A boiler heating surface primary-secondary detection method and system

Abstract

This invention provides a method and system for mother-daughter type inspection of boiler heating surfaces. The method includes: acquiring a three-dimensional structural model of the boiler heating surface and preset inspection task instructions; dividing the three-dimensional structural model into a gridded tube bank area to generate an initial task allocation set for the corresponding gridded inspection area; controlling a main unit carrying at least one set of daughter units to enter the boiler furnace, collecting global environmental data of the tube bank in real time, calibrating the three-dimensional structural model in real time, and simultaneously completing the global positioning and pose closed-loop control of the main unit in the furnace; dynamically updating the initial task allocation set based on the calibration results, generating a main unit movement path sequence and corresponding target inspection task packages for each daughter unit; and synchronously controlling each daughter unit to perform multi-dimensional inspection of the heating surface tube wall within the corresponding gridded inspection area. This achieves accurate allocation and efficient execution of inspection tasks. Combined with a full-coverage verification and supplementary test closed-loop mechanism, it ensures full-area coverage of the heating surface inspection.

Description

Technical Field

[0001] This invention relates to the field of boiler operation and maintenance testing technology, specifically to a mother-daughter type testing method and system for boiler heating surfaces. Background Technology

[0002] Power plant boilers are the core power equipment in thermal power generation and combined heat and power (CHP) systems. The health of the heating surface piping, consisting of the water-cooled walls, superheater, reheater, and economizer, directly determines the boiler's service life, safety, reliability, and the unit's operating economy. With the normalization of high-parameter operation, deep peak shaving, and coal blending strategies in thermal power units, boiler "four-pipe" leakage accidents are frequent, highlighting the increasing demand for efficient, accurate, and comprehensive inspection of the heating surface piping.

[0003] Currently, boiler furnace heating surface inspection and maintenance still primarily rely on manual entry. The boiler furnace interior is a confined space with multiple layers of closely spaced tubes. Traditional manual inspection suffers from inherent shortcomings: numerous blind spots, long inspection cycles, low data accuracy, high risks associated with working at heights, and high labor costs. This makes it difficult to meet the core requirements of modern power systems for safe management, economical operation, and refined maintenance of boiler equipment. Especially within the complex three-dimensional space formed by closely spaced tubes, traditional inspection equipment cannot simultaneously handle multiple operation modes such as single-tube climbing, tube-to-tube transitions, tube panel crossings, obstacle traversal, and large-area wall scanning. The inspection coverage is severely insufficient, failing to achieve convenient full-path inspection of the heating surface piping.

[0004] Existing solutions for inspecting heated surfaces inside furnaces mostly adopt a structure that integrates a single mobile carrier with an inspection unit. Due to the limitations of the complex tube array structure inside the furnace, they cannot simultaneously achieve large-scale, rapid, and convenient inspection as well as localized, high-precision, and detailed inspection. They are also difficult to adapt to the full-scenario inspection needs of different furnace types and tube array structures, and cannot solve the requirements for full-path coverage and multi-mode collaborative operation within densely packed tube arrays. This has become a key problem that urgently needs to be solved in this field. Summary of the Invention

[0005] Based on this, and in response to the shortcomings of existing technologies, the present invention provides a method and system for detecting the mother-daughter type of boiler heating surface, in order to solve the problem that existing technologies cannot simultaneously achieve large-scale and local high-precision fine detection, and are difficult to adapt to the full-scenario detection needs of different boiler types and different tube bank structures.

[0006] To solve the above-mentioned technical problems, the first aspect of the present invention proposes: A method for detecting the mother-daughter type of boiler heating surface, the method comprising: Obtain the three-dimensional structural model of the heating surface inside the boiler furnace and the preset detection task instructions, divide the three-dimensional structural model into a pipe row area grid, and generate an initial task allocation set for the corresponding grid detection area; The machine controls the mother machine to carry at least one set of daughter machines into the boiler furnace. The global sensing device on the mother machine collects global environmental data of the tube bank in the furnace in real time, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. Based on the calibrated 3D structural model and real-time positioning results, the initial task allocation set is dynamically updated to generate the mother machine's movement path sequence and the corresponding target detection task package for each sub-machine. According to the movement path sequence, the master machine is controlled to move to the corresponding task point, and the deployment and docking of the sub-machines are completed in sequence. Simultaneously, each sub-machine is controlled to perform multi-dimensional inspection of the heated surface pipe wall in the corresponding gridded inspection area, and the inspection data transmitted back by the sub-machines is received and the data is bound to the spatial coordinates. Based on the detection data returned by all slave units, complete the full coverage verification of the boiler heating surface detection area, generate supplementary detection tasks for uncovered areas, and repeatedly execute the operations of moving the main unit, deploying slave units for detection and retrieval until all detection tasks are completed.

[0007] The beneficial effects of this invention are as follows: By dividing the boiler heating surface tube bank into grids and using a mother-daughter collaborative operation architecture, this invention effectively solves the core pain point of traditional detection schemes that cannot simultaneously achieve large-scale convenient and local high-precision fine-grained detection. It can adapt to the detection needs of different boiler types and tube bank structures in all scenarios. Through global environmental perception of the mother machine and real-time calibration of the three-dimensional structural model, the system's adaptability to the complex tube bank environment inside the furnace and its global positioning accuracy are improved. Through dynamic task updates and mother-daughter collaborative operation, the accurate allocation and efficient execution of detection tasks are achieved. Combined with a full-coverage verification and supplementary test closed-loop mechanism, the full-area coverage of heating surface detection is ensured, while significantly improving detection efficiency and data accuracy, and reducing the safety risks of manual furnace operations.

