.

[0037] Additional aspects and advantages of the present invention will be set forth in the following description, and some of them will be apparent from the description, or may be understood by practicing the present invention. In addition, the various aspects and advantages of the present invention may be realized and obtained by the method steps and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of a system framework shown in an embodiment of the present invention;

[0039] Figure 2 is a schematic diagram of performing image segmentation in the image cropping step in an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of a model for performing tower target recognition in an embodiment of the present invention;

[0041] Figure 4 Schematic diagram of a model for classifying tower status in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0044] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0045] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0046] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0047] In view of the characteristics of dispersed inspection and centralized statistics of post-disaster teams, this embodiment proposes a distributed power grid line disaster loss intelligent statistics system for drone inspections: on the one hand, the inspection team carries intelligent terminals and directly performs intelligent identification after the on-site drone inspection; on the other hand, the statistical platform collects the identification results of each terminal and conducts centralized summary statistics. The system conforms to the current operation mode of post-disaster inspections of power grids, and saves time-consuming manual image viewing, reporting and communication, statistical calculations, etc. in the process of disaster loss statistics, which can effectively support the rapid statistics of power grid disaster losses.

[0048] See also Figure 1 This embodiment provides a distributed power grid line damage intelligent statistical system for drone inspections, including: a statistical platform and at least one intelligent terminal.

[0049] The intelligent terminal communicates with at least one patrol drone.

[0050] In one embodiment, the smart terminal can be a drone controller that communicates with a patrol drone. The operator can control the corresponding patrol drone through the smart terminal, issue control commands to it and receive data sent back by the patrol drone. This is suitable for patrol missions with short patrol routes, where one drone can complete the patrol mission.

[0051] In one embodiment, the smart terminal can be a host terminal, multiple patrol drones communicate with the host terminal, multiple operators control the corresponding patrol drones through the drone controller, and each patrol drone transmits data back to the drone controller and the smart terminal at the same time. This is suitable for patrol missions with long patrol routes, and scenarios where multiple drones need to coordinate to perform patrol missions.

[0052] The intelligent terminal obtains the patrol image data uploaded by the patrol drone and generates corresponding damage information; in this embodiment, a line damage target recognition model based on machine learning is embedded in the intelligent terminal, and the machine learning method can be a deep learning technology, such as a convolutional neural network (CNN). CNN can extract deep features of the image and improve the accuracy and efficiency of target recognition. It can also be an SVM support vector machine, which achieves target classification by finding the optimal hyperplane in a high-dimensional space, and has good generalization ability and high accuracy. In addition, a random forest algorithm can also be used. The algorithm can effectively process large-scale data sets by constructing multiple decision trees and performing integrated learning, and has strong robustness. A variety of machine learning algorithms can also be combined to further improve the accuracy and efficiency of recognition. Through these machine learning methods, the intelligent terminal can quickly identify the damage of the patrol line offline, providing strong support for the rapid repair and restoration of power supply of the power grid.

[0053] In this embodiment, the disaster damage information includes identifying the disaster damage target in the patrol image data through the line disaster damage target recognition model, and automatically calculating the actual position of the identified disaster damage target to obtain the position information of the disaster damage target.

[0054] The statistical platform is used to respond to the network access request of the intelligent terminal and access the intelligent terminal. When the intelligent terminal is connected to the statistical platform network, the generated disaster loss information is uploaded to the statistical platform.

[0055] The statistical platform dynamically manages the smart terminals and comprehensively summarizes the disaster loss information uploaded by the smart terminals to obtain all the disaster loss information.

[0056] Among them, a smart terminal information list is configured in the statistical platform, and the smart terminal information list stores the first characteristic information of at least one authorized smart terminal, and the first characteristic information is pre-entered into the statistical platform by the staff.

[0057] The network access request initiated by the smart terminal includes the second characteristic information of the local device, which is entered into the smart terminal by the patrol personnel performing the patrol task, as well as the basic information of the local device automatically read by the smart terminal.

[0058] The statistical platform responds to the network access request initiated by the smart terminal, performs feature matching of the first characteristic information and the second characteristic information, and matches the second characteristic information of the current smart terminal with all the first characteristic information in the smart terminal information list; if any of the first characteristic information successfully matches the second characteristic information of the current smart terminal, the smart terminal that currently initiates the network access request is connected.

