Decoking system and method for thermal power plant boiler by using unmanned aerial vehicle to carry vibrator

Abstract

The invention belongs to the technical field of thermal power generation boiler maintenance, and relates to a decoking system and method for a thermal power plant boiler by using an unmanned aerial vehicle to carry a vibrator. Comprising a multi-rotor unmanned aerial vehicle; the composite vibration decoking module is arranged below the multi-rotor unmanned aerial vehicle; the multi-mode control module is connected with the multi-rotor unmanned aerial vehicle and the composite vibration decoking module; and the AI coking identification system is arranged on the multi-rotor unmanned aerial vehicle and is connected with the multi-mode control module. According to the invention, through unmanned, intelligent and accurate removal of the coke in the boiler, and through fusion of AI visual identification, self-adaptive vibration decoking and unmanned aerial vehicle autonomous control, the safety, the coverage range and the operation efficiency of decoking operation are greatly improved, and unmanned, intelligent and accurate removal of the coke in the boiler is realized.

Description

Technical Field

[0001] This invention belongs to the field of thermal power boiler maintenance technology, and relates to a coke removal system and method for thermal power plant boilers using a drone-mounted vibrator. Background Technology

[0002] In thermal power plants, the boiler, as the core heat energy conversion equipment, directly affects the safety and economy of the unit. During coal combustion, ash melts at high temperatures and adheres to the inner wall of the furnace, easily forming coke deposits. In severe cases, this can lead to deterioration of heat transfer on the heating surfaces, localized overheating, and even tube rupture. Therefore, regularly removing coke deposits from inside the boiler is a crucial maintenance measure to ensure stable unit operation. Traditional decoking methods mainly rely on manual methods using long-handled mechanical hammering or high-pressure water jet cleaning. These operations are usually carried out after the boiler is shut down, resulting in high labor intensity, long operation cycles, and limited coverage. In recent years, some power plants have attempted to use fixed rapping devices or robotic wall-climbing equipment for automated decoking; however, the former has a limited operating area, and the latter is difficult to adapt to complex furnace structures and the need for working at heights.

[0003] The aforementioned existing technical solutions still have significant drawbacks: manual or semi-automatic methods cannot achieve comprehensive coverage of high and concealed areas of the boiler, and operators must enter high-temperature, confined spaces, facing significant safety risks such as falls and burns; at the same time, they lack real-time sensing capabilities and cannot accurately adjust decoking parameters according to the type of coking, resulting in incomplete removal or damage to the furnace wall. In addition, traditional methods are inefficient and cannot meet the development trend of intelligent and unmanned operation and maintenance in modern power plants. Summary of the Invention

[0004] To address the problems in the prior art, this invention provides a coke removal system and method for thermal power plant boilers using a drone equipped with a vibrator, which significantly improves the safety, coverage, and efficiency of coke removal operations.

[0005] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention provides a coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, comprising: Multi-rotor drones; A composite vibration defocusing module is located below the multi-rotor UAV; A multimodal control module, which is connected to the multi-rotor UAV and the composite vibration defocusing module; The AI ​​coking recognition system is installed on the multi-rotor drone and connected to the multimodal control module.

[0006] Preferably, the composite vibration descaling module includes an electromagnetic vibrator, a mechanical impact head, and a vibration parameter adaptive control unit, wherein the vibration parameter adaptive control unit is connected to the electromagnetic vibrator and the mechanical impact head respectively; and the vibration parameter adaptive control unit is connected to the multimodal control module.

[0007] Preferably, the AI ​​focus recognition system includes a visible light camera, an infrared thermal imager, an edge computing unit, and a heat map generation module; the visible light camera and the infrared thermal imager are respectively connected to the image input terminal of the edge computing unit; the data output terminal of the edge computing unit is connected to the heat map generation module; and the heat map generation module is connected to the multimodal control module.

[0008] Preferably, the edge computing unit is equipped with a hybrid neural network model based on YOLOv5s and Transformer.