[0008] Furthermore, the steps of acquiring the three-dimensional structural model of the boiler's internal heating surface and the preset detection task instructions, dividing the three-dimensional structural model into a pipe-row area grid, and generating an initial task allocation set corresponding to the gridded detection area include: Extract the tube bank structural parameters of the water-cooled wall, superheater, reheater, and economizer from the three-dimensional structural model of the boiler heating surface, including tube diameter, tube spacing, number of rows, bending angle, and spatial layout coordinates; Based on the detection accuracy requirements, coverage and operation cycle in the preset detection task instructions, the tube array structure is divided into grid areas, taking a single heated surface tube as the smallest unit, to generate multiple non-overlapping grid detection areas. For each gridded detection area, match the corresponding submachine detection operation parameters, including detection range, detection type, operation path and data acquisition requirements. Combine the movable range of the master machine and the operation radius of the submachine to generate the initial task allocation set containing multiple sets of submachine-detection area pairing relationships.

[0009] Furthermore, the control machine, equipped with at least one set of slave units, enters the boiler furnace. Through a global sensing device mounted on the machine, it collects real-time global environmental data of the boiler's pipe rows, performs real-time calibration of the three-dimensional structural model, and simultaneously completes the steps of global positioning and closed-loop attitude control of the machine within the furnace, including: The machine is controlled to move along a preset initial path inside the furnace. Through the lidar and binocular vision device mounted on the machine, three-dimensional point cloud data and image data of the tube bank structure inside the furnace are collected in real time to generate a real-time point cloud map of the furnace environment. The real-time environmental point cloud map is matched and the pre-stored three-dimensional structural model of the boiler heating surface is registered with feature points to identify the deviation between the model and the actual structure, and the pipe row coordinates and spatial structure of the three-dimensional structural model are calibrated in real time. Based on the registered coordinate system, and combined with the fusion data from the inertial measurement unit, wheeled odometer, and visual odometer on the mother machine, the real-time eight-degree-of-freedom pose parameters of the mother machine are calculated. Based on the deviation between the pose parameters and the preset path, the pose closed-loop control of the mother machine is completed.

[0010] Furthermore, the step of dynamically updating the initial task allocation set based on the calibrated 3D structural model and real-time positioning results to generate the mother machine's movement path sequence and the corresponding target detection task packages for each sub-machine includes: Based on the calibrated 3D structural model, the actual obstacles, irregular tube structures and impassable areas inside the furnace are identified, and the boundaries of the gridded detection areas in the initial task allocation set are adjusted and merged / split. Based on the real-time positioning results of the mother machine, the passable range, the rated operating capacity of each submachine, and the remaining power, the submachine tasks are redistributed in the adjusted gridded detection area, and the pairing relationship between submachine and detection area is updated. Based on the updated pairing relationship, with the optimization objectives of minimizing the total movement path of the mother machine and minimizing the number of deployment and retrieval times of the daughter machines, the mother machine movement path sequence is planned and generated. At the same time, for each daughter machine, a target detection task package is generated, which includes a single-pipe climbing path, inter-pipe transition rules, and detection parameter configuration for the corresponding gridded detection area.

[0011] Furthermore, the steps of controlling the mother machine to move to the corresponding task location according to the movement path sequence, sequentially deploying and docking the submachines, synchronously controlling each submachine to perform multi-dimensional inspection of the heated surface pipe wall within the corresponding gridded inspection area, and receiving the inspection data transmitted back by the submachines and binding the data with spatial coordinates include: Control the mother machine to move to the target task point in the movement path sequence, complete the mother machine attitude lock and docking platform leveling, drive the delivery and recovery mechanism to perform the daughter machine delivery operation, and attach the daughter machine to the target heated surface tube wall; The corresponding target inspection task package is sent to the submachine, and the submachine is controlled to perform single pipe climbing, pipe transition and obstacle avoidance operations along the operation path in the task package within the gridded inspection area. Simultaneously, the ultrasonic thickness measuring device, eddy current detection device and visual inspection device on the submachine are controlled to perform multi-dimensional inspection operations of pipe wall thickness, internal defects and surface cracks. The system receives the detection data and real-time pose data transmitted back from the slave unit in real time, and binds the detection data with the spatial coordinates and pipe segment number of the corresponding heated surface tube to generate a detection dataset with spatial positioning tags. After the slave unit completes its work in the corresponding gridded detection area, it is controlled to return to the docking point, and the mother machine's deployment and retrieval mechanism is driven to complete the docking and retrieval of the slave unit.

[0012] Furthermore, the steps of completing the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generating supplementary detection tasks for uncovered areas, and repeatedly executing the mother unit movement, slave unit deployment and retrieval operations until all detection tasks are completed include: Based on the detection dataset with spatial positioning labels, all pipe section areas that have been inspected are marked in the calibrated 3D structural model of the boiler heating surface, and a detection coverage heat map is generated. The detection coverage heat map is compared with the target coverage area in the preset detection task instruction to identify the uncovered pipe section area or the invalid detection data, and mark it as the area to be retested; For the area to be retested, a retesting task package is generated, the master machine's movement path sequence and the slave machine's operation task are updated, and the master machine and slave machine are controlled to repeatedly execute the corresponding retesting operation until the detection coverage meets the preset task requirements and all detection tasks are completed.

[0013] Furthermore, the method also includes: Defect features are extracted from the detection dataset bound to spatial coordinates to identify abnormal data such as pipe wall thinning, internal defects, and surface cracks. The spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are then extracted. In the calibrated 3D structural model of the boiler heating surface, the defect points are visually marked, and a boiler heating surface inspection report containing the defect location, defect type, and defect level is generated.