[0059] In one embodiment, after the smart terminal is turned on, it self-checks the network operation status of the local communication module, such as the 4G / 5G module operation status, the WIFI module operation status, the Bluetooth module operation status, the LORA module operation status, etc. When the network operation status of any communication module is normal, the patrol personnel enter the task characteristic information according to the patrol task. The task characteristic information includes the patrol task area and ID. At the same time, the smart terminal reads the basic information of the local machine, including the local device ID, longitude and latitude, and the execution task Task ID, etc. The basic information of the local machine and the task characteristic information form the second characteristic information. The control interface of the smart terminal has a built-in status send button. After clicking the status send button, the second characteristic information is filled in the status frame and a network access request is initiated. The format of the status frame is as follows:

[0060]

[0061] When the intelligent terminal successfully accesses the network, it interacts with the statistical platform to perform network operations. When performing network operations, the intelligent terminal uploads the disaster loss information stored locally to the statistical platform. When the intelligent terminal fails to access the network, it performs local stand-alone operations. When performing local stand-alone operations, the intelligent terminal generates disaster loss information and stores it locally.

[0062] In one embodiment, when initiating a network access request, the intelligent terminal sends a status frame via a socket, and simultaneously starts a first timer Timer1 and a counter Counter.

[0063] The intelligent terminal sends status frames at a period of T1 set by the first timer Timer1, and sends status frames again after the time of sending status frames exceeds T1. After continuously sending status frames exceeds the count number C1 set by the counter Counter, it prompts that the status sending fails, otherwise it prompts that the status sending is successful.

[0064] After the status is sent successfully, the intelligent terminal turns on the listening function and starts the second timer Timer2 to wait for the network access indication fed back from the statistical platform; after the status is sent unsuccessfully, the intelligent terminal sets the network status to 0 and stops network interaction.

[0065] The statistics platform is deployed on the cloud server, dynamically listens to the status frame sent by the smart terminal, parses the received status frame to obtain the second characteristic information, and matches the second characteristic information with the first characteristic information in the configuration smart terminal information list. If the match is successful, the network access instruction is fed back to the smart terminal, and if the match fails, no response is made. Multiple information in the first characteristic information and the second characteristic information is matched to prevent other devices from mistakenly accessing the statistics platform network.

[0066] If the smart terminal receives the network access indication from the statistical platform within the set time T2 of the second timer Timer2, after successful information matching, it will prompt that the network access is successful and enable the data upload function; if the smart terminal has not received the network access indication from the statistical platform after a time greater than T2, it will prompt that the network access has not been achieved.

[0067] In one embodiment, the process of the statistical platform matching the second characteristic information with the first characteristic information in the configuration intelligent terminal information list is as follows:

[0068] After receiving the status frame fed back by the intelligent terminal, the statistics platform parses the status frame;

[0069] The first characteristic information read includes device ID, longitude, latitude, mission area and ID (respectively denoted as D id ,Log,Lat,T area , T id ), and the first characteristic information corresponding to the intelligent terminal information list in the platform is: {(D′ id,i ,Log′ i ,Lat′ i ,T′ area,i ,T′ id,i )}, i∈[1,K]) for matching.

[0070] Among them, D′ id,i is the ID of the ith device registered on the platform, T′ area,i The administrative region where the post-disaster inspection mission is assigned to the i-th device, T′ id,i is the task ID assigned to the i-th device in the patrol administrative area, Log′ i is the longitude assigned to the ith device on the platform, Lat′ i The latitude assigned to the ith device on the platform.

[0071] This embodiment uses an exemplary procedure to match the first feature information with the second feature information:

[0072]

[0073] Wherein, reply=1 represents a successful match, and reply=0 represents a failed match.

[0074] In case of network access failure and single-machine operation mode, the smart terminal has a battery and can obtain the patrol images by reading the SD card storing the drone patrol images after the power grid disaster through the card reader interface after startup.

[0075] In one embodiment, the intelligent terminal reads attribute information of the patrol image, the attribute information including shooting height and image size, and cuts the image according to the shooting height and image size to obtain M sub-images;

[0076] In one embodiment, the step of performing image cropping includes:

[0077] ① Get the folder path of the drone patrol image;

[0078] ② Traverse the image file names in the folder and extract the name of the first image;

[0079] ③ Read the image attributes and match and extract the longitude flon, latitude flat, shooting height H, image height fh and image width fw;

[0080] ④ The number of divisions in the width direction of the image is Gw=floor(H / 50), and the number of divisions in the height direction of the image is Gh=floor(H / 40), where floor is a rounding function.