[0009] Preferably, the multi-rotor drone includes: case; A six-axis redundant power system is mounted on the housing; The inertial measurement unit (IMU) is located inside the housing. A lidar obstacle avoidance module is disposed outside the housing; The gimbal suspension mechanism is located below the housing and is fixedly connected to the composite vibration desiccant module.

[0010] Preferably, the multimodal control module includes: Dual-link communication module, connecting to the ground control station; A dynamic obstacle avoidance module is connected to the lidar obstacle avoidance module. The control module is connected to the dual-link communication module, the dynamic obstacle avoidance module, and the six-axis redundant power system, respectively.

[0011] Preferably, the dual-link communication module includes a 5G communication unit and a WiFi communication unit; the input terminals of the 5G communication unit and the WiFi communication unit are both connected to the ground control station, and the output terminals of the 5G communication unit and the WiFi communication unit are both connected to the control module.

[0012] Secondly, the present invention provides a method for removing coke from a thermal power plant boiler using a vibrator mounted on a drone, comprising the following steps: The AI ​​coking identification system identifies and locates coking areas inside the boiler and generates coking information. The multimodal control module receives the coking information and controls the multi-rotor UAV to fly to the target coking area based on the coking information; The multimodal control module controls the composite vibration decoking module to start, and performs vibration impact decoking operation on the target coking area; After the descorching operation is completed, the target descorched area is re-identified by the AI ​​descorching recognition system to verify the descorching effect.

[0013] Preferably, the method by which the multi-modal control module controls the start-up of the composite vibration desiccant module includes: Based on the recognition results of the target coking area by the AI ​​coking recognition system, the coking type is determined to be either loose coking or hard coking. If the coking is loose, the composite vibration decoking module is controlled to operate in a preset vibration mode with a frequency of 50Hz to 200Hz. If the coking is hard, the composite vibration decoking module is controlled to operate in a preset impact mode, and the impact force of the impact mode is 5N~80N.

[0014] Preferred options also include: The multimodal control module controls the multi-rotor UAV to perform a comprehensive inspection of the boiler's inner wall based on a preset spiral scanning path. During the inspection process, when the AI ​​coking recognition system detects coking, it automatically performs a coking removal operation. After the decoking operation is completed, continue to inspect along the preset path until all inspection points are completed or a stop command is received.

[0015] Compared with the prior art, the present invention has the following beneficial effects: By integrating an AI-powered coking identification system into a high-temperature resistant multi-rotor drone, autonomous identification and location of coking deposits inside boilers are achieved. Combined with a composite vibration coking removal module, the vibration frequency and impact force can be adaptively adjusted according to the hardness of the coking deposits, avoiding damage to the furnace wall or incomplete removal caused by uniform processing. The drone's flexible flight capability overcomes the problem of traditional equipment being unable to reach high places and blind spots, improving coking removal coverage and operational safety. This invention solves the technical problems of high risk, low efficiency, and insufficient intelligence in manual coking removal, achieving unmanned, precise, and efficient coking removal results. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1This is a schematic diagram of a coke removal system for a thermal power plant boiler that utilizes a vibrator mounted on a drone, according to the present invention.

[0018] The components include: 1. Multi-rotor drones; 2. Composite vibration desiccation module; 3. AI coking recognition system. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0024] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0025] The present invention will now be described in further detail with reference to the accompanying drawings: The first objective of this invention is to provide a coke removal system for thermal power plant boilers that utilizes a vibrator mounted on a drone, comprising: Multi-rotor drone 1; The composite vibration defocusing module 2 is located below the multi-rotor UAV 1; A multimodal control module is connected to the multi-rotor UAV 1 and the composite vibration defocusing module 2; The AI ​​coking recognition system 3 is installed on the multi-rotor drone 1 and connected to the multimodal control module.

