Power plant safety monitoring method and system based on unmanned aerial vehicle

Abstract

The invention discloses a power plant safety monitoring method and system based on an unmanned aerial vehicle, and the method comprises the steps: determining a to-be-inspected target region in a power plant and an inspection unmanned aerial vehicle corresponding to the target region in response to an inspection instruction; determining a plurality of inspection task points corresponding to the target area according to the inspection instruction, and generating an inspection path of the inspection unmanned aerial vehicle according to the inspection task points; controlling the inspection unmanned aerial vehicle according to the inspection path to obtain inspection image data corresponding to the target area; and performing safety analysis on the power plant according to the inspection image data to obtain the safety state of the power plant, thereby improving the accuracy of safety monitoring.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a power plant safety monitoring method and system based on unmanned aerial vehicle (UAV) technology. Background Art

[0002] Using digital technology and means to strengthen power plant safety management is an important research direction to improve power plant operation safety. With the development of science and technology, more and more new power plants or renovated power plants are gradually using smart sensors, operation and maintenance robots, drones and other smart equipment for smart inspection and management of power plants, which can not only improve the quality and accuracy of power plant safety management, but also reduce the manpower and time costs of power plant safety operation and maintenance.

[0003] In the existing technology, the use of drones for power plant monitoring has become one of the important means of intelligent inspection of power plants due to the flexible and easy-to-operate features of drones. Existing drone monitoring technology is mainly managed and operated by professional drone operators to inspect and monitor designated scenes or blind spots in power plants. However, this operation method requires professionals to operate drones for a long time in actual application, resulting in an increase in human resources, and is limited by manual operation of personnel, making it impossible for drones to achieve precise flight control and inspection control during the inspection process. Summary of the invention

[0004] In order to solve the above technical problems, the present invention discloses a power plant safety monitoring method and system based on drones, which are used to improve the intelligence of power plant safety monitoring.

[0005] In order to achieve the above object, the present invention discloses a power plant safety monitoring method based on a drone, comprising:

[0006] In response to the inspection instruction, determine the target area to be inspected in the power plant and the inspection drone corresponding to the target area;

[0007] Determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path of the inspection drone according to the inspection task points;

[0008] Control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area;

[0009] A safety analysis is performed on the power plant according to the inspection image data to obtain a safety status of the power plant.

[0010] The present invention discloses a power plant safety monitoring method based on drones. First, in response to an input inspection instruction, the target area to be inspected inside the power plant and the inspection drone corresponding to the target area are determined. Thus, through the corresponding relationship between the inspection drone and the target area, the task volume of the drone inspection is reduced while avoiding the waste of drone resources, thereby improving the efficiency of the inspection. Secondly, according to the inspection instruction, several task points to be inspected in the target area are determined, so as to generate an inspection path for the inspection drone according to the task points, so that the inspection drone can inspect all task points according to the inspection path, avoiding inspection omissions caused by manual operation, thereby improving the accuracy of the inspection. Then, when the inspection drone is controlled to execute the inspection instruction according to the inspection path, the corresponding inspection image data is obtained, so as to perform a safety analysis of the power plant according to the inspection image data covering all inspection task points, thereby improving the accuracy of the safety analysis.

[0011] As a preferred example, the determining of the target area to be inspected in the power plant and the inspection drone corresponding to the target area includes:

[0012] Determine the type of inspection task according to the inspection instruction; wherein the type of inspection task includes daily inspection task and emergency inspection task;

[0013] When it is determined that the inspection task is a daily inspection task, the regional division information and drone deployment information of the power plant are obtained, and the inspection drone corresponding to each inspection area in the power plant is determined;

[0014] When it is determined that the inspection task is an emergency inspection task, the location information of the emergency inspection area is obtained to determine the inspection drone corresponding to the emergency inspection area according to the location information and the drone deployment information.

[0015] The present invention divides the inspection tasks into daily inspections and emergency inspections. On the one hand, the daily inspection routes of the inspection drones are controlled to be consistent through the daily inspections, thereby ensuring the consistency of the daily inspection contents and the accuracy of the daily inspections. On the other hand, emergency inspection tasks are set so that when an emergency occurs, the safety situation of the emergency inspection area can be obtained in time, thereby improving the efficiency of the inspections.

[0016] As a preferred example, the generating of the inspection path of the inspection drone according to the inspection task point includes:

[0017] Retrieving a three-dimensional environment model corresponding to the power plant, and obtaining location information of each inspection task point in the three-dimensional environment model;

[0018] According to the location information, an inspection path of the inspection drone in the target area is generated by a preset path planning algorithm.

[0019] The present invention uses a three-dimensional model of a power plant to construct the inspection path, so that the inspection drone can avoid power plant equipment in the power plant when inspecting according to the inspection path, thereby ensuring the safety of the drone flight and the safety of the power plant.

[0020] As a preferred example, the controlling the inspection drone according to the inspection path includes:

[0021] Controlling the inspection drone to take off vertically to a preset height;

[0022] According to the preset height, obtaining the environmental information corresponding to the inspection path in real time;

[0023] When it is determined according to the environmental information that there is an obstacle on the inspection path, the inspection drone is controlled to perform an obstacle avoidance action.

[0024] The present invention obtains in real time the environmental information of the inspection drone when it flies along the inspection path, so as to control the inspection drone to avoid obstacles in time according to the environmental information, thereby reducing the risk during the flight of the drone.

[0025] As a preferred example, the obtaining of inspection image data corresponding to the target area includes:

[0026] When the inspection task is an equipment detection task in an emergency inspection task, the target image data corresponding to each inspection task point is obtained to determine whether there is an abnormal target at the current inspection task point according to the target image data;

[0027] When the abnormal target exists, the inspection drone is controlled to track the abnormal target in real time, and an abnormal target image corresponding to the abnormal target is acquired.

[0028] The present invention classifies the emergency inspection tasks to determine whether to perform safety inspections on personnel or equipment according to the classification, thereby avoiding the drone from executing useless inspection processes and improving the efficiency of inspection.

[0029] As a preferred example, the obtaining of inspection image data corresponding to the target area includes:

[0030] When the inspection task is an equipment detection task in an emergency inspection task, the equipment image data corresponding to each inspection task point is obtained to determine whether an abnormality occurs in the equipment at the current inspection task point according to the equipment image data;

[0031] When it is determined that the device is abnormal, an abnormal device image corresponding to the device is acquired.

[0032] The present invention classifies the emergency inspection tasks to determine whether to perform safety inspections on personnel or equipment according to the classification, thereby avoiding the drone from executing useless inspection processes and improving the efficiency of inspection.

[0033] As a preferred example, before performing safety analysis on the power plant according to the inspection image data, the method includes:

[0034] Obtaining the positioning information of the inspection drone in real time;

[0035] According to the positioning information and the location information of the inspection task point, determining whether the inspection drone has completed the inspection task;

[0036] When it is determined that the inspection drone completes the inspection task, the inspection drone is controlled to return.

[0037] The present invention positions the inspection drone to ensure that the inspection drone performs safety inspection on all inspection task points in the inspection task, thereby improving the accuracy of the inspection.