[0081] ⑤For details, see Figure 2 , determine the specific range of each segmented image, with the upper left corner of the image as the coordinate origin, the width direction as the x-axis, and the height direction as the y-axis. Starting from the upper left corner of the entire image, the coordinates of the upper left corner and the lower right corner of the first segmented sub-image are: (0,0) and (int(fw / Gw), int(fh / Gh)). Along the width direction, the coordinates of the upper left corner and the lower right corner of the second segmented sub-image are: (int(fw / Gw)-Cw,0) and (int(2*fw / Gw), int(fh / Gh)), where Cw is the sub-image overlap value in the width direction to avoid incomplete cutting of the target object due to image segmentation. Similarly, when segmenting in the height direction, the boundary of the sub-image is determined according to the overlap value Ch in the height direction.

[0082] Through the above segmentation, when images are taken at different heights, the damaged objects captured can occupy a suitable proportion in the cropped image when the size of the objects changes.

[0083] The line damage target recognition model is used to identify damage targets in multiple sub-graphs, and the identified damage targets are output.

[0084] In one embodiment, in view of the diverse forms of collapsed towers and the limited number of samples, the line damage target recognition model adopts a hierarchical approach to identify collapsed tower damage targets: first identify all towers, then identify collapsed towers, and then identify broken towers. The improved Unet algorithm is used to identify all towers. The characteristic of the improved Unet is that in the upsampling decoding recovery process, the feature graphs of the downsampling encoding stage are fused through the attention mechanism SE, such as Figure 3 shown.

[0085] Among them, in the attention mechanism model SE, parameter a 2 , b is consistent with the feature map size in the left downsampling encoding process in Unet, d 2 , c is consistent with the feature map size of the upsampling decoding connection on the right side of Unet, and:

[0086] Where u is the input feature map (d×d×c), u(i,j) is the element value in the feature map, z is the global average pooling result of the feature element in the d×d dimensional feature map, and Fs is the corresponding global average pooling operator;

[0087] s=Fe(W1,W2,z)=σ(W2δ(W1z)), where W1, W2 are channel weights, δ(·) is the Dirac function, σ(·) is the Sigmoid function, s is the result of weighted calculation and activation function operation on Fs, and Fe is the corresponding weighted activation operator;

[0088] q=Fc(s,u)=su; where Fc is the operator for weighted calculation of s and u.

[0089] Through the attention mechanism, the features of power line towers are better extracted and integrated. 2 ×2 feature map Fi calculation (Fi includes pixel position and pixel classification), then the target extraction area set T of all power line towers in the map can be obtained all .

[0090] Based on the tower targets obtained, the random forest algorithm is used to further classify the status of normal towers, tower collapse and fracture, such as Figure 4 shown.

[0091] First, the target area of ​​the i-th tower in Tall is selected, expanded into a 1-dimensional vector, and a random forest (including n tree random trees);

[0092] Then, the 1-dimensional vector is input into the random forest, and each random tree performs classification calculation on the vector and obtains the corresponding output (0-normal tower, 1-fallen tower, 2-broken tower);

[0093] Finally, vote on the output results of each random tree, and take the majority result as the output of the final tower pole state classification, denoted as St i 。

[0094] Meanwhile, in order to reduce the occurrence of multiple scattered regions in target segmentation in case of disasters, the designed loss function is as follows:

[0095]

[0096] Among them, A is the labeled mask region, B is the segmented region, and ||·|| is the number of pixel points in the region. When the overlap degree between the segmented region and the labeled mask region is less than CON, the loss function is 1, thereby reducing the generation of small-scale segmented regions.

[0097] After completing the disaster damage template recognition step, restore and splice the M subgraphs, and at the same time automatically splice the disaster damage targets recognized in each subgraph Mi to obtain the set of coordinates of the disaster damage targets in the corresponding inspection image.

[0098] In one embodiment, the splicing process mainly performs a merging and duplicate removal process on the repeated disaster damage targets in the subgraphs, and the specific rules are as follows:

[0099] 1) Select subgraphs Mi and Mj, and the center distance is less than Dcon (Dcon is a set threshold).