[0026] Among them, the multi-rotor UAV 1 serves as the core mobile carrier, enabling it to fly flexibly in the complex space inside the boiler and achieve accessibility to areas that are difficult to cover by traditional methods, such as vertical shafts, screen-type superheaters, and the area around burners; the composite vibration decoking module 2 is installed on the lower part of the UAV and is used to perform physical vibration or impact actions after approaching the target to loosen and remove the coking layer; the multi-modal control module is responsible for coordinating flight control commands and decoking operation logic to ensure stable operation of the system in a dynamic environment; and the AI ​​coking recognition system 3 is deployed on the UAV to collect real-time image information of the boiler's inner wall and complete the detection and positioning of the coking area based on artificial intelligence algorithms.

[0027] Specifically, the multi-rotor UAV 1 adopts a multi-axis layout design, possessing strong load capacity and flight stability. It can carry the composite vibration decoking module 2 and the AI ​​coking recognition system 3 into the boiler to perform tasks. This platform can adapt to high-temperature, dusty, and electromagnetic interference environments, supports long-term hovering and precise trajectory tracking, and is suitable for precision operations in confined spaces. In practical applications, the multi-rotor UAV 1 can be configured in quadcopter, hexacopter, or octaco configurations, depending on actual load requirements. Its power system uses a brushless motor paired with a carbon fiber propeller, improving thrust-to-weight ratio and energy efficiency. The airframe can be made of high-temperature resistant engineering plastics or lightweight alloy materials, ensuring strength while reducing overall weight.

[0028] The composite vibration decoking module 2 is located below the multi-rotor UAV 1 and is typically fixedly connected via a rigid bracket or shock-absorbing suspension mechanism to prevent the reaction force during operation from affecting the flight attitude. This module can generate high-frequency micro-amplitude vibrations or instantaneous impact forces, transferring energy to the coke layer attached to the boiler's heating surface, causing it to fracture and detach due to fatigue. This module can integrate multiple actuation modes, such as electromagnetic drive, piezoelectric ceramic excitation, or pneumatic hammering, switching operating states according to different coking characteristics to improve its applicability.

[0029] The multimodal control module establishes a two-way communication connection with the multi-rotor UAV 1 and the composite vibration desiccation module 2, receives identification information from the AI ​​coking recognition system 3, and generates flight path planning, attitude adjustment commands, and start / stop commands for the desiccation device accordingly. This module has the capability to switch between multiple control modes, such as remote control, semi-automatic cruise, and fully autonomous operation modes, which can be flexibly selected according to site conditions.

[0030] The core hardware of the multimodal control module is the main control computer (such as a high-performance embedded industrial control board), which interconnects with the entire system through rich communication interfaces (CAN, UART, Ethernet, etc.). The power management unit connects via I... 2 The C / ADC interface connects to the main control computer, responsible for powering the entire module and monitoring its power status. The signal conditioning circuit, a crucial component of the data channel, connects to the outputs of various external sensors (sensors on the multi-rotor UAV 1) on one end and to the input port of the main control computer on the other. It filters, amplifies, and converts the format of the raw signals; simultaneously, it amplifies the control signals output by the main control computer to drive the UAV's power system and the composite vibration defocusing module 2. At the software level, a flight control algorithm (such as a cascaded PID controller) runs on the main control computer. This algorithm processes the fused sensor data in real time and calculates control commands for the UAV's power system. The task scheduling logic, acting as a higher-level state machine, sequentially calls flight control, obstacle avoidance, and defocusing sub-functions based on AI recognition results and operational procedures. The fault diagnosis mechanism continuously monitors the health of the power supply, communication links, and sensor data. Upon detecting an anomaly, it immediately takes over control and executes preset safety strategies (such as hovering or returning to base).

[0031] The AI-powered coking identification system 3, mounted on a multi-rotor UAV 1, directly participates in data acquisition and preliminary analysis, avoiding the latency issues caused by transmitting all raw images back to the ground station and improving response speed. This system acquires visual information about the boiler's interior through cameras or other optical sensors, and combines this with a deep learning model to perform semantic segmentation and classification of coking areas in the images, outputting information labels including location coordinates, area size, and severity. This identification result serves as a key input to trigger subsequent coking removal actions, realizing a shift from "passive discovery" to "active intervention."