[0014] The second aspect of the present invention proposes: A boiler heating surface mother-daughter type detection system, characterized in that the system comprises: The planning module is used to obtain the three-dimensional structural model of the heating surface inside the boiler and the preset detection task instructions, divide the three-dimensional structural model into a pipe row area grid, and generate an initial task allocation set for the corresponding grid detection area. The positioning module is used to control the mother machine to carry at least one set of daughter machines into the boiler furnace. It collects global environmental data of the tube bank in the furnace in real time through the global sensing device on the mother machine, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. The update module is used to dynamically update the initial task allocation set based on the calibrated 3D structural model and real-time positioning results, and generate the mother machine's movement path sequence and the corresponding target detection task package for each submachine. The operation module is used to control the mother machine to move to the corresponding task point according to the movement path sequence, and to complete the deployment and docking and retrieval of the sub-machines in sequence. It also controls each sub-machine to perform multi-dimensional inspection of the heated surface pipe wall in the corresponding gridded inspection area, and receives the inspection data transmitted back by the sub-machines and completes the binding of the data with spatial coordinates. The verification module is used to complete the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generate supplementary test tasks for uncovered areas, and repeatedly execute the operation of moving the main unit, deploying slave units for detection and retrieval until all detection tasks are completed.

[0015] Furthermore, the planning module is specifically used for: Extract the tube bank structural parameters of the water-cooled wall, superheater, reheater, and economizer from the three-dimensional structural model of the boiler heating surface, including tube diameter, tube spacing, number of rows, bending angle, and spatial layout coordinates; Based on the detection accuracy requirements, coverage and operation cycle in the preset detection task instructions, the tube array structure is divided into grid areas, taking a single heated surface tube as the smallest unit, to generate multiple non-overlapping grid detection areas. For each gridded detection area, match the corresponding submachine detection operation parameters, including detection range, detection type, operation path and data acquisition requirements. Combine the movable range of the master machine and the operation radius of the submachine to generate the initial task allocation set containing multiple sets of submachine-detection area pairing relationships.

[0016] Furthermore, the positioning module is specifically used for: The machine is controlled to move along a preset initial path inside the furnace. Through the lidar and binocular vision device mounted on the machine, three-dimensional point cloud data and image data of the tube bank structure inside the furnace are collected in real time to generate a real-time point cloud map of the furnace environment. The real-time environmental point cloud map is matched and the pre-stored three-dimensional structural model of the boiler heating surface is registered with feature points to identify the deviation between the model and the actual structure, and the pipe row coordinates and spatial structure of the three-dimensional structural model are calibrated in real time. Based on the registered coordinate system, and combined with the fusion data from the inertial measurement unit, wheeled odometer, and visual odometer on the mother machine, the real-time eight-degree-of-freedom pose parameters of the mother machine are calculated. Based on the deviation between the pose parameters and the preset path, the pose closed-loop control of the mother machine is completed.

[0017] Furthermore, the update module is specifically used for: Based on the calibrated 3D structural model, the actual obstacles, irregular tube structures and impassable areas inside the furnace are identified, and the boundaries of the gridded detection areas in the initial task allocation set are adjusted and merged / split. Based on the real-time positioning results of the mother machine, the passable range, the rated operating capacity of each submachine, and the remaining power, the submachine tasks are redistributed in the adjusted gridded detection area, and the pairing relationship between submachine and detection area is updated. Based on the updated pairing relationship, with the optimization objectives of minimizing the total movement path of the mother machine and minimizing the number of deployment and retrieval times of the daughter machines, the mother machine movement path sequence is planned and generated. At the same time, for each daughter machine, a target detection task package is generated, which includes a single-pipe climbing path, inter-pipe transition rules, and detection parameter configuration for the corresponding gridded detection area.

[0018] Furthermore, the operation module is specifically used for: Control the mother machine to move to the target task point in the movement path sequence, complete the mother machine attitude lock and docking platform leveling, drive the delivery and recovery mechanism to perform the daughter machine delivery operation, and attach the daughter machine to the target heated surface tube wall; The corresponding target inspection task package is sent to the submachine, and the submachine is controlled to perform single pipe climbing, pipe transition and obstacle avoidance operations along the operation path in the task package within the gridded inspection area. Simultaneously, the ultrasonic thickness measuring device, eddy current detection device and visual inspection device on the submachine are controlled to perform multi-dimensional inspection operations of pipe wall thickness, internal defects and surface cracks. The system receives the detection data and real-time pose data transmitted back from the slave unit in real time, and binds the detection data with the spatial coordinates and pipe segment number of the corresponding heated surface tube to generate a detection dataset with spatial positioning tags. After the slave unit completes its work in the corresponding gridded detection area, it is controlled to return to the docking point, and the mother machine's deployment and retrieval mechanism is driven to complete the docking and retrieval of the slave unit.

[0019] Furthermore, the verification module is specifically used for: Based on the detection dataset with spatial positioning labels, all pipe section areas that have been inspected are marked in the calibrated 3D structural model of the boiler heating surface, and a detection coverage heat map is generated. The detection coverage heat map is compared with the target coverage area in the preset detection task instruction to identify the uncovered pipe section area or the invalid detection data, and mark it as the area to be retested; For the area to be retested, a retesting task package is generated, the master machine's movement path sequence and the slave machine's operation task are updated, and the master machine and slave machine are controlled to repeatedly execute the corresponding retesting operation until the detection coverage meets the preset task requirements and all detection tasks are completed.

[0020] Furthermore, the verification module is specifically used for: Defect features are extracted from the detection dataset bound to spatial coordinates to identify abnormal data such as pipe wall thinning, internal defects, and surface cracks. The spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are then extracted. In the calibrated 3D structural model of the boiler heating surface, the defect points are visually marked, and a boiler heating surface inspection report containing the defect location, defect type, and defect level is generated.

[0021] The third aspect of the present invention proposes: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the boiler heating surface mother-daughter type detection method as described above.