[0038] On the other hand, the present invention discloses a power plant safety monitoring system based on a drone, including a security control module, a computing and communication module, a safety inspection module and a safety analysis module;

[0039] The placement control module is used to respond to the inspection instruction and determine the target area to be inspected in the power plant and the inspection drone corresponding to the target area;

[0040] The edge computing module is used to determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path of the inspection drone according to the inspection task points;

[0041] The safety inspection module is used to control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area;

[0042] The safety analysis module is used to perform safety analysis on the power plant according to the inspection image data to obtain the safety status of the power plant.

[0043] The present invention discloses a power plant safety monitoring system based on drones. First, in response to an input inspection instruction, the target area to be inspected inside the power plant and the inspection drone corresponding to the target area are determined. Thus, through the corresponding relationship between the inspection drone and the target area, the task volume of the drone inspection is reduced while avoiding the waste of drone resources, thereby improving the efficiency of the inspection. Secondly, according to the inspection instruction, several task points to be inspected in the target area are determined, so as to generate an inspection path for the inspection drone according to the task points, so that the inspection drone can inspect all task points according to the inspection path, avoiding inspection omissions caused by manual operation, thereby improving the accuracy of the inspection. Then, when the inspection drone is controlled to execute the inspection instruction according to the inspection path, the corresponding inspection image data is obtained, so as to perform a safety analysis of the power plant according to the inspection image data covering all inspection task points, thereby improving the accuracy of the safety analysis.

[0044] As a preferred example, the placement control module includes an inspection task unit and a drone scheduling unit;

[0045] The inspection task unit is used to determine the type of inspection task according to the inspection instruction; wherein the type of inspection task includes daily inspection task and emergency inspection task;

[0046] The drone scheduling unit is used to obtain the area division information and drone deployment information of the power plant when it is determined that the inspection task is a daily inspection task, and determine the inspection drone corresponding to each inspection area in the power plant; when it is determined that the inspection task is an emergency inspection task, obtain the location information of the emergency inspection area, and determine the inspection drone corresponding to the emergency inspection area based on the location information and the drone deployment information.

[0047] The present invention divides the inspection tasks into daily inspections and emergency inspections. On the one hand, the daily inspection routes of the inspection drones are controlled to be consistent through the daily inspections, thereby ensuring the consistency of the daily inspection contents and the accuracy of the daily inspections. On the other hand, emergency inspection tasks are set so that when an emergency occurs, the safety situation of the emergency inspection area can be obtained in time, thereby improving the efficiency of the inspections.

[0048] As a preferred example, the edge computing module includes a positioning unit and a path planning unit;

[0049] The positioning unit is used to retrieve the three-dimensional environment model corresponding to the power plant and obtain the position information of each inspection task point in the three-dimensional environment model;

[0050] The path planning unit is used to generate an inspection path for the inspection drone in the target area according to the location information through a preset path planning algorithm.

[0051] The present invention uses a three-dimensional model of a power plant to construct the inspection path, so that the inspection drone can avoid power plant equipment in the power plant when inspecting according to the inspection path, thereby ensuring the safety of the drone flight and the safety of the power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 : A schematic diagram of a flow chart of a power plant safety monitoring method based on a drone disclosed in an embodiment of the present invention;

[0053] Figure 2 : is a structural schematic diagram of a power plant safety monitoring system based on a drone disclosed in an embodiment of the present invention;

[0054] Figure 3 : A schematic structural diagram of a power plant safety monitoring system based on a drone disclosed in another embodiment of the present invention;

[0055] Figure 4 : A schematic diagram of the structure of a computing and communication device disclosed in another embodiment of the present invention;

[0056] Figure 5 : is a structural schematic diagram of a central control device disclosed in another embodiment of the present invention;

[0057] Figure 6 : A schematic diagram of a process of controlling a drone to conduct inspections based on a power plant safety monitoring system disclosed in another embodiment of the present invention. DETAILED DESCRIPTION

[0058] 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.

[0059] Embodiment 1

[0060] This embodiment discloses a power plant safety monitoring method based on drones, which is used to improve the accuracy of power plant safety monitoring. Figure 1 , mainly including steps 101 to 104, wherein the steps are mainly:

[0061] Step 101: In response to an inspection instruction, determine a target area to be inspected in a power plant and an inspection drone corresponding to the target area.

[0062] In this embodiment, the step mainly includes: determining the type of inspection task according to the inspection instruction; wherein the types of inspection tasks include daily inspection tasks and emergency inspection tasks; when the inspection task is determined to be a daily inspection task, obtaining the area division information and drone deployment information of the power plant, and determining the inspection drone corresponding to each inspection area in the power plant; when the inspection task is determined to be an emergency inspection task, obtaining the location information of the emergency inspection area, and determining the inspection drone corresponding to the emergency inspection area according to the location information and the drone deployment information.

[0063] In this embodiment, this step divides the inspection tasks into daily inspections and emergency inspections. On the one hand, the daily inspection route of the inspection drone is controlled to be consistent through the daily inspection, thereby ensuring the consistency of the daily inspection content and the accuracy of the daily inspection. On the other hand, emergency inspection tasks are set so that when an emergency occurs, the safety situation of the emergency inspection area can be obtained in time to improve the efficiency of the inspection.

[0064] Step 102: Determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path for the inspection drone according to the inspection task points.

[0065] In this embodiment, this step mainly includes: calling up the three-dimensional environmental model corresponding to the power plant, and obtaining the position information of each inspection task point in the three-dimensional environmental model; based on the position information, generating the inspection path of the inspection drone in the target area through a preset path planning algorithm.

[0066] In this embodiment, this step uses the three-dimensional model of the power plant to construct the inspection path so that the inspection drone can avoid the power plant equipment in the power plant when inspecting according to the inspection path, thereby ensuring the safety of the drone flight and the safety of the power plant.

[0067] Step 103: Control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area.

[0068] In this embodiment, the steps mainly include: controlling the inspection drone to take off vertically to a preset height; acquiring environmental information corresponding to the inspection path in real time according to the preset height; and when it is determined that there are obstacles on the inspection path according to the environmental information, controlling the inspection drone to perform obstacle avoidance actions.

[0069] Among them, when the inspection task is an equipment detection task in an emergency inspection task, the target image data corresponding to each of the inspection task points is obtained to determine whether there is an abnormal target at the current inspection task point based on the target image data; when the abnormal target exists, the inspection drone is controlled to track the abnormal target in real time, and the abnormal target image corresponding to the abnormal target is obtained.

[0070] When the inspection task is an equipment detection task in an emergency inspection task, the equipment image data corresponding to each inspection task point is obtained to determine whether the equipment at the current inspection task point has an abnormality based on the equipment image data; when it is determined that the equipment has an abnormality, the abnormal equipment image corresponding to the equipment is obtained.

[0071] In this embodiment, this step obtains the environmental information of the inspection drone in real time when it flies along the inspection path, so as to control the inspection drone to avoid obstacles in time according to the environmental information, thereby reducing the risk of the drone during flight. At the same time, the emergency inspection tasks are classified to determine whether to perform safety inspections on personnel or equipment according to the classification, so as to avoid the drone from performing useless inspection processes and improve the efficiency of inspections.

[0072] Step 104: Perform safety analysis on the power plant according to the inspection image data to obtain the safety status of the power plant.

[0073] In this embodiment, this step mainly includes: acquiring the positioning information of the inspection drone in real time; judging whether the inspection drone has completed the inspection task based on the positioning information and the location information of the inspection task point; and when it is determined that the inspection drone has completed the inspection task, controlling the inspection drone to return.