[0032] The composite vibration descaling module 2 includes an electromagnetic vibrator, a mechanical impact head, and a vibration parameter adaptive control unit; the vibration parameter adaptive control unit is connected to the electromagnetic vibrator and the mechanical impact head respectively; and the vibration parameter adaptive control unit is connected to the multi-modal control module.

[0033] This module comprises two different types of actuation devices: an electromagnetic vibrator and a mechanical impact head. The electromagnetic vibrator, driven by an alternating electromagnetic field, generates high-frequency reciprocating motion and is suitable for removing loose or powdery coke deposits. Its working principle involves using rapid periodic vibration to induce micro-crack propagation within the coke layer, thereby causing loosely adhered coke to detach. The mechanical impact head, on the other hand, applies instantaneous high-energy impacts to the surface of hard coke, relying on localized stress concentration to break the interface between the coke and the metal wall. It is suitable for removing dense, high-strength ash deposits. Both devices can operate independently or in tandem, forming a multi-mode coke removal capability and enhancing the system's adaptability to complex coking conditions.

[0034] The vibration parameter adaptive control unit establishes electrical connections with both the electromagnetic vibrator and the mechanical impact head to regulate their output parameters, such as vibration frequency, amplitude, impact force, and duration. Simultaneously, this control unit communicates with the multimodal control module to receive control commands and environmental sensing information (e.g., coking type identification results) from the upper-level system. Based on this, the control unit automatically selects the optimal operating mode and parameter combination according to a preset logic or algorithm model. For example, upon receiving a "loose coking" judgment signal, the control unit will activate the electromagnetic vibrator and set its operating frequency to the mid-to-high frequency range (e.g., 100Hz) to avoid unnecessary structural disturbances caused by low-frequency, large-amplitude vibrations. When facing hard coking, it activates the mechanical impact head and adjusts the output impact force to an appropriate level (e.g., 30N) to ensure effective crushing while preventing overload damage to the furnace wall material.

[0035] The AI ​​focus recognition system 3 includes a visible light camera, an infrared thermal imager, an edge computing unit, and a heat map generation module; the visible light camera and the infrared thermal imager are respectively connected to the image input terminal of the edge computing unit; the data output terminal of the edge computing unit is connected to the heat map generation module; and the heat map generation module is connected to the multimodal control module.

[0036] The visible light camera is used to collect visible light images of the boiler's inner wall and coking areas, acquiring the geometric contours, distribution patterns, and relative positions of the coking material. This camera possesses high-resolution imaging capabilities, clearly capturing millimeter-level edge features of the coking blocks and providing accurate spatial visual data under suitable lighting conditions. Its installation position is fixed at the front of the UAV platform or below the gimbal, ensuring the field of view covers the effective range of the composite vibration coking removal module 2. In some optional embodiments, the visible light camera can be replaced with an industrial camera with low-light enhancement capabilities to adapt to the locally dim environment inside the boiler; a wide-angle lens configuration can also be used to expand the field of view and improve inspection efficiency.

[0037] Infrared thermal imagers are used to detect temperature distribution differences in different areas within a boiler, identifying "active coking" areas caused by uneven combustion or coking buildup. These areas typically exhibit a significantly higher temperature gradient than the surrounding walls (temperature difference ≥15℃), serving as an important indicator of potential coking development. Infrared thermal imagers, through non-contact temperature measurement, can operate effectively even in smoke-filled or low-light conditions, compensating for the blind spots of visible light cameras in low-contrast scenes. Their detectors have a resolution of at least 640×512 pixels and a thermal sensitivity ≤50mK, enabling precise detection of minute temperature changes. Furthermore, infrared thermal imagers can integrate short-wave infrared (SWIR) or multispectral imaging capabilities, further enhancing their ability to penetrate and identify semi-transparent or minute coke layers.