[0100] 2) Set Ti as the set of segmented targets in subgraph Mi, Ti = {Ti,1, Ti,2, Ti,3,... Ti,p},

[0101] Tj as the set of segmented targets in subgraph Mj, Tj = {Tj,1, Tj,2, Tj,3,... Tj,q}, and T*,* is the disaster damage target area recognized in the subgraph.

[0102] Judge whether there is an overlap between the segmented target area Ti,p in Ti and the segmented target area Tj,q in Tj. If the areas overlap, merge the two areas in the spliced graph;

[0103] 3) Set Cti,p as the center position of the area of Ti,p, and Ctj,q as the center position of the area of Tj,q. If Dis(Cti,p, Ctj,q) < Dcc (Dis represents distance calculation, and Dcc is a set threshold), then retain the area with the larger area in Ti,p and Tj,q.

[0104] 4) Repeat steps 1-3 to complete the merging and deduplication processing of the segmentation targets for all sub-graphs. In the final graph, the set of disaster loss targets is: TGdis = {TGn1, TGn2, ...}.

[0105] In one embodiment, the attribute information further includes longitude and latitude, and the step of calculating the location information of the disaster damage target according to the identified disaster damage target includes:

[0106] The coordinate set of the disaster target corresponding to the patrol image is extracted: TGdis = {TGn1, TGn2, ...}, and the position of the disaster target is further calculated: TCn = {(xtcn1, ytcn1), (xtcn2, ytcn2), ...}, and the longitude, latitude and shooting height and other information are extracted at the same time.

[0107] Convert the corresponding patrol drone from the geodetic coordinate system to the world coordinate system, read the heading angle information, and establish the coordinate conversion relationship;

[0108] Taking the coordinates of the corresponding patrol drone as the reference point, the location information of the disaster target is calculated according to the coordinate set, longitude and latitude of the disaster target and the shooting height of the corresponding patrol image.

[0109] Take (xtcn1, ytcn1) as an example to solve the position: from the pixel coordinate system to the plane coordinate system, then to the camera coordinate system, then to the drone coordinate system, and then to the geographic coordinate system. Combining the transformation relationship of each coordinate system, we get the following:

[0110]

[0111] Where f is the focal length, K is a constant, and K = f, c x and c y It is the deviation between the center of the camera's photosensitive chip and the optical axis on the x-axis and y-axis, which can be approximated to 0. (α, β) are the pitch angle and azimuth angle between the camera and the drone, and (ψ, θ, w) are the pitch angle, roll angle, and yaw angle of the drone, respectively. In actual flight, β = 0, ψ = 0, θ = 0 can be set to obtain the following relationship:

[0112]

[0113] Furthermore, the geodetic longitude and latitude can be calculated:

[0114] B = arctan(E / N).

[0115] L = arcsin(D / sqrt(N^2+E^2+D^2))+L' (where L' is the latitude of the drone).

[0116] Thus, the longitude and latitude corresponding to the disaster damage target (xtcn1, ytcn1) are obtained.

[0117] In one embodiment, after the statistical platform is connected to the intelligent terminal, the intelligent terminal uploads the disaster loss information to the statistical platform. The specific steps are:

[0118] The intelligent terminal reads the patrol image and generates a unique UUID (Universally Unique Identifier).

[0119] The intelligent terminal encapsulates the UUID, the target identification frame coordinates of all disaster-damaged targets in the patrol image, the shooting time, and the location information of the disaster-damaged targets into a location data frame and uploads it to the statistical platform using a socket.

[0120] After completing the location data frame transmission, the intelligent terminal will encapsulate the corresponding patrol image and UUID together, and upload them to the statistical platform using socket.

[0121] The format of the location data frame is as follows:

[0122]

[0123] Among them, the location information structure is as follows:

[0124]

[0125]

[0126] In one embodiment, the smart terminal has a batch uploading function;

[0127] When uploading in batches, the intelligent terminal uniformly transmits the position data frames of all patrol images to be transmitted at one time.

[0128] When uploading in batches, the intelligent terminal uploads the patrol images to be transmitted one by one, and listens to the feedback from the statistical platform at the same time. If the statistical platform feedbacks the UUID information of the currently transmitted patrol image, the transmission of the current patrol image is terminated and the next patrol image is transmitted until all images are transmitted.