[0074] In this embodiment, this step ensures that the inspection drone performs safety inspections on all inspection task points in the inspection task by positioning the inspection drone, thereby improving the accuracy of the inspection.

[0075] On the other hand, this embodiment also discloses a power plant safety monitoring system based on drones. For the specific structure of the monitoring system, please refer to Figure 2 , including a security control module 201, a computing and communication module 202, a security inspection module 203 and a security analysis module 204.

[0076] The placement control module 201 is used to respond to the inspection instruction and determine the target area to be inspected in the power plant and the inspection drone corresponding to the target area.

[0077] The edge computing module 202 is used to determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path for the inspection drone according to the inspection task points.

[0078] The safety inspection module 203 is used to control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area.

[0079] The safety analysis module 204 is used to perform safety analysis on the power plant according to the inspection image data to obtain the safety status of the power plant.

[0080] In this embodiment, the placement control module 201 includes an inspection task unit and a drone scheduling unit.

[0081] The inspection task unit is used to determine the type of inspection task according to the inspection instruction; wherein the type of inspection task includes daily inspection task and emergency inspection task.

[0082] The drone scheduling unit is used to obtain the area division information and drone deployment information of the power plant when it is determined that the inspection task is a daily inspection task, and determine the inspection drone corresponding to each inspection area in the power plant; when it is determined that the inspection task is an emergency inspection task, obtain the location information of the emergency inspection area, and determine the inspection drone corresponding to the emergency inspection area based on the location information and the drone deployment information.

[0083] In this embodiment, the edge computing module 202 includes a positioning unit and a path planning unit.

[0084] The positioning unit is used to retrieve the three-dimensional environment model corresponding to the power plant and obtain the position information of each inspection task point in the three-dimensional environment model.

[0085] The path planning unit is used to generate an inspection path for the inspection drone in the target area according to the location information through a preset path planning algorithm.

[0086] The present embodiment provides a power plant safety monitoring method and system based on drones. First, in response to the input inspection instruction, the target area to be inspected inside the power plant and the inspection drone corresponding to the target area are determined. Thus, through the correspondence between the inspection drone and the target area, the task volume of the drone inspection is reduced while avoiding the waste of drone resources, thereby improving the inspection efficiency. Secondly, according to the inspection instruction, several task points to be inspected in the target area are determined, so as to generate the inspection path of the inspection drone according to the task points, so that the inspection drone can inspect all task points according to the inspection path, avoiding inspection omissions caused by manual operation, and thus improving the inspection accuracy. Then, when the inspection drone is controlled to execute the inspection instruction according to the inspection path, the corresponding inspection image data is obtained, so as to perform the safety analysis of the power plant according to the inspection image data covering all inspection task points, thereby improving the accuracy of the safety analysis.

[0087] Embodiment 2

[0088] In the prior art, manual control methods are often used to operate drones to inspect the safety of power plants. In the process of using drones for inspections, first of all, professionals who are proficient in using drones are required to operate and manage the recovery of drones throughout the process. That is, a large amount of human resources are required in the prior art to support the operation of drones, resulting in an increase in labor costs, and the process is cumbersome and time-consuming, which does not meet the requirements for efficient operation of power plants. At the same time, in the process of manually operating drones, it is limited by the personnel's own reasons, such as misoperation causing unexpected movements of drones, intentional sabotage by operators, insufficient training or carelessness of operators in power plants, and flying drones into prohibited flying areas or operators' lack of proficiency in using drones may cause the drone to fall or fly into prohibited flying areas, thereby causing collisions or even crashes between drones and power plant equipment, thereby posing a danger to the safety of power plants and drones.

[0089] Furthermore, when manually operating drones to conduct safe inspections of power plants, high-precision flight trajectory control cannot be guaranteed when encountering complex environmental areas. In the regular routine inspections of power plants, manually operating drones will inevitably cause deviations in the flight trajectory of each inspection, and it is impossible to ensure that the route and content of each inspection are consistent, which may cause inconsistencies in the content of routine inspections. At the same time, the manual operation of drones cannot adapt well to all environments. In some harsh environments, power plant staff usually cannot work conveniently. For example, in high-temperature environments, it is necessary to consider the thermal tolerance of staff; in high-voltage environments, it is necessary to avoid injuries to staff due to electric shock or other dangers; in the presence of radiation, it is necessary to avoid the impact of radiation on staff.

[0090] In order to solve the above technical problems, in certain implementations of this embodiment, a power plant safety monitoring system based on drones is provided to realize the automatic operation of drones, thereby improving the automation level and monitoring accuracy of power plant safety monitoring. Figure 3 , mainly including a central control device 301, a data transmission device 302, a computing and communication device 303 and an on-site equipment device 304.

[0091] Reference Figure 3 The central control device 301 is used to receive an inspection instruction sent from the outside, determine the target area to be inspected inside the power plant and the inspection drone to inspect the target area according to the inspection instruction, and then generate an inspection task corresponding to the inspection drone. At the same time, in the process of generating the inspection task, a number of inspection task points corresponding to the target area are obtained, so as to generate an inspection path corresponding to the inspection drone according to the location information of the inspection task points.

[0092] Among them, the data transmission device 302 is signal-connected with the central control device 301, and is used to receive the inspection path and the inspection task corresponding to the inspection drone, and forward them to the edge computing device 303 corresponding to the inspection drone.

[0093] The edge computing device 303 is connected to the on-site equipment device 304 by signal, and is used to send the inspection path and the inspection task to the inspection drone 305 in the on-site equipment device 304.

[0094] The inspection drone 305 performs the inspection task according to the inspection path, obtains the inspection image data corresponding to the target area, and uploads the inspection image data to the central control device 301 through the data transmission device 302 and the computing communication device 303 in sequence, so that the central control device 301 can perform a safety analysis on the power plant according to the inspection image data to obtain the safety status of the power plant.

[0095] Specifically, in order to solve the technical problem of low efficiency of drone inspections in the prior art, in certain implementations of this embodiment, the power plant is divided into areas according to safety requirements to obtain several key inspection areas corresponding to the power plant, such as inspection area A, inspection area B, inspection area C, etc. In a specific application example, the power plant is divided into multiple inspection areas such as cooling tower and chimney areas, warehouse and parking lot areas, main plant corridors and bridge areas, and then multiple drones are deployed in each inspection area according to the division of the areas. The drones only inspect the corresponding inspection areas, and multiple drones located in the same inspection area cooperate with each other to complete the inspection of the corresponding inspection areas. As a preferred example, refer to Figure 3 In the on-site equipment device 304, according to the division of the area, multiple inspection drones 305 are deployed in each inspection area, so as to obtain corresponding inspection image data through the inspection drones 305, and then perform safety analysis on the inspection area.

[0096] Furthermore, in order to ensure that the power plant safety monitoring system obtains accurate monitoring results of the power plant, the inspection drone 305 is configured to have industrial-grade reliability so that the inspection drone 305 can work continuously for a long time in complex and harsh environments. In certain implementations of this embodiment, the inspection drone 305 is configured to include at least the following hardware configurations: (1) High-resolution camera: with 4K or higher video acquisition capabilities, supporting night vision and thermal imaging functions to ensure all-weather monitoring. (2) Powerful communication module: should support 5G, LTE or dedicated wireless frequency bands to ensure low-latency, high-bandwidth communication with communication towers. (3) Long-life battery: The battery life must meet long-term flight requirements and support quick replacement or charging. (4) GPS and inertial navigation system: ensure accurate positioning and path planning, and support automatic cruising and task scheduling.