[0038] The edge computing unit is responsible for receiving raw image signals from visible light cameras and infrared thermal imagers, and performing key algorithmic tasks such as dual-source data fusion, target detection, and feature extraction. Deployed on a UAV, this unit boasts strong computing power and low-latency response characteristics, enabling real-time analysis during flight and avoiding the communication burden and time lag associated with transmitting massive amounts of image data back to the ground station. The edge computing unit establishes a stable data channel with the two sensors through a dedicated interface, synchronously acquiring and calibrating the timestamps and spatial coordinates of multimodal images to ensure the accuracy of subsequent fusion processing.

[0039] The heatmap generation module transforms the recognition results output by the edge computing unit into a visual graphical interface, generating two-dimensional or three-dimensional heatmaps containing information such as coking location, area size, and severity level. This heatmap visually displays the coking risk level of each area using a color gradient; for example, blue represents areas without coking, yellow represents light coking, and red represents severe coking, helping operators quickly grasp the overall situation. Simultaneously, this module packages the structured data and sends it to the multimodal control module as input parameters for automatically planning flight paths and selecting decoking modes.

[0040] The edge computing unit is equipped with a hybrid neural network model based on YOLOv5s and Transformer. YOLOv5s (You Only Look Once version 5 small) is a single-stage object detection model with a small number of parameters, fast inference speed, and suitable for embedded platform deployment. Transformer is a deep learning architecture based on a self-attention mechanism, adept at capturing long-range dependencies and contextual semantic relationships between image features. The two are not simply connected in series or parallel, but rather employ a feature-level fusion strategy: after the YOLOv5s backbone network extracts multi-scale local texture and geometric features, the deep feature maps are compressed into serialized tokens via a spatial-channel joint recalibration module. These tokens are then input into a lightweight Transformer encoder (containing 2-4 encoder layers, each with 8 self-attention mechanisms). The Transformer further weights and semantically corrects global criteria such as the spatial distribution pattern of the focal region, thermal radiation continuity, and edge blurring degree, ultimately outputting optimized bounding box coordinates, confidence scores, and focal type classification results.

[0041] The multi-rotor UAV 1 includes: case; A six-axis redundant power system is mounted on the housing; The inertial measurement unit (IMU) is located inside the housing. A lidar obstacle avoidance module is disposed outside the housing; The gimbal suspension mechanism is located below the housing and is fixedly connected to the composite vibration desiccant module.

[0042] The housing protects internal electronic components from the high-temperature environment inside the boiler. This housing can be made of heat-resistant engineering plastics (such as polyetheretherketone, PEEK) or lightweight metal alloys (such as titanium-aluminum alloy), possessing excellent thermal insulation and mechanical strength, capable of withstanding continuous operating temperatures above 50°C and instantaneous temperatures up to 80°C. Furthermore, the housing surface can be coated with a ceramic-based thermal insulation coating to further enhance thermal protection; alternatively, an air gap or phase change material layer can be placed between the housing and internal components to achieve passive temperature control.

[0043] The six-axis redundant power system is mounted on the fuselage, providing the lift and attitude control required for flight. This system comprises six independently driven motor-propeller assemblies, symmetrically mounted on external support structures. If one power unit fails, the remaining five motors can adjust their speed differences to achieve torque balance, maintaining the aircraft in hover and performing a safe return-to-home maneuver. This redundancy significantly enhances the system's flight safety and fault tolerance. Alternatively, an eight-axis (eight-rotor) configuration can be used to achieve higher load capacity and stronger single-point failure tolerance; or a tiltrotor structure can be used to change the thrust direction during specific flight phases to enhance maneuverability.

[0044] The Inertial Measurement Unit (IMU) is housed inside the aircraft's casing and is used to acquire the aircraft's three-axis acceleration, angular velocity, and attitude information in real time. This unit typically integrates a three-axis accelerometer, a three-axis gyroscope, and a magnetometer, with a sampling frequency of at least 200Hz and a data output delay of less than 10ms. Its installation position is close to the aircraft's center of gravity to reduce motion coupling errors. This IMU provides core sensing data to the flight control system, supporting accurate attitude calculation and dynamic response control.