[0022] The fourth aspect of the present invention proposes: A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the boiler heating surface mother-daughter type detection method as described above. Attached Figure Description

[0023] Figure 1 A flowchart of the boiler heating surface mother-daughter type detection method provided in the first embodiment of the present invention; Figure 2 This is a structural block diagram of the boiler heating surface mother-daughter type detection system provided in the third embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0025] Please see Figure 1The first embodiment of this invention provides a mother-daughter type detection method for boiler heating surfaces. This method effectively solves the core pain point of traditional detection schemes, which cannot simultaneously achieve large-scale convenient and localized high-precision detailed detection, through gridded division of boiler heating surface tube banks and a mother-daughter collaborative operation architecture. It can adapt to the full-scenario detection needs of different boiler types and tube bank structures. Through global environmental perception of the mother machine and real-time calibration of the three-dimensional structural model, the system's adaptability to complex tube bank environments within the furnace and its global positioning accuracy are improved. Through dynamic task updates and mother-daughter collaborative operation, accurate allocation and efficient execution of detection tasks are achieved. Combined with a full-coverage verification and supplementary testing closed-loop mechanism, full-area coverage of heating surface detection is ensured, while significantly improving detection efficiency and data accuracy, and reducing the safety risks of manual furnace operations.

[0026] Specifically, the first embodiment of the present invention provides: A method for detecting the mother-daughter type of boiler heating surface, wherein the method includes: Step S10: Obtain the three-dimensional structural model of the heating surface inside the boiler and the preset detection task instructions; divide the three-dimensional structural model into a pipe row area grid; and generate an initial task allocation set for the corresponding gridded detection area. It should be noted that the boiler heating surface comprises four core components: water-cooled walls, superheaters, reheaters, and economizers. The tube bank structure is complex and spans a large space, with significant differences in testing requirements in different areas. By acquiring a three-dimensional structural model of the boiler heating surface and pre-set testing task instructions, it is possible to understand the basic spatial structure of the tube bank inside the furnace and the core objectives and boundary conditions of this operation in advance. Furthermore, by dividing the three-dimensional structural model into tube bank areas into grids, the entire heating surface, which originally spans a huge area and has a complex structure, can be broken down into multiple independently executable and clearly defined testing units. This solves the problem of not being able to accurately decompose and allocate large-scale testing tasks. The resulting initial task allocation set provides an initial task framework for the collaborative operation of the mother and daughter boilers, realizing the pre-emptive and refined decomposition of testing tasks.

[0027] Step S20: Control the mother machine to carry at least one set of daughter machines into the boiler furnace. The global sensing device on the mother machine collects global environmental data of the tube bank in the furnace in real time, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. It should be noted that the pre-stored 3D structural model is a theoretical model from the design phase, and there are inherent differences between it and the actual tube deformation, new obstacles on site, and structural deviations after the boiler is in operation. At the same time, the furnace is a confined space without satellite navigation, and traditional positioning methods are not applicable. By controlling the mother machine to carry the daughter machine into the furnace and collecting global environmental data, the actual tube structure and spatial environment inside the furnace can be restored. This allows for real-time calibration of the 3D structural model, eliminating the deviation between design and reality, providing a precise spatial reference for subsequent path planning and task allocation, and completing the global positioning and posture closed-loop control of the mother machine. This solves the problem of precise positioning of the mobile carrier in a confined space, ensuring that the mother machine can move stably and accurately in the complex tube environment, and providing a stable operating platform for the subsequent deployment and retrieval of the daughter machine.

[0028] Step S30: Based on the calibrated 3D structural model and real-time positioning results, dynamically update the initial task allocation set to generate the mother machine's movement path sequence and the corresponding target detection task package for each submachine. It should be noted that the initial task allocation set is generated based on a theoretical model and may not match the actual field environment after calibration. By combining the calibrated 3D structural model with the real-time positioning results of the machine, the initial task allocation set is dynamically updated. This can avoid inaccessible areas on site, adapt to the actual structure of the pipe array, and ensure that the task allocation is fully adapted to the field environment. The generation of the machine's movement path sequence and the slave machine's target detection task package can not only reduce the number of invalid movements of the machine and the number of slave machine deployments and retrievals through path optimization, thereby improving the overall operation efficiency, but also break down the macroscopic detection task into specific operation instructions that the slave machine can directly execute, achieving a precise conversion from macroscopic tasks to microscopic execution actions.

[0029] Step S40: Control the mother machine to move to the corresponding task point according to the moving path sequence, and sequentially complete the deployment and docking of the sub-machines. Simultaneously control each sub-machine to perform multi-dimensional detection of the heated surface pipe wall in the corresponding gridded detection area, receive the detection data returned by the sub-machines and complete the binding of data with spatial coordinates. It should be noted that the mother unit undertakes the core functions of large-scale movement, deployment and retrieval of slave units, and data transfer, solving the need for large-scale inspection. The slave units, on the other hand, undertake the functions of single-pipe climbing and local fine-grained inspection, solving the need for high-precision inspection. Together, they break through the core limitations of traditional inspection solutions. By controlling the mother unit to move precisely to the task location to complete the deployment and retrieval of slave units, it can ensure that the slave units accurately enter the target inspection area and control the slave units to perform multi-dimensional inspection operations. It can comprehensively collect core health data such as pipe wall thickness, internal defects, and surface cracks. Furthermore, binding the inspection data with spatial coordinates solves the pain points of traditional inspection data being unable to accurately locate and trace the defect location, providing accurate spatial positioning basis for subsequent defect analysis and operation and maintenance decisions.

[0030] Step S50: Based on the detection data returned by all slave units, complete the full coverage verification of the boiler heating surface detection area, generate supplementary detection tasks for uncovered areas, and repeatedly execute the mother unit movement, slave unit deployment and detection and retrieval operations until all detection tasks are completed.

[0031] It should be noted that the core requirement for boiler heating surface inspection is full coverage without blind spots to avoid leaks in the four tubes due to missed inspections. However, in actual operation, there may be situations such as deviation of the sub-machine path, invalid inspection data, and partial coverage. Full coverage verification can comprehensively verify the completed inspection work, accurately identify the uncovered or invalid areas, generate supplementary inspection tasks for these areas, and repeat the corresponding operation process to form a complete closed loop of "execution-verification-supplementary inspection". This fundamentally ensures the full coverage requirement of the inspection work and solves the problems of many blind spots and high missed inspection rate of traditional inspection methods.