[0097] Further, in order to make up for the shortcomings of the inspection drone 305 during the inspection process, refer to Figure 3 According to the above regional division of the power plant, security equipment 306 can be set in different areas of the power plant. The security equipment 306 includes equipment such as access control system and video monitoring system, which are used to provide physical security information of different areas of the power plant, such as access control information, monitoring video and other physical data.

[0098] In this embodiment, reference is made to the settings of the inspection drone 305 and the security equipment 306 to obtain safety monitoring data corresponding to the power plant at different angles, such as the inspection image data fed back by the inspection drone 305 and the physical monitoring information fed back by the security equipment 306, so as to improve the effect of safety monitoring of the power plant.

[0099] In certain implementations of this embodiment, reference Figure 3 , the computing communication device 303 obtains the data uploaded by the inspection drone 305 and sends the data to the central control device. In the process of data uploading, since the inspection tasks are divided into daily and emergency, that is, before the central control device 301 issues an abnormal alarm based on the uploaded data, it is necessary to issue an alarm for the current abnormality in a timely manner. Figure 4 A communication tower 401, a network switch 402, an edge computing box 403 and other information transmission equipment are set in the computing communication device 303.

[0100] Reference Figure 4, the communication tower 401 is used to receive real-time data from the inspection drone 305 and transmit it to the network switch 402. In particular, based on the inspection process, the inspection drone 305 is distributed in different areas of the power plant. In order to be able to receive the data sent by the inspection drone in a timely manner, the communication tower 401 needs to have wide-area coverage capability and high-performance wireless access technology. In certain implementations of this embodiment, the communication tower 401 is set to include at least the following hardware configurations: (1) Multi-band antenna system: supports multiple communication frequency bands, including 5G, Wi-Fi 6, LTE and dedicated frequency bands to ensure stable connection between the drone and ground equipment. (2) Redundant power supply design: It should have dual power supply redundancy or uninterruptible power supply (UPS) support to ensure uninterrupted communication at critical moments.

[0101] The network switch 402 is responsible for diverting the data sent by the inspection drone 305, with one part being transmitted to the edge computing box 403 for local preprocessing, and the other part being uploaded to the data transmission device 302. Among them, the diverted data is not exactly the same. The data transmitted to the edge computing box 402 is usually data that requires local rapid response and preprocessing, such as real-time video streams of drones, real-time monitoring data of sensors, etc. After preprocessing, this type of data may only upload key information or results obtained after analysis to the upper-level system. The data uploaded to the data transmission device 302 may include original data, streamlined data after preprocessing, and other data that does not require immediate processing but requires centralized storage or subsequent analysis. This diversion method ensures the real-time nature of key tasks, while reducing unnecessary data transmission and improving the overall operating efficiency of the system.

[0102] Specifically, in the process of diverting the data uploaded by the inspection drone 305, the network switch 402 reasonably distributes the data uploaded by the inspection drone 305 to the edge computing box 403 and the upper system according to different data processing requirements and priorities. Part of the data is transmitted to the edge computing box 403 for localized preprocessing to reduce the load of the upper network and the data processing pressure of the central control layer; the other part is uploaded to the data transmission device 302 for further centralized processing and storage.

[0103] Furthermore, based on the fact that the network switch 402 is responsible for data transmission and diversion, in order to ensure the smoothness of the data transmission process and meet the real-time transmission requirements of large-scale data. In certain implementations of this embodiment, the network switch 402 is set to support industrial standard protocols and have high bandwidth, low latency and redundancy capabilities. Specifically, it includes at least the following hardware configurations: (1) High-performance ports: at least 10 Gigabit Ethernet ports, supporting optical fiber and copper cable connections to ensure high-speed data transmission. (2) QoS (Quality of Service) support: ensure priority processing of different data streams (such as video, control instructions, alarm information). (3) Redundant design: support stacking, dual uplink redundancy and fast fault switching functions to improve system reliability.

[0104] After the edge computing box 403 receives the data diverted by the network switch 402, it can quickly respond to security events by performing preliminary analysis and filtering of the data locally, especially when dealing with tasks with high real-time requirements such as abnormal behavior detection or emergency response. At the same time, the edge computing box 403 is close to the data source and can perform preliminary data analysis locally, reducing the bandwidth and transmission time required for data to be transmitted back to the central server, and reducing network latency. Furthermore, through local preprocessing, the edge computing box 403 can filter unnecessary or redundant data and reduce the amount of data uploaded to the central control layer, thereby saving bandwidth resources and avoiding network congestion.

[0105] Specifically, the edge computing box 403 should have efficient processing capabilities and reliable hardware configuration to support local processing and analysis of real-time data and reduce dependence on central servers. In certain implementations of this embodiment, it includes at least the following hardware configurations: (1) Multi-core processor: high-performance CPU or GPU for real-time data processing and AI algorithm reasoning. (2) Large-capacity memory and storage: at least 16GB of memory and 1TB of SSD storage to meet the needs of edge computing tasks. (3) Durable industrial design: dustproof, waterproof, shockproof, and adaptable to harsh environments.

[0106] In this embodiment, refer to Figure 3 When the computing communication device 303 obtains the data uploaded by the inspection drone 305, the security data obtained by the security device 306 can also be transmitted to the upper layer through the network switch for centralized processing.

[0107] The data transmission equipment 302 is the data aggregation and transmission backbone of the entire power plant safety monitoring system, ensuring efficient data transmission between various layers, and mainly includes data transmission equipment such as network switches. This layer transmits data to the central control device 301 safely and efficiently through various transmission equipment, ensuring the overall connectivity and data processing capabilities of the system.

[0108] In some implementations of this embodiment, the central control device 301 is responsible for the overall monitoring and decision-making of the power plant. Figure 5 A core aggregation switch 501 , a security host computer 502 , a security server 503 , a cloud backup server 504 , and a control center 505 may be set in the central control device 301 .

[0109] Reference Figure 5 The core aggregation switch 501 receives all security data from the data transmission device 302, and distributes the data to the security host computer 502 and the security server 503 for processing and storage. Among them, the core aggregation switch 501, as the core backbone of the power plant safety monitoring system, needs to have ultra-high throughput and reliable redundancy mechanism to support all data traffic of the central control layer. In certain implementations of this embodiment, the core aggregation switch 501 is set to include at least the following hardware configurations: (1) Multi-port support: at least support 40G / 100G high-speed ports to ensure the aggregation and transmission of all security data. (2) Hardware acceleration: integrated hardware forwarding engine to ensure line-speed data forwarding and adapt to large-scale data center environments. (3) High availability design: provide dual power supply and fan redundancy, support hot-swappable modules, and ensure long-term stable operation of the equipment.

[0110] The security host computer 502 is used for real-time monitoring and task distribution. Specifically, during the real-time monitoring process, the security host computer 502 directly obtains real-time security data from the edge computing box 403, the network switch 402, the communication tower 401, etc., such as the high-resolution video stream transmitted by the drone, the environmental data fed back by the sensor, the status information of the access control system, etc., and presents these data to the operator through a graphical interface to ensure that the operator can view the situation of the key areas of the power plant in real time in the control center. At the same time, the security host computer 502 is preset with a threshold. When certain key parameters exceed the normal range, such as abnormal chimney temperature, foreign objects or unknown persons detected by the drone, etc., the security host computer 502 will send an alarm signal to notify the operator to take emergency measures. Among them, the security host computer 502 can also collect real-time video streams from the video surveillance system, allowing the operator to view the real-time picture of a specific area, and even replay historical videos to investigate previous events.