[0045] The lidar obstacle avoidance module is located externally on the casing and is used to acquire 3D point cloud data of the surrounding environment. This module emits frequency-modulated continuous wave (FMCW) or pulsed laser beams, calculates distance by measuring echo time, and has a scanning frequency exceeding 10Hz. The detection range is 0.3m to 30m, with an angular resolution of 0.5°. Its output data is used to construct a local environment map, which, combined with Simultaneous Localization and Mapping (SLAM) algorithms, enables autonomous path planning and dynamic obstacle avoidance. The lidar is deployed at multiple locations on the front and sides of the aircraft, forming multi-view coverage and avoiding blind spots. In scenarios where dense smoke and dust cause severe optical signal attenuation, millimeter-wave radar or ultrasonic sensors can be used as supplementary or alternatives.

[0046] The gimbal suspension mechanism is located below the housing and is fixedly connected to the composite vibration descaling module 2, used to support and stabilize the descaling device. This gimbal features a six-axis active stabilization structure, integrating a servo motor and attitude feedback control system. It can counteract external disturbances during flight, keeping the descaling module perpendicular to the target wall or contacting the coking area at a preset angle. Its maximum load capacity is no less than 15kg, meeting the requirements for mounting heavy-duty vibrators. In practical applications, the gimbal can have a telescopic arm structure, allowing the descaling module to move back and forth within a certain range to adapt to furnace walls of different curvatures; or it can use a universal joint connection, allowing the end effector to freely adjust its angle, improving operational flexibility.

[0047] The multimodal control module includes: Dual-link communication module, connecting to the ground control station; A dynamic obstacle avoidance module is connected to the lidar obstacle avoidance module. The control module is connected to the dual-link communication module, the dynamic obstacle avoidance module, and the six-axis redundant power system, respectively.

[0048] The dual-link communication module includes a 5G communication unit and a WiFi communication unit; the input terminals of the 5G communication unit and the WiFi communication unit are both connected to the ground control station, and the output terminals of the 5G communication unit and the WiFi communication unit are both connected to the control module.

[0049] The dual-link communication module establishes a two-way data transmission channel between the UAV platform and the ground control station, supporting long-distance, high-bandwidth, and low-latency communication requirements. This module employs two independent communication systems operating in parallel, such as a 5G communication unit and a WiFi communication unit, serving as backups for each other. If one link is interrupted due to electromagnetic interference or signal attenuation, the other link automatically takes over the communication task, preventing control disconnection. This design is suitable for operating conditions involving dense metal structures and complex electromagnetic environments within boilers, enhancing the system's communication robustness.

[0050] The dynamic obstacle avoidance module receives point cloud data from the lidar obstacle avoidance module, combines it with SLAM algorithms to generate a real-time 3D model of the local environment, and identifies the location, shape, and movement trend of obstacles. This module incorporates a path replanning strategy, enabling it to instantly calculate detour trajectories and output obstacle avoidance commands when detecting sudden obstacles (such as suspended pipes or incompletely cooled slag deposits). Its processing cycle is less than 100ms, meeting the millisecond-level response requirements of high-speed flight.

[0051] The control module's input is connected to a dual-link communication module to obtain ground remote control commands or preset task sequences. It also receives obstacle avoidance suggestions from the dynamic obstacle avoidance module and, based on the current flight attitude (provided by the IMU inertial measurement unit) and power system status, comprehensively determines the motor output parameters of the six-axis redundant power system. This module employs an embedded real-time operating system (RTOS), supporting concurrent multi-task processing to ensure the strict timing of the flight control law. In fault conditions, the control module can automatically initiate an emergency hovering procedure based on attitude anomalies reported by the IMU, using the remaining five motors to maintain airframe balance and achieve a safe emergency landing or return. Furthermore, without manual intervention, the control module can also drive the UAV to automatically cruise along a preset spiral scanning path to complete a full furnace coverage inspection task.