[0032] Second Embodiment Furthermore, the steps of acquiring the three-dimensional structural model of the boiler's internal heating surface and the preset detection task instructions, dividing the three-dimensional structural model into a pipe-row area mesh, and generating an initial task allocation set corresponding to the meshed detection area include: Extract the tube bank structural parameters of the water-cooled wall, superheater, reheater, and economizer from the three-dimensional structural model of the boiler heating surface, including tube diameter, tube spacing, number of rows, bending angle, and spatial layout coordinates; Based on the detection accuracy requirements, coverage and operation cycle in the preset detection task instructions, the tube array structure is divided into grid areas, taking a single heated surface tube as the smallest unit, to generate multiple non-overlapping grid detection areas. For each gridded detection area, match the corresponding submachine detection operation parameters, including detection range, detection type, operation path and data acquisition requirements. Combine the movable range of the master machine and the operation radius of the submachine to generate the initial task allocation set containing multiple sets of submachine-detection area pairing relationships.

[0033] It should be noted that this embodiment clarifies the specific implementation path for grid-based partitioning and the generation of the initial task allocation set. The three sub-steps form a complete logic of "structural parameter extraction - grid-based region partitioning - task matching and allocation," which is a necessary support for the implementation of the pre-planning stage. The first sub-step is the basic premise for grid-based partitioning. By extracting core structural parameters such as pipe diameter, pipe spacing, number of rows, bending angle, and spatial layout coordinates of the four major heating surfaces in the three-dimensional structural model, the actual spatial layout pattern of the pipe rows can be fully grasped. This provides an accurate structural basis for subsequent region partitioning and avoids problems such as unreasonable region partitioning and mismatch with the actual pipe rows due to missing structural parameters. The second sub-step clarifies the core principles and minimum unit of grid division. Using a single heated surface pipe as the minimum unit for area division is because the core object of heated surface inspection is a single pipe. Based on this, it can be ensured that the boundary of each grid area is fully adapted to the pipe row structure, and there will be no problems such as crossing pipe rows or blurring area boundaries. At the same time, the division is combined with the accuracy requirements, coverage and operation cycle of the inspection task. The grid size can be flexibly adjusted according to different task requirements to achieve precise adaptation between area division and task requirements. The generation of non-overlapping grid areas avoids the problems of duplicate operation and operation boundary conflict in subsequent task allocation. The third sub-step transforms the process from area division to task allocation. By matching corresponding sub-machine detection operation parameters to each gridded detection area, the detection requirements of each area have clear execution standards, avoiding arbitrariness in operation. At the same time, by combining the mobile range of the main machine with the operating radius of the sub-machine to pair the sub-machine with the detection area, it can be ensured that the paired detection area is within the sub-machine's operating capability and that the main machine can reach the corresponding deployment point. This avoids the risk of tasks failing to be implemented from the initial planning stage. The final generated initial task allocation set provides a clear and executable initial task framework for the entire detection operation.

[0034] Furthermore, the control machine, equipped with at least one set of slave units, enters the boiler furnace. Through a global sensing device mounted on the machine, it collects real-time global environmental data of the boiler's pipe rows, performs real-time calibration of the three-dimensional structural model, and simultaneously completes the steps of global positioning and closed-loop attitude control of the machine within the furnace, including: The machine is controlled to move along a preset initial path inside the furnace. Through the lidar and binocular vision device mounted on the machine, three-dimensional point cloud data and image data of the tube bank structure inside the furnace are collected in real time to generate a real-time point cloud map of the furnace environment. The real-time environmental point cloud map is matched and the pre-stored three-dimensional structural model of the boiler heating surface is registered with feature points to identify the deviation between the model and the actual structure, and the pipe row coordinates and spatial structure of the three-dimensional structural model are calibrated in real time. Based on the registered coordinate system, and combined with the fusion data from the inertial measurement unit, wheeled odometer, and visual odometer on the mother machine, the real-time eight-degree-of-freedom pose parameters of the mother machine are calculated. Based on the deviation between the pose parameters and the preset path, the pose closed-loop control of the mother machine is completed.

[0035] It should be noted that by controlling the machine to move along a preset initial path, and using a LiDAR and binocular vision device to collect 3D point cloud data and image data in real time, the advantages of both sensors can be combined to generate a high-precision, high-fidelity real-time point cloud map of the furnace environment. This map completely recreates the actual structure of the tube banks, on-site obstacles, and spatial environment inside the furnace, providing a real-world data benchmark for subsequent model calibration and positioning. Moving along the preset initial path ensures that the machine can cover the core detection area inside the furnace and collect complete environmental data. Furthermore, by matching feature points and registering coordinates between the real-time environmental point cloud map and the pre-stored 3D structural model, the coordinate system of the real-world data collected on-site can be unified with the theoretical model. At the same time, the structural deviations between the theoretical model and the actual tube banks can be identified, and the tube bank coordinates and spatial structure of the 3D structural model can be calibrated in real time. This ensures that the calibrated 3D model perfectly matches the real environment inside the furnace, providing an accurate and reliable spatial benchmark for subsequent path planning and task allocation, avoiding the problem of paths and tasks planned based on theoretical models not matching the actual on-site conditions.

[0036] In the confined environment of the furnace without GNSS signals, single-sensor positioning is prone to cumulative errors and interference from the metallic environment. By fusing multiple sensors, including inertial measurement units, wheeled odometers, and visual odometers, the shortcomings of each sensor can be complemented, cumulative errors can be eliminated, and positioning accuracy can be significantly improved. The real-time eight-DOF pose parameters of the mother machine are calculated, including not only the three-dimensional spatial coordinates and three-axis attitude angles of the mother machine, but also core parameters such as the displacement of the walking mechanism and the attitude of the docking platform, which can fully characterize the real-time operating status of the mother machine. The pose closed-loop control based on the deviation between the pose parameters and the preset path can correct the movement deviation of the mother machine in real time, ensuring that the mother machine moves accurately along the planned path and accurately reaches the target point for the deployment of the submachine, providing a stable and accurate platform foundation for the subsequent deployment and recovery of the submachine.