[0111] In the process of task issuance, the security host computer 502 issues inspection tasks based on operator input or preset automated programs, such as controlling drones to perform specific inspection tasks or triggering access control systems to lock a certain area. The security host computer can also collect real-time video streams from video surveillance systems, allowing operators to view real-time images of specific areas and even replay historical videos to investigate previous events.

[0112] The security server 503 is used for data analysis and storage. Among them, the security server 503 can identify the normal operation mode and abnormal behavior of personnel and equipment in the power plant by analyzing the drone and sensor data. If it is detected that the personnel are not wearing safety equipment, abnormal movement path, etc., an analysis report will be generated to prompt potential safety hazards. At the same time, the security server can conduct in-depth analysis of the data of drones and security equipment, identify the health status of the equipment, predict when maintenance or replacement is required, and reduce the risk of equipment failure. Further, the security server 503 can perform statistical analysis on the security data collected over a long period of time to find the trend of safety hazards. For example, through video analysis, it can be found that there are more security events in certain areas within a specific time period, thereby providing a basis for optimizing security strategies. Preferably, the security server 503 analyzes the data of the video surveillance system through a pattern recognition algorithm, and can promptly identify the behavior of unauthorized personnel entering key areas, or equipment abnormalities found during drone inspections, such as oil leakage, temperature abnormalities, etc. At the same time, based on the data analysis results, the security server 503 can generate action suggestions to help operators make decisions. For example, when an abnormal temperature change is found, it is recommended to dispatch a drone for detailed inspection, or it is recommended to adjust the authorization policy of the access control system to restrict the entry of suspicious persons. These analysis results can not only provide timely decision support for operators, but also provide a data-driven basis for future security optimization.

[0113] The cloud backup server 504 is used for remote backup of data. Through the control center, the staff can monitor the security status of the entire power plant and issue drone missions and emergency response instructions through the host computer.

[0114] In this embodiment, the security server 503 and the cloud backup server 504 are responsible for storing, analyzing and backing up large amounts of data, and must have high computing power and storage scalability. In certain implementations of this embodiment, the security server 503 and the cloud backup server 504 are configured with at least the following hardware structure settings: (1) Multi-core processor: at least a dual-core Xeon processor or a multi-core processor of the same level to support virtualization and data-intensive tasks. (2) Large-capacity storage: a combination of high-speed SSDs and HDDs in a RAID configuration, providing more than tens of TB of storage space to support long-term data storage. (3) High-bandwidth network interface: dual 10 Gigabit Ethernet interfaces to ensure high-speed data transmission and cloud synchronization backup.

[0115] On the other hand, in some implementations of this embodiment, reference may be made to Figures 3 to 5The power plant safety monitoring system shown controls the inspection drone in the monitoring system to perform safety monitoring of the power plant. Figure 6 , mainly including step 601 to step 604, the steps are mainly:

[0116] Step 601: Determine a target area to be inspected and an inspection drone to inspect the target area based on a central control device, so as to generate an inspection task corresponding to the inspection drone.

[0117] In this embodiment, this step mainly includes: the central control device receives an operation instruction sent by an operator, and determines whether it is a routine inspection or an emergency inspection according to the operation instruction.

[0118] In the case of daily inspection, the daily inspection tasks are imported and the functions are set up through the preset UAV task planning software in the control center of the central control device. The central control device uses the task scheduling algorithm (such as load balancing scheduling) to allocate tasks to specific UAVs based on the operator's pre-set settings, BIM or other three-dimensional models, historical inspection data, and key area information of the power plant.

[0119] When it is an emergency inspection, the operator or central control device automatically identifies that a specific device or area in the power plant needs an emergency inspection. The mission planning software combines real-time sensor data, BIM models or other 3D models and historical inspection records to generate special mission requirements. The system calls the mission scheduling algorithm to assign the appropriate drone to perform the mission. The scheduling algorithm takes into account parameters such as the remaining power of the drone, the distance to the target location, and the sensor status.

[0120] Among them, before the central control device obtains the three-dimensional model, it can initially construct a three-dimensional model of the power plant based on the drone and use 3D modeling and environmental reconstruction algorithms during daily inspections to create or update an accurate three-dimensional environmental model for use in drone path planning and for power plant personnel to conduct remote inspection and maintenance planning. Specifically, the following algorithms are required in the process of constructing the three-dimensional model, including: (1) Three-dimensional reconstruction algorithm: Use SfM (Structure from Motion), multi-view geometry, and point cloud processing technology to generate a three-dimensional environmental model of the power plant to help operators better understand the on-site situation. (2) Point cloud fusion and modeling algorithm: Combine LiDAR point cloud and visual information, and use the ICP (Iterative Closest Point) algorithm to match and fuse point clouds to generate a detailed 3D model.

[0121] When the drone performs 3D modeling and environmental reconstruction tasks in the power plant, the relevant algorithms are deployed as follows: The 3D reconstruction algorithm is divided into different parts and deployed on the drone body and security server. The preliminary data processing of the algorithm, such as image and point cloud acquisition and processing, is deployed on the drone body. The drone uploads the processed data to the security server of the central control layer to complete complex 3D reconstruction and refined model creation. The point cloud fusion and modeling algorithm is divided into different parts and deployed on the edge computing box and security server. The preliminary fusion of point cloud data can be processed in the edge computing box, and the complex point cloud matching and modeling work is completed in the security server of the central control layer.

[0122] Step 602: Plan the inspection path of the inspection drone according to the inspection task, and send the inspection path and the inspection task to the inspection drone.

[0123] In this embodiment, this step mainly includes: the central control device automatically performs optimal route planning, combines AI algorithm or other path planning algorithm to generate the optimal route. Among them, considering the environment, equipment distribution and other information, ensure that the drone covers all key points and sends the inspection task to the drone.

[0124] Specifically, in this embodiment, in order to meet the inspection task requirements of the control center, the multiple drones in the power plant should be able to generate customized inspection routes based on the inspection points set by the control center, and efficiently plan the inspection paths according to the complex layout of the power plant to perform multiple tasks at the same time. In order to achieve the above technical objectives, in this embodiment, the following algorithms can be deployed in the drones, edge computing boxes and control centers, which should include: (1) Global path planning algorithm: A*, Dijkstra, RRT (Rapidly-exploring Random Tree) and other algorithms and their derivative algorithms should be used to plan the global inspection routes of drones to ensure the efficiency and coverage of inspections. (2) Dynamic path adjustment algorithm: During the inspection process, if the environment changes or an emergency occurs, the path can be adjusted in real time based on dynamic planning and graph search technology (such as D* algorithm, PRM path generator). (3) Task scheduling and multi-drone coordination algorithm: Based on mixed integer programming (MIP), reinforcement learning, task allocation algorithm (such as Hungarian algorithm) and other technologies, collaborative work, task scheduling and load balancing between multiple drones are achieved. This ensures that drones do not collide in confined areas while assigning different tasks such as target tracking or equipment detection.