[0052] The second objective of this invention is to provide a method for removing slag from a thermal power plant boiler using a vibrator mounted on a drone, comprising the following steps: S1. The coking area inside the boiler is identified and located by the AI ​​coking identification system 3, and coking information is generated. S2. The multimodal control module receives the coking information and controls the multi-rotor UAV 1 to fly to the target coking area according to the coking information; S3. The multimodal control module controls the composite vibration decoking module 2 to start, and performs vibration impact decoking operation on the target coking area; S4. After the descorching operation is completed, the target coking area is re-identified by the AI ​​coking recognition system 3 to verify the descorching effect.

[0053] This method utilizes an AI vision system to accurately identify and locate coking deposits. Combined with a multimodal control module, it guides a drone to autonomously reach the target area. Then, based on the characteristics of the coking deposits, it intelligently activates the corresponding vibration / impact mode for removal. A secondary identification process forms a quality closed-loop verification. This method not only solves the technical problems of high safety risks, numerous blind spots, and low efficiency inherent in traditional manual coking removal, but also significantly improves the reliability and consistency of operations through intelligent closed-loop control. It provides a practical and feasible technical path for unattended boiler maintenance in thermal power plants.

[0054] The method by which the multimodal control module controls the start-up of the composite vibration desiccant module 2 includes: Based on the recognition results of the target coking area by the AI ​​coking recognition system 3, the coking type is determined to be either loose coking or hard coking. If the coking is loose, the composite vibration decoking module 2 is controlled to operate in a preset vibration mode with a frequency of 50Hz to 200Hz. If the coking is hard, the composite vibration decoking module 2 is controlled to operate in a preset impact mode, and the impact force of the impact mode is 5N~80N.

[0055] Loose coke typically appears as granular or flaky deposits with low adhesion strength. It is mainly formed by the deposition of unburned carbon particles or light ash. In visible light images, it exhibits a loose and porous structure, while in infrared images, it shows a small temperature difference (generally <25℃). It has weak mechanical strength and is easily detached due to high-frequency vibration. Hard coke, on the other hand, is a dense, molten coke mass, often found near high-temperature combustion zones. It has high adhesion and hardness, and its appearance is glassy or layered. In infrared thermography, it shows a significant temperature difference (≥30℃), indicating that it is still in an active growth state. It is difficult to remove and requires directional impact energy for effective breakage and peeling.

[0056] The AI ​​coking identification module of this system directly determines the working mode of the composite vibration coking removal module 2 through classification output. After receiving the image recognition results, the multi-modal control module schedules the corresponding actuator according to the coking type: if it is determined to be loose coking, the high-frequency electromagnetic vibration mode is activated; if it is determined to be hard coking, the controllable impact mode is switched. Both modes use a vibration parameter adaptive control unit to adjust key parameters (such as frequency or impact force) in real time to ensure that the working intensity matches the actual coking characteristics. This invention effectively solves the problems of "overly agitated soft coking" or "incomplete removal of hard coking" caused by fixed parameters in traditional coking removal methods by introducing machine vision-based intelligent recognition logic and combining it with a switchable vibration / impact dual-mode execution strategy. While ensuring the safety of the boiler body, this system significantly improves the targeting, energy efficiency ratio and success rate of coking removal operations, thereby achieving the goal of efficient, energy-saving and safe intelligent coking removal.

[0057] The method also includes: The multimodal control module controls the multi-rotor UAV 1 to perform a comprehensive inspection of the boiler's inner wall based on a preset spiral scanning path. During the inspection process, when the AI ​​coking recognition system 3 detects coking, it automatically performs a coking removal operation. After the decoking operation is completed, continue to inspect along the preset path until all inspection points are completed or a stop command is received.

[0058] The preset spiral scanning path is based on the boiler's central axis, distributed in an equidistant spiral from bottom to top or top to bottom. The spacing between adjacent layers is set according to the detection accuracy of the lidar obstacle avoidance module and the effective recognition range of the visible light camera, typically ranging from 0.3m to 1.0m, ensuring that no area is missed in image acquisition. Comprehensive inspection refers to the process by which the UAV continuously acquires visual and thermal imaging data of the entire inner wall surface of the boiler according to the predetermined path.