[0129]

[0130]

[0131] In one embodiment, the steps of dynamically managing smart terminals and aggregating disaster information uploaded by the smart terminals by the statistical platform include:

[0132] The statistical platform receives the location data frame uploaded by the smart terminal and parses it. If the same UUID information already exists on the statistical platform, the corresponding location data frame will be deleted. If the same UUID information does not exist on the statistical platform, new disaster damage information will be created for recording and whether there is an original patrol image in the Xinjiang disaster damage information will be marked. At the same time, the update status frame will be fed back to the smart terminal:

[0133]

[0134] In one embodiment, after creating new disaster damage information, the statistical platform extracts the disaster damage location information in the newly created disaster damage information, and calculates the distance with the disaster damage location information corresponding to other UUIDs. When the calculated distance is less than the preset distance, the disaster damage information is merged according to the number of disaster damage targets. The merge processing method can be to delete other disaster damage information when the number of disaster damage targets in the newly created disaster damage information is greater than that of other disaster damage information. The number of duplicate records of disaster damage information is reduced through the merge processing.

[0135] In one embodiment, the statistical platform responds to the image upload request of the smart terminal, obtains the UUID of the currently uploaded inspection image, and if the disaster damage information corresponding to the UUID in the statistical platform does not have the original image, the existing inspection image is fed back to the smart terminal; if the disaster damage information corresponding to the UUID in the statistical platform already has the original image, the currently uploaded inspection image is received.

[0136] In one embodiment, the steps of dynamically managing smart terminals and aggregating disaster information uploaded by the smart terminals by the statistical platform include:

[0137] The statistical platform maps the received disaster location information to a specific area, and identifies the disaster target in the patrol image according to the target identification frame of the disaster target; thus obtaining a detailed and readable record of the disaster information.

[0138] The statistical platform filters and summarizes the disaster loss information according to time and specific areas, and displays the disaster loss distribution points on the map to provide commanders with intuitive disaster loss information.

[0139] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0140] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0141] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0142] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent statistical system for the damage of distributed power grid lines for drone inspection, characterized in that: include: Statistics platform and at least one intelligent terminal; The intelligent terminal communicates with at least one patrol drone, obtains patrol image data uploaded by the patrol drone and generates disaster damage information; A line damage target recognition model based on machine learning is embedded in the intelligent terminal. The damage information includes the damage target identified by the line damage target recognition model and the location information of the damage target calculated based on the identified damage target. The statistical platform is used to respond to network access requests from smart terminals, access smart terminals, dynamically manage smart terminals and summarize disaster loss information uploaded by smart terminals; Among them, the statistical platform is configured with a smart terminal information list, which stores the first characteristic information of the authorized smart terminal. The network access request initiated by the smart terminal contains the second characteristic information of the local terminal. The statistical platform matches the first characteristic information and the second characteristic information in response to the network access request, and accesses the corresponding smart terminal after the match is successful.

2. According to claim 1, a distributed power grid line damage intelligent statistical system for drone inspection is characterized by: After the intelligent terminal is turned on, it self-checks the network operation status of the local machine. When the network operation status is normal, the patrol personnel enter the task characteristic information according to the patrol task, and the intelligent terminal reads the basic information of the local machine, combines the basic information of the local machine and the task characteristic information into the second characteristic information and initiates a network access request; When the intelligent terminal successfully accesses the network, it interacts with the statistical platform to perform network operations. When the intelligent terminal fails to access the network, it performs local stand-alone operations.

3. According to claim 2, a distributed power grid line damage intelligent statistical system for drone inspection is characterized by: When initiating a network access request, the intelligent terminal sends a status frame by means of a socket and simultaneously starts a first timer and a counter, wherein the status frame includes second characteristic information; The intelligent terminal sends the status frame at a period set by the first timer, and sends the status frame again after the time set by the first timer has expired. If the number of times the status frame has been continuously sent exceeds the count number set by the counter, the status sending fails, otherwise the status sending succeeds. After the status is sent successfully, the intelligent terminal turns on the listening function and starts the second timer to wait for the network access indication fed back by the statistics platform; The statistical platform dynamically listens to the status frame sent by the intelligent terminal, parses the received status frame to obtain the second characteristic information, and matches the second characteristic information with the first characteristic information. If the match is successful, the statistical platform feeds back the network access instruction to the intelligent terminal, and does not respond if the match fails. If the intelligent terminal receives the network access indication fed back by the statistical platform within the set time of the second timer, the network access is successful, otherwise the network access fails.