[0125] Furthermore, the path planning and task scheduling algorithm involves the coordination and path optimization of a single or multiple drones. The algorithm can be deployed in drones, edge computing boxes and control centers: Global path planning algorithm: This type of algorithm is complex and requires global layout information of the power plant. It is usually deployed on the security server of the central control layer. The server plans the global path according to the inspection task and sends the instruction to the drone. Dynamic path adjustment algorithm: In order to cope with emergencies, the dynamic path adjustment algorithm is partially deployed on the local processing unit of the drone to ensure that local adjustments can be made when communication is not smooth. The algorithm can also be deployed on the edge computing device for partial calculations, update the path in real time and transmit the data to the drone to complete the adjustment. Task scheduling and multi-drone coordination algorithm: This algorithm needs to handle the task allocation and coordination of multiple drones. It is mainly deployed in the security server or security host computer of the central control layer to generate task scheduling strategies and send them to each drone through the network.

[0126] Step 603: The inspection drone flies to the inspection task point according to the inspection path, obtains image data corresponding to each inspection task point, and determines whether an abnormality occurs at the current inspection task point.

[0127] In this embodiment, this step mainly includes: the central control device synchronizes the task instructions through a remote communication algorithm (such as LoRa or 5G transmission protocol). Then, when the inspection drone obtains the task data, it completes the self-check before takeoff, including sensor status, GPS signal, IMU initialization, etc.

[0128] Specifically, during the self-inspection process of the inspection drone, the drone needs to perceive and process data from multiple sensors in a complex power plant environment, and then combine data from multiple sensors such as cameras, LiDAR, GPS, IMU, etc. to improve positioning accuracy and environmental perception capabilities, thereby ensuring the safety and efficiency of the drone in inspection tasks.

[0129] Furthermore, in order to achieve the above functions, the system should deploy a multimodal sensor fusion algorithm on the drone body, wherein the sensor data fusion algorithm is based on Kalman filtering, extended Kalman filtering (EKF), unscented Kalman filtering (UKF) and other technologies, which fuses data from multiple sensors (such as IMU, GPS, LiDAR, camera) to improve positioning accuracy and environmental perception capabilities. Next, the multi-source data integration and analysis algorithm is deployed on the drone body to integrate and analyze the data of multimodal sensors through Bayesian estimation, information theory, deep learning and other methods, so as to comprehensively evaluate the state, position and environmental conditions of the drone.

[0130] Among them, complex multi-source data analysis is initially processed by the drone body and then transmitted to the edge computing device. The final integrated analysis can be performed on the security server of the central control layer to comprehensively evaluate the environment and drone status. Therefore, such algorithms need to be divided into different parts and deployed on the drone body, edge computing devices and security servers.

[0131] Next, the drone starts the flight control system, starts the operation of algorithms such as PID control or model predictive control (MPC), and enters autonomous flight mode. The drone takes off vertically to the specified height, uses IMU, GPS, environmental map and other technologies to navigate, and flies to the first specified waypoint in the task list. Among them, the system monitors the position, speed, altitude and other parameters of the drone in real time, and the control center keeps the data synchronized. During this period, the obstacle avoidance system is started, the environment is perceived in real time through SLAM and visual sensors, and the three-dimensional model information is synchronized and updated. After arriving at the inspection point, the drone uses algorithms such as target detection to begin identifying designated equipment or areas.

[0132] Specifically, in the process of the inspection drone, in order to ensure the autonomous flight and stability of the inspection drone and avoid obstacles at the same time, in order to complete this function, it is necessary to deploy flight control and navigation algorithms. Preferably, the inspection drone can be deployed with: (1) attitude control and stability algorithm: use PID control, model predictive control (MPC) and quadrotor dynamics model to adjust the attitude and stability of the drone to ensure that the drone can fly stably under different wind speeds and environmental conditions. (2) Autonomous navigation and positioning algorithm: including GPS / INS fusion, visual inertial navigation (VIO), SLAM (Simultaneous Localization and Mapping) and other technologies to help drones achieve autonomous positioning, navigation and path tracking in the complex environment of power plants. (3) Obstacle avoidance algorithm: obstacle avoidance algorithms based on laser radar (LiDAR), ultrasound, and visual perception (such as optical flow method, deep learning obstacle avoidance) can perceive surrounding obstacles in real time and plan a safe flight path. The drone-based power plant intelligent inspection security system should use the above-mentioned technologies to ensure that the drone flies safely and stably in different areas (such as cooling towers, chimneys, plant buildings, etc.).

[0133] Furthermore, flight control and navigation algorithms are directly related to the autonomous flight and obstacle avoidance of the drone, and need to be deployed in the flight control system of the drone itself. These algorithms rely on the computing hardware inside the drone to process data in real time. The specific algorithm deployment objects are: Attitude control and stability algorithm: deployed in the flight controller (Flight Control Unit, FCU) of the drone, which is the core control hardware of the drone, and adjusts the flight attitude and stability of the drone in real time. The unit also integrates the drone's inertial measurement unit (IMU) and other sensors to provide feedback on the state of motion. Autonomous navigation and positioning algorithm: This type of algorithm mainly relies on GPS, INS (inertial navigation system) and visual inertial navigation (VIO) and other devices on the drone, and is deployed in combination with SLAM and other technologies. These algorithms can be directly deployed in the navigation processing unit of the drone. Obstacle avoidance algorithm: This algorithm relies on the perception devices on the drone, such as laser radar (LiDAR), ultrasound and camera, and is deployed in the perception and control module of the drone to perceive the surrounding environment in real time and avoid obstacles.

[0134] When the inspection drone flies to each inspection task point according to the deployed algorithm, the inspection drone obtains inspection data corresponding to the type of the inspection task. When the inspection task is a daily inspection, the inspection drone determines whether there is an abnormality (such as oil leakage, equipment wear, etc.) at the inspection point based on the acquired image data.

[0135] Specifically, in the process of identifying anomalies, the inspection drone needs to complete the target detection and tracking tasks at the specified location. In order to automatically identify abnormal phenomena during the inspection process and issue an alarm in time, the system needs to deploy visual perception and object detection algorithms. These should include: (1) Image processing and feature extraction algorithms: Classic image feature extraction methods such as SIFT, SURF, and ORB are used to identify and track specific power plant equipment or landmarks. (2) Object detection and classification algorithms: YOLO, Faster R-CNN, SSD and other algorithms based on deep learning are used to detect objects such as personnel, vehicles, and equipment in the power plant, and classify and identify them. (3) Scene segmentation and semantic analysis algorithms: Use convolutional neural networks (CNNs) such as UNet and SegNet to perform pixel-level segmentation of images and analyze the structure, equipment layout, etc. in the power plant environment. (4) Target tracking algorithms: Use visual algorithms such as Deepsort and TLD to track designated moving targets in the power plant. (5) Human behavior analysis algorithm: Through human posture estimation (such as OpenPose) and action recognition algorithms (such as temporal convolutional network TCN), the behavior of personnel in the power plant is analyzed, such as not wearing a helmet, personnel falling, dangerous behavior, etc. After configuring the above algorithms, the drone can monitor the behavior of personnel. For example, when a worker does not wear a helmet or enters a dangerous area, the system will automatically generate an alarm. At the same time, it can identify problems such as oil leakage and water leakage of equipment, or monitor critical situations such as fire and smoke, and upload this information to the control center for analysis and response.