[0037] Furthermore, the step of dynamically updating the initial task allocation set based on the calibrated 3D structural model and real-time positioning results to generate the mother machine's movement path sequence and the corresponding target detection task packages for each sub-machine includes: Based on the calibrated 3D structural model, the actual obstacles, irregular tube structures and impassable areas inside the furnace are identified, and the boundaries of the gridded detection areas in the initial task allocation set are adjusted and merged / split. Based on the real-time positioning results of the mother machine, the passable range, the rated operating capacity of each submachine, and the remaining power, the submachine tasks are redistributed in the adjusted gridded detection area, and the pairing relationship between submachine and detection area is updated. Based on the updated pairing relationship, with the optimization objectives of minimizing the total movement path of the mother machine and minimizing the number of deployment and retrieval times of the daughter machines, the mother machine movement path sequence is planned and generated. At the same time, for each daughter machine, a target detection task package is generated, which includes a single-pipe climbing path, inter-pipe transition rules, and detection parameter configuration for the corresponding gridded detection area.

[0038] It should be noted that the calibrated 3D model has reproduced the actual obstacles, irregular structures, and impassable areas inside the furnace. However, the initial gridded detection area is based on a theoretical model and may have issues such as areas crossing impassable zones or containing irregularly shaped pipes, preventing the submachines from operating. By adjusting the boundaries of the gridded detection area and merging or splitting the areas, each gridded detection area can be adapted to the actual pipe structure on site, avoiding impassable areas and ensuring that each area is within the submachine's operational range, providing a reasonable regional basis for subsequent task allocation. To achieve better optimal task allocation, the real-time positioning results of the main unit, the accessible range, and the rated operating capacity and remaining power of each submachine are combined. This avoids problems in the initial allocation such as submachines not being able to cover areas, the main unit not being able to reach the deployment point, and submachines having insufficient power to complete the operation. By re-allocating tasks and updating the pairing relationship between submachines and detection areas, the operation tasks of each submachine are fully matched with its own capabilities, maximizing the operation capacity of each submachine and improving overall operation efficiency. Furthermore, by optimizing the mother machine's movement path sequence to minimize the total movement path of the mother machine and the number of deployment and retrieval operations of the slave machines, the unnecessary movement of the mother machine can be minimized, the frequency of slave machine deployment and retrieval can be reduced, and the overall operation efficiency can be significantly improved and the inspection cycle can be shortened. The target inspection task package generated for each slave machine clarifies the core execution instructions such as the single-pipe climbing path, inter-pipe transition rules, and inspection parameter configuration during the operation of the slave machine. This allows the slave machine to complete all inspection operations autonomously based on the task package without real-time human intervention, realizing the autonomous execution of inspection operations, while ensuring that the inspection operations in different areas meet the corresponding accuracy and parameter requirements.

[0039] Furthermore, the steps of controlling the mother machine to move to the corresponding task location according to the movement path sequence, sequentially deploying and docking the submachines, synchronously controlling each submachine to perform multi-dimensional inspection of the heated surface pipe wall within the corresponding gridded inspection area, and receiving the inspection data transmitted back by the submachines and binding the data with spatial coordinates include: Control the mother machine to move to the target task point in the movement path sequence, complete the mother machine attitude lock and docking platform leveling, drive the delivery and recovery mechanism to perform the daughter machine delivery operation, and attach the daughter machine to the target heated surface tube wall; The corresponding target inspection task package is sent to the submachine, and the submachine is controlled to perform single pipe climbing, pipe transition and obstacle avoidance operations along the operation path in the task package within the gridded inspection area. Simultaneously, the ultrasonic thickness measuring device, eddy current detection device and visual inspection device on the submachine are controlled to perform multi-dimensional inspection operations of pipe wall thickness, internal defects and surface cracks. The system receives the detection data and real-time pose data transmitted back from the slave unit in real time, and binds the detection data with the spatial coordinates and pipe segment number of the corresponding heated surface tube to generate a detection dataset with spatial positioning tags. After the slave unit completes its work in the corresponding gridded detection area, it is controlled to return to the docking point, and the mother machine's deployment and retrieval mechanism is driven to complete the docking and retrieval of the slave unit.

[0040] It is important to note that the precise movement of the mother unit to the target task location is fundamental to ensuring the successor unit can enter the target inspection area. Locking the mother unit's attitude and leveling the docking platform eliminates attitude deviations caused by the mother unit's parking in the pipe array environment, ensuring the deployment and retrieval mechanism can accurately and stably execute deployment actions. This avoids deployment failures and inaccurate pipe wall contact caused by platform tilting or attitude swaying. The deployment and retrieval mechanism precisely attaches the successor unit to the target heated pipe wall, allowing it to smoothly enter operational mode and begin single-pipe climbing and inspection operations. By issuing target inspection task packages to the successor unit, it understands all the operational requirements and execution rules, enabling autonomous operation. The successor unit performs single-pipe climbing, inter-pipe transitions, and obstacle avoidance along the planned path, achieving convenient movement throughout the entire area within the confined space of densely packed pipes. This solves the problems of traditional inspection equipment's inability to move flexibly between densely packed pipes and insufficient coverage. Simultaneously, through the synergy of ultrasonic thickness measurement devices, eddy current detection devices, and visual inspection devices, multi-dimensional detection of all types of defects, including thinning of pipe wall thickness, internal buried defects, and surface cracks, is achieved. This comprehensively covers the core inspection needs of boiler heating surface tubes, avoiding the omissions caused by single inspection methods and ensuring the comprehensiveness and accuracy of inspection data. In traditional inspection methods, the inspection data and pipe location cannot be precisely correlated, making it difficult to accurately locate defects after they are discovered, which brings great difficulties to subsequent operation and maintenance. By receiving inspection data and real-time pose data from the slave unit in real time, each set of inspection data is bound to the corresponding spatial coordinates and pipe segment number of the heating surface tube, so that each piece of inspection data has a unique corresponding spatial location. The generated inspection dataset with spatial positioning tags enables full-process traceability of inspection data and precise location of defects, providing a precise basis for subsequent defect analysis and maintenance plan formulation. After completing all operations, the slave unit returns to the preset docking point, ensuring that the mother unit's deployment and recovery mechanism accurately docks with the slave unit. The deployment and recovery mechanism completes the docking and recovery of the slave unit, ensuring that the slave unit is safely and stably recovered to the mother unit. This provides a foundation for the slave unit to be transferred to the next work point, charged, and maintained, while avoiding the risk of the slave unit being left inside the furnace, thus completing a complete closed loop for a single set of testing tasks.