4. According to the distributed power grid line damage intelligent statistical system for drone inspections according to claim 1, it is characterized in that: The process of the intelligent terminal acquiring the patrol image data uploaded by the patrol drone and generating disaster damage information includes: The intelligent terminal reads the attribute information of the patrol image, including the shooting height and the image size, and cuts the image according to the shooting height and the image size to obtain multiple sub-images; Using the line damage target recognition model, damage targets are recognized on multiple sub-graphs, and the recognized damage targets are output; The multiple sub-images are restored and spliced, and the identified damaged targets are spliced ​​at the same time to obtain the coordinate set of the damaged targets corresponding to the patrol image.

5. According to claim 4, a distributed power grid line damage intelligent statistical system for drone inspection is characterized in that: The attribute information also includes longitude and latitude, and the step of calculating the location information of the disaster damage target according to the identified disaster damage target includes: Extract the coordinate set of the damaged target corresponding to the patrol image, and extract the longitude, latitude and shooting height accordingly; Convert the corresponding patrol drone from the geodetic coordinate system to the world coordinate system, read the heading angle information, and establish the coordinate conversion relationship; Taking the coordinates of the corresponding patrol drone as the reference point, the location information of the disaster target is calculated according to the coordinate set, longitude and latitude of the disaster target and the shooting height of the corresponding patrol image.

6. The intelligent statistical system for distributed power grid line damage for drone inspection according to claim 1 is characterized in that: After the statistical platform is connected to the intelligent terminal, the intelligent terminal uploads the disaster loss information to the statistical platform. The specific steps are as follows: The intelligent terminal reads the patrol image and generates a unique UUID; The intelligent terminal encapsulates the UUID, the target identification frame coordinates of all damaged targets in the patrol image, the shooting time, and the location information of the damaged targets into a location data frame and uploads it to the statistical platform using a socket; After completing the location data frame transmission, the intelligent terminal encapsulates the corresponding patrol image and UUID together and uploads them to the statistical platform using a socket; The intelligent terminal has a batch uploading function; When uploading in batches, the intelligent terminal uniformly transmits all the position data frames of the patrol images to be transmitted at one time; When uploading in batches, the intelligent terminal uploads the patrol images to be transmitted one by one, and listens to the feedback from the statistical platform at the same time. If the statistical platform feedbacks the UUID information of the currently transmitted patrol image, the transmission of the current patrol image is terminated and the next patrol image is transmitted until all images are transmitted.

7. The intelligent statistical system for distributed power grid line damage for drone inspection according to claim 6 is characterized in that: The steps of dynamically managing the intelligent terminals and summarizing the disaster loss information uploaded by the intelligent terminals by the statistical platform include: The statistical platform receives the location data frame uploaded by the smart terminal and parses it. If the same UUID information already exists on the statistical platform, the corresponding location data frame will be deleted. If the same UUID information does not exist on the statistical platform, new disaster loss information will be recorded and it will be marked in the Xinjiang disaster loss information whether there is an original patrol image.

8. The intelligent statistical system for distributed power grid line damage for drone inspection according to claim 7 is characterized by: After creating new disaster damage information, the statistical platform extracts the disaster damage location information in the new disaster damage information and calculates the distance with the disaster damage location information corresponding to other UUIDs. When the calculated distance is less than the preset distance, the disaster damage information is merged according to the number of disaster damage targets.

9. The intelligent statistical system for distributed power grid line damage for drone inspection according to claim 7 is characterized by: The statistical platform responds to the image upload request of the smart terminal and obtains the UUID of the currently uploaded inspection image. If there is no original image for the disaster damage information corresponding to the UUID in the statistical platform, the existing inspection image is fed back to the smart terminal; if there is an original image for the disaster damage information corresponding to the UUID in the statistical platform, the currently uploaded inspection image is received.

10. The intelligent statistical system for distributed power grid line damage for drone inspection according to claim 7 is characterized in that: The steps of dynamically managing the intelligent terminals and summarizing the disaster loss information uploaded by the intelligent terminals by the statistical platform include: The statistical platform maps the received disaster location information to a specific area and identifies the disaster target in the patrol image according to the target identification frame of the disaster target; The statistical platform filters and summarizes the disaster loss information according to time and specific area, and displays the disaster loss distribution points on the map.

Images