[0136] Furthermore, based on the requirement of high computing power and real-time data processing for visual perception and object detection algorithms, the deployment locations of the above algorithms are as follows: Image processing and feature extraction algorithms: Since image information needs to be processed quickly, the algorithms can be deployed on the visual processing unit of the drone body to process the camera data of the drone in real time. Object detection and classification algorithms: Algorithms based on deep learning have high requirements for computing resources. Some processing (such as preliminary detection results) can be performed on the drone body, and then transmitted to the edge computing device through the network for preprocessing, and finally uploaded to the security server of the central control layer to complete complex object detection and classification tasks. Therefore, such algorithms need to be divided into different parts and deployed on the drone body, edge computing devices and security servers. Scene segmentation and semantic analysis algorithms: This algorithm has higher requirements for computing resources and is usually run on the security server of the central control layer to facilitate scene segmentation and environmental analysis from the image data transmitted by the drone. Target tracking algorithms and human behavior analysis algorithms: Since it involves real-time tracking of dynamic targets, some calculations need to be performed on the drone body, especially when processing preliminary tracking. More complex behavior analysis can be uploaded to edge computing devices or security servers of the central control layer for deep processing.

[0137] When the inspection task is a routine inspection, the inspection drone determines whether there is an abnormality (such as oil leakage, equipment wear, etc.) at the inspection point based on the acquired image data.

[0138] When the inspection task is an emergency inspection, the type of the emergency inspection task needs to be determined. When the emergency inspection task is a target tracking task, the drone will feed back the tracking target information (suspicious persons, vehicles, etc.) to the control center in real time.

[0139] When the emergency inspection task is an equipment monitoring task, it is determined whether there is an abnormality such as oil leakage, equipment wear, etc. at the inspection point.

[0140] Step 604: When it is determined that an abnormality exists, the inspection drone reports the abnormality information and returns after completing the inspection task.

[0141] In this embodiment, this step mainly includes: when the inspection drone determines that there is an abnormality at the inspection point, the drone generates and uploads an abnormality report to the control center. The abnormality information includes location, type, severity, etc. At the same time, the control center generates an early warning or schedules a maintenance task.

[0142] Determine whether the drone has completed all the inspection tasks in the list. If not, continue the inspection. If it has been completed, the drone starts the return program and uses the path planning algorithm to calculate the best return route. The drone returns to the take-off and landing point, gradually lowers the flight altitude, uses ranging sensors, IMU, etc. to accurately control the landing position, and lands smoothly on the designated drone platform.

[0143] After the drone completes its return flight, the central control device summarizes the flight data and inspection report, analyzes the drone's flight trajectory, sensor data, and possible anomalies, generates a final inspection report, and stores the data or uploads it to the cloud for backup.

[0144] Specifically, in the process of uploading data by the inspection drone, in order to ensure that the massive data generated during the inspection of multiple drones can be smoothly transmitted to the control center and to ensure low-latency real-time monitoring, the drone-based power plant safety monitoring system should deploy remote communication and data transmission algorithms, which should include: (1) Data compression and transmission algorithms: Through compressed sensing and video encoding technology (such as H.264 / HEVC), optimize the transmission of large-scale video and sensor data to ensure low-latency and efficient transmission. (2) Network optimization and load balancing algorithms: Use network traffic prediction, dynamic bandwidth allocation and load balancing algorithms to optimize the data link between drone control centers and ensure priority transmission of mission-critical data. Ensuring stable and efficient data transmission between drones and control centers is the key to remote communication and data transmission algorithms, and the relevant algorithms are deployed as follows:

[0145] Data compression and transmission algorithm: Deployed in the communication module of the drone body, compression algorithms (such as H.264 / HEVC and other video compression technologies) are used to reduce the amount of data transmission. Some pre-processed data can also be compressed in the edge computing box of the computing communication layer.

[0146] Network optimization and load balancing algorithm: This algorithm needs to be deployed in the network switch of the data transmission layer to ensure the optimization of the data transmission link and the reasonable allocation of dynamic bandwidth, and to ensure the real-time upload of key data of the drone. At the same time, the security server of the central control layer also deploys relevant algorithms to ensure the global load balancing of the system.

[0147] In this embodiment, refer to Figures 3 to 5 The safety monitoring system, the algorithm used in step 601 to step 607 can be deployed on the local equipment device 304 (drone and other security local equipment) that meets the corresponding hardware requirements, the computing communication device 303 (edge ​​computing device, network switch, etc.), the data transmission device 302 (network switch and other communication equipment), the central control device 301 (security server, cloud backup server, etc.). Further, in order to ensure the long-term stable operation of the power plant safety monitoring system, it is necessary to consider the maintainability and upgrade strategy of the system to ensure the smooth operation of each algorithm. The system design should include modular software and hardware architecture so that each component can be upgraded and replaced independently. The algorithm should be updated through the software version management tool to ensure that the new algorithm version is compatible with the existing system and smoothly transition. In terms of hardware, the equipment should have a pluggable design to facilitate rapid replacement or upgrade in case of failure, ensuring that it can support the full load operation of the algorithm. Regular system maintenance plans should include hardware inspection and cleaning, software updates and patch installation, and performance monitoring and optimization. Through automated maintenance tools and alarm systems, operation and maintenance personnel can promptly detect and handle potential problems in the system to ensure that the system is always in optimal operating condition.

[0148] The present embodiment discloses a power plant safety monitoring system based on drones and a monitoring method thereof. First, in actual application scenarios, the automated drone can accurately and efficiently perform inspection tasks according to the preset schedule and route without interference from human factors. It can cover a wide area in a short time, greatly reducing the time investment of manual inspections, and at the same time increasing the inspection frequency. This efficiency improvement is particularly significant in large-scale power plants, which can significantly reduce equipment downtime and maintenance intervals. The automatic inspection is based on AI algorithms for data collection and analysis, eliminating subjective errors and negligence that may occur in manual inspections. Through precise path planning, sensor fusion and visual recognition technology, each inspection can provide consistent and standardized data, providing a reliable basis for subsequent analysis and equipment health assessment. Secondly, drone automatic inspection is particularly suitable for dangerous environments in power plants such as high temperature, high pressure, and chemical hazardous areas, avoiding direct exposure of staff and ensuring personnel safety. At the same time, drones can operate flexibly in narrow and inaccessible areas, further reducing the risks of manual operation. Then the drone automatic inspection system can quickly identify abnormal conditions of equipment or environment through multi-sensor fusion and real-time monitoring technology, and immediately feedback to the control center. This rapid response mechanism, combined with AI's predictive maintenance algorithm, can significantly reduce the incidence of equipment failures and avoid losses caused by sudden accidents, while manual operations are relatively lagging in response speed and accuracy. At the same time, the drone automation solution reduces the number of operators and inspection frequency in manual inspections, reducing the operating costs of power plants. In the long run, the automation solution can further reduce maintenance and replacement costs by reducing equipment failures and extending equipment life. At the same time, with the popularization of automatic inspections, the system maintenance costs are gradually reduced, and the initial investment can be quickly amortized. Furthermore, in a complex power plant environment, multiple automated drones can coordinate work to achieve multi-tasking. This makes inspections not limited to a single task or area, and can simultaneously monitor different areas and different equipment, avoiding the inefficiency of serial execution of tasks in manual inspections. Preferably, the automatic inspection system can be dynamically adjusted according to the real-time situation of the power plant. Drones can quickly respond to emergencies or abnormal signals, conduct emergency inspections or track directional targets without waiting for manual operation instructions, greatly shortening the response time. Finally, the automatic inspection system can continuously collect equipment operating status, environmental data and inspection information to form a complete data archive. This data can be used for long-term trend analysis, helping power plants to perform accurate predictive maintenance and improve equipment operating efficiency and reliability. Through machine learning algorithms, the system can continuously optimize inspection strategies and improve the accuracy of future inspections. Compared with manual control, the drone automated inspection system can maximize battery life and reduce energy waste through path optimization, intelligent charging scheduling and other means.In addition, the automated inspection system can reduce carbon emissions caused by vehicle inspections and personnel transfers, and has significant environmental benefits. The automated inspection system can be deeply integrated with the power plant's SCADA system, equipment management platform, security system, etc., providing comprehensive data aggregation and analysis functions. Through the analysis of drone inspection data, the power plant management can obtain the operating status from a global perspective, support more intelligent operational decisions, and improve the overall management level of the power plant.