[0059] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, characterized in that, include: Multi-rotor unmanned aerial vehicle (1); A composite vibration defocusing module (2) is disposed below the multi-rotor UAV (1); A multimodal control module is connected to the multi-rotor UAV (1) and the composite vibration defocusing module (2); The AI ​​coking recognition system (3) is installed on the multi-rotor UAV (1) and connected to the multimodal control module.

2. The coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 1, is characterized in that... The composite vibration descorching module (2) includes an electromagnetic vibrator, a mechanical impact head, and a vibration parameter adaptive control unit. The vibration parameter adaptive control unit is connected to the electromagnetic vibrator and the mechanical impact head, respectively; and the vibration parameter adaptive control unit is connected to the multimodal control module.

3. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 1, is characterized in that... The AI ​​coma recognition system (3) includes a visible light camera, an infrared thermal imager, an edge computing unit, and a heat map generation module; the visible light camera and the infrared thermal imager are respectively connected to the image input terminal of the edge computing unit; the data output terminal of the edge computing unit is connected to the heat map generation module; and the heat map generation module is connected to the multimodal control module.

4. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 3, is characterized in that... The edge computing unit is equipped with a hybrid neural network model based on YOLOv5s and Transformer.

5. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 1, is characterized in that... The multi-rotor unmanned aerial vehicle (1) includes: case; A six-axis redundant power system is mounted on the housing; The inertial measurement unit (IMU) is located inside the housing. A lidar obstacle avoidance module is disposed outside the housing; The gimbal suspension mechanism is located below the housing and is fixedly connected to the composite vibration descorching module (2).

6. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 5, is characterized in that... The multimodal control module includes: Dual-link communication module, connecting to the ground control station; A dynamic obstacle avoidance module is connected to the lidar obstacle avoidance module. The control module is connected to the dual-link communication module, the dynamic obstacle avoidance module, and the six-axis redundant power system, respectively.

7. A coke removal system for a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 6, is characterized in that... The dual-link communication module includes a 5G communication unit and a WiFi communication unit; the input terminals of the 5G communication unit and the WiFi communication unit are both connected to the ground control station, and the output terminals of the 5G communication unit and the WiFi communication unit are both connected to the control module.

8. A method for removing slag from a thermal power plant boiler using a vibrator mounted on a drone, characterized in that, The system based on any one of claims 1 to 7 includes the following steps: The coking area inside the boiler is identified and located by the AI ​​coking identification system (3), and coking information is generated. The multimodal control module receives the coking information and controls the multi-rotor UAV (1) to fly to the target coking area according to the coking information; The multimodal control module controls the composite vibration decoking module (2) to start, and performs vibration impact decoking operation on the target coking area; After the descorching operation is completed, the target descorched area is re-identified by the AI ​​descorching recognition system (3) to verify the descorching effect.

9. A method for removing coke from a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 8, is characterized in that... The method by which the multimodal control module controls the start of the composite vibration decoking module (2) includes: Based on the recognition results of the target coking area by the AI ​​coking recognition system (3), the coking type is determined to be either loose coking or hard coking; If the coking is loose, the composite vibration decoking module (2) is controlled to work in a preset vibration mode with a frequency of 50Hz~200Hz. If the coking is hard, the composite vibration decoking module (2) is controlled to work in a preset impact mode, and the impact force of the impact mode is 5N~80N.

10. A method for removing coke from a thermal power plant boiler using a vibrator mounted on a drone, as described in claim 9, is characterized in that... Also includes: The multimodal control module controls the multi-rotor UAV (1) to conduct a comprehensive inspection of the boiler inner wall based on a preset spiral scanning path; During the inspection process, when the AI ​​coking identification system (3) detects coking, it automatically performs a coking removal operation; After the decoking operation is completed, continue to inspect along the preset path until all inspection points are completed or a stop command is received.

Images