[0041] Furthermore, the steps of completing the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generating supplementary detection tasks for uncovered areas, and repeatedly executing the mother unit movement, slave unit deployment and retrieval operations until all detection tasks are completed include: Based on the detection dataset with spatial positioning labels, all pipe section areas that have been inspected are marked in the calibrated 3D structural model of the boiler heating surface, and a detection coverage heat map is generated. The detection coverage heat map is compared with the target coverage area in the preset detection task instruction to identify the uncovered pipe section area or the invalid detection data, and mark it as the area to be retested; For the area to be retested, a retesting task package is generated, the master machine's movement path sequence and the slave machine's operation task are updated, and the master machine and slave machine are controlled to repeatedly execute the corresponding retesting operation until the detection coverage meets the preset task requirements and all detection tasks are completed.

[0042] It's worth noting that the detection dataset with spatial positioning labels allows for precise marking of each inspected pipe segment within a accurately matched 3D structural model of the actual site. This clearly reconstructs the distribution of inspected and uninspected areas, generating a detection coverage heatmap that intuitively and visually presents the detection coverage of the entire heated surface. This makes uncovered and low-coverage areas immediately apparent, providing a direct and accurate basis for subsequent coverage comparison and uncovered area identification. By comparing the detection coverage heatmap with the target coverage range in the preset detection task instructions, uncovered areas that haven't been inspected and pipe segments with invalid detection data can be accurately identified. These areas are marked as areas requiring supplementary testing, clarifying the specific targets and scope of supplementary testing, avoiding inefficient work caused by blind supplementary testing, and fundamentally eliminating the risk of missed detections. The third sub-step is the execution phase of the supplementary testing closed loop. A dedicated supplementary testing task package is generated for the area to be tested, which clarifies the scope, accuracy requirements, and operational parameters of the supplementary testing, ensuring that the supplementary testing meets the task requirements. At the same time, the movement path sequence of the main machine and the operation tasks of the slave machine are updated, enabling the supplementary testing to be executed efficiently and accurately. By repeatedly executing the corresponding operation process until the detection coverage fully meets the preset task requirements, a complete closed loop of "detection-verification-supplementary testing-re-verification" is formed. From the process perspective, it thoroughly ensures full coverage and no blind spots in the detection of boiler heating surfaces, solving the core pain points of traditional detection methods, such as many blind spots and high missed detection rates.

[0043] Furthermore, the method also includes: Defect features are extracted from the detection dataset bound to spatial coordinates to identify abnormal data such as pipe wall thinning, internal defects, and surface cracks. The spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are then extracted. In the calibrated 3D structural model of the boiler heating surface, the defect points are visually marked, and a boiler heating surface inspection report containing the defect location, defect type, and defect level is generated.

[0044] It is worth noting that by extracting defect features from the detection dataset bound to spatial coordinates, abnormal data such as pipe wall thinning, internal defects, and surface cracks can be accurately identified from a large amount of normal detection data. Simultaneously, the spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are extracted, achieving accurate defect identification and parameter quantification. This provides quantitative data support for subsequent defect assessment and operation and maintenance decisions. Visualizing and marking defect locations in a 3D structural model matching the actual site allows operation and maintenance personnel to intuitively and clearly grasp the specific location and distribution of defects. The generated boiler heating surface inspection report fully includes core information such as the location, type, and level of defects, providing authoritative and accurate decision-making basis for boiler operation and maintenance, life assessment, and operational strategy adjustments. This allows the entire detection method to not only achieve the data collection objective but also realize the transformation from detection data to operation and maintenance value.

[0045] Please see Figure 2 The third embodiment of the present invention provides: A boiler heating surface mother-daughter type detection system, wherein the system includes: The planning module is used to obtain the three-dimensional structural model of the heating surface inside the boiler and the preset detection task instructions, divide the three-dimensional structural model into a pipe row area grid, and generate an initial task allocation set for the corresponding grid detection area. The positioning module is used to control the mother machine to carry at least one set of daughter machines into the boiler furnace. It collects global environmental data of the tube bank in the furnace in real time through the global sensing device on the mother machine, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. The update module is used to dynamically update the initial task allocation set based on the calibrated 3D structural model and real-time positioning results, and generate the mother machine's movement path sequence and the corresponding target detection task package for each submachine. The operation module is used to control the mother machine to move to the corresponding task point according to the movement path sequence, and to complete the deployment and docking and retrieval of the sub-machines in sequence. It also controls each sub-machine to perform multi-dimensional inspection of the heated surface pipe wall in the corresponding gridded inspection area, and receives the inspection data transmitted back by the sub-machines and completes the binding of the data with spatial coordinates. The verification module is used to complete the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generate supplementary test tasks for uncovered areas, and repeatedly execute the operation of moving the main unit, deploying slave units for detection and retrieval until all detection tasks are completed.