[0149] Secondly, in terms of the network structure and equipment deployment design of the UAV-based power plant intelligent inspection and security system, a targeted network architecture solution and technical requirements for equipment deployment are provided for the UAV-based automated inspection solution. This solution ensures efficient interconnection between UAVs, communication towers, edge computing equipment, core switching equipment and security servers in various areas of the power plant. Through a reasonably designed network architecture, inspection data can be transmitted, analyzed and fed back in real time, effectively improving the response speed of inspection tasks. Compared with traditional manual operation solutions, the scalability of this network architecture lays the foundation for the integration of new equipment or new functions in the future.

[0150] Furthermore, in terms of the algorithm deployment strategy design for the UAV-based power plant intelligent inspection and security system, targeted algorithm deployment solutions, algorithm deployment location requirements, and algorithm system maintenance requirements are provided for the UAV-based automated inspection solution. This algorithm deployment strategy optimizes the intelligence level of power plant UAVs, allowing the system to effectively perform multiple intelligent operations such as task allocation, path planning, target detection and tracking. These designs enhance the intelligence, flexibility, and fault recovery capabilities of the system, thereby improving the safety and efficiency of power plant operations.

[0151] Finally, in terms of the design of the application process solution for the drone-based intelligent inspection and security system of power plants, the drone-based automated inspection solution provides application processes for daily routine inspections of power plants and application processes for emergency situations or customized inspection tasks in specific areas of power plants. The overall process design not only improves the flexibility and efficiency of inspections, but also provides faster response capabilities and higher levels of protection for the safe operation of power plants, thereby reducing equipment failure rates and accident rates.

[0152] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A power plant safety monitoring method based on drones, characterized in that: include: In response to the inspection instruction, determine the target area to be inspected in the power plant and the inspection drone corresponding to the target area; Determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path of the inspection drone according to the inspection task points; Control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area; A safety analysis is performed on the power plant according to the inspection image data to obtain a safety status of the power plant.

2. The power plant safety monitoring method based on drone according to claim 1 is characterized in that: The step of determining a target area to be inspected in the power plant and an inspection drone corresponding to the target area includes: Determine the type of inspection task according to the inspection instruction; wherein the type of inspection task includes daily inspection task and emergency inspection task; When it is determined that the inspection task is a daily inspection task, the regional division information and drone deployment information of the power plant are obtained, and the inspection drone corresponding to each inspection area in the power plant is determined; When it is determined that the inspection task is an emergency inspection task, the location information of the emergency inspection area is obtained to determine the inspection drone corresponding to the emergency inspection area according to the location information and the drone deployment information.

3. A power plant safety monitoring method based on drone according to claim 2, characterized in that: Generating the inspection path of the inspection drone according to the inspection task points includes: Retrieving a three-dimensional environment model corresponding to the power plant, and obtaining location information of each inspection task point in the three-dimensional environment model; According to the location information, an inspection path of the inspection drone in the target area is generated by a preset path planning algorithm.

4. The power plant safety monitoring method based on drone according to claim 1 is characterized in that: The controlling the inspection drone according to the inspection path includes: Controlling the inspection drone to take off vertically to a preset height; According to the preset height, obtaining the environmental information corresponding to the inspection path in real time; When it is determined according to the environmental information that there is an obstacle on the inspection path, the inspection drone is controlled to perform an obstacle avoidance action.

5. The power plant safety monitoring method based on drone according to claim 2 is characterized in that: The obtaining of inspection image data corresponding to the target area includes: When the inspection task is an equipment detection task in an emergency inspection task, the target image data corresponding to each inspection task point is obtained to determine whether there is an abnormal target at the current inspection task point according to the target image data; When the abnormal target exists, the inspection drone is controlled to track the abnormal target in real time, and an abnormal target image corresponding to the abnormal target is acquired.

6. The power plant safety monitoring method based on drone according to claim 2 is characterized in that: The obtaining of inspection image data corresponding to the target area includes: When the inspection task is an equipment detection task in an emergency inspection task, the equipment image data corresponding to each inspection task point is obtained to determine whether an abnormality occurs in the equipment at the current inspection task point according to the equipment image data; When it is determined that the device is abnormal, an abnormal device image corresponding to the device is acquired.

7. The power plant safety monitoring method based on drone according to claim 3 is characterized in that: Before performing safety analysis on the power plant according to the inspection image data, the method includes: Obtaining the positioning information of the inspection drone in real time; According to the positioning information and the location information of the inspection task point, determining whether the inspection drone has completed the inspection task; When it is determined that the inspection drone completes the inspection task, the inspection drone is controlled to return.

8. A power plant safety monitoring system based on drones, characterized in that: Including security control module, computing and communication module, security inspection module and security analysis module; The placement control module is used to respond to the inspection instruction and determine the target area to be inspected in the power plant and the inspection drone corresponding to the target area; The edge computing module is used to determine a number of inspection task points corresponding to the target area according to the inspection instruction, so as to generate an inspection path of the inspection drone according to the inspection task points; The safety inspection module is used to control the inspection drone according to the inspection path to obtain inspection image data corresponding to the target area; The safety analysis module is used to perform safety analysis on the power plant according to the inspection image data to obtain the safety status of the power plant.

9. The power plant safety monitoring system based on drone according to claim 8 is characterized in that: The security control module includes an inspection task unit and a drone dispatching unit; The inspection task unit is used to determine the type of inspection task according to the inspection instruction; wherein the type of inspection task includes daily inspection task and emergency inspection task; The drone scheduling unit is used to obtain the area division information and drone deployment information of the power plant when it is determined that the inspection task is a daily inspection task, and determine the inspection drone corresponding to each inspection area in the power plant; when it is determined that the inspection task is an emergency inspection task, obtain the location information of the emergency inspection area, and determine the inspection drone corresponding to the emergency inspection area based on the location information and the drone deployment information.

10. The power plant safety monitoring system based on drone according to claim 9, characterized in that: The edge computing module includes a positioning unit and a path planning unit; The positioning unit is used to retrieve the three-dimensional environment model corresponding to the power plant and obtain the position information of each inspection task point in the three-dimensional environment model; The path planning unit is used to generate an inspection path for the inspection drone in the target area according to the location information through a preset path planning algorithm.

Images