[0046] Furthermore, the planning module is specifically used for: Extract the tube bank structural parameters of the water-cooled wall, superheater, reheater, and economizer from the three-dimensional structural model of the boiler heating surface, including tube diameter, tube spacing, number of rows, bending angle, and spatial layout coordinates; Based on the detection accuracy requirements, coverage and operation cycle in the preset detection task instructions, the tube array structure is divided into grid areas, taking a single heated surface tube as the smallest unit, to generate multiple non-overlapping grid detection areas. For each gridded detection area, match the corresponding submachine detection operation parameters, including detection range, detection type, operation path and data acquisition requirements. Combine the movable range of the master machine and the operation radius of the submachine to generate the initial task allocation set containing multiple sets of submachine-detection area pairing relationships.

[0047] Furthermore, the positioning module is specifically used for: The machine is controlled to move along a preset initial path inside the furnace. Through the lidar and binocular vision device mounted on the machine, three-dimensional point cloud data and image data of the tube bank structure inside the furnace are collected in real time to generate a real-time point cloud map of the furnace environment. The real-time environmental point cloud map is matched and the pre-stored three-dimensional structural model of the boiler heating surface is registered with feature points to identify the deviation between the model and the actual structure, and the pipe row coordinates and spatial structure of the three-dimensional structural model are calibrated in real time. Based on the registered coordinate system, and combined with the fusion data from the inertial measurement unit, wheeled odometer, and visual odometer on the mother machine, the real-time eight-degree-of-freedom pose parameters of the mother machine are calculated. Based on the deviation between the pose parameters and the preset path, the pose closed-loop control of the mother machine is completed.

[0048] Furthermore, the update module is specifically used for: Based on the calibrated 3D structural model, the actual obstacles, irregular tube structures and impassable areas inside the furnace are identified, and the boundaries of the gridded detection areas in the initial task allocation set are adjusted and merged / split. Based on the real-time positioning results of the mother machine, the passable range, the rated operating capacity of each submachine, and the remaining power, the submachine tasks are redistributed in the adjusted gridded detection area, and the pairing relationship between submachine and detection area is updated. Based on the updated pairing relationship, with the optimization objectives of minimizing the total movement path of the mother machine and minimizing the number of deployment and retrieval times of the daughter machines, the mother machine movement path sequence is planned and generated. At the same time, for each daughter machine, a target detection task package is generated, which includes a single-pipe climbing path, inter-pipe transition rules, and detection parameter configuration for the corresponding gridded detection area.

[0049] Furthermore, the operation module is specifically used for: Control the mother machine to move to the target task point in the movement path sequence, complete the mother machine attitude lock and docking platform leveling, drive the delivery and recovery mechanism to perform the daughter machine delivery operation, and attach the daughter machine to the target heated surface tube wall; The corresponding target inspection task package is sent to the submachine, and the submachine is controlled to perform single pipe climbing, pipe transition and obstacle avoidance operations along the operation path in the task package within the gridded inspection area. Simultaneously, the ultrasonic thickness measuring device, eddy current detection device and visual inspection device on the submachine are controlled to perform multi-dimensional inspection operations of pipe wall thickness, internal defects and surface cracks. The system receives the detection data and real-time pose data transmitted back from the slave unit in real time, and binds the detection data with the spatial coordinates and pipe segment number of the corresponding heated surface tube to generate a detection dataset with spatial positioning tags. After the slave unit completes its work in the corresponding gridded detection area, it is controlled to return to the docking point, and the mother machine's deployment and retrieval mechanism is driven to complete the docking and retrieval of the slave unit.

[0050] Furthermore, the verification module is specifically used for: Based on the detection dataset with spatial positioning labels, all pipe section areas that have been inspected are marked in the calibrated 3D structural model of the boiler heating surface, and a detection coverage heat map is generated. The detection coverage heat map is compared with the target coverage area in the preset detection task instruction to identify the uncovered pipe section area or the invalid detection data, and mark it as the area to be retested; For the area to be retested, a retesting task package is generated, the master machine's movement path sequence and the slave machine's operation task are updated, and the master machine and slave machine are controlled to repeatedly execute the corresponding retesting operation until the detection coverage meets the preset task requirements and all detection tasks are completed.

[0051] Furthermore, the verification module is specifically used for: Defect features are extracted from the detection dataset bound to spatial coordinates to identify abnormal data such as pipe wall thinning, internal defects, and surface cracks. The spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are then extracted. In the calibrated 3D structural model of the boiler heating surface, the defect points are visually marked, and a boiler heating surface inspection report containing the defect location, defect type, and defect level is generated.

[0052] The fourth embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the boiler heating surface mother-daughter type detection method as described above.

[0053] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the boiler heating surface mother-daughter detection method as described above.

[0054] In summary, the boiler heating surface mother-daughter type detection method and system provided by this invention can effectively solve the core pain point of traditional detection schemes that cannot simultaneously achieve large-scale convenient and local high-precision fine detection through the grid division of boiler heating surface tube banks and the mother-daughter type collaborative operation architecture. It can adapt to the detection needs of different boiler types and tube bank structures in all scenarios. Through the global environment perception of the mother machine and the real-time calibration of the three-dimensional structural model, the system's adaptability to the complex tube bank environment in the furnace and the global positioning accuracy are improved. Through dynamic task updates and mother-daughter type collaborative operation, the accurate allocation and efficient execution of detection tasks are realized. Combined with the full coverage verification and supplementary test closed-loop mechanism, the full area coverage of heating surface detection is guaranteed, while the detection operation efficiency and data accuracy are greatly improved, and the safety risks of manual furnace operation are reduced.

[0055] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for detecting the mother-daughter type of boiler heating surface, characterized in that, The method includes: Obtain the three-dimensional structural model of the heating surface inside the boiler furnace and the preset detection task instructions, divide the three-dimensional structural model into a pipe row area grid, and generate an initial task allocation set for the corresponding grid detection area; The machine controls the mother machine to carry at least one set of daughter machines into the boiler furnace. The global sensing device on the mother machine collects global environmental data of the tube bank in the furnace in real time, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. Based on the calibrated 3D structural model and real-time positioning results, the initial task allocation set is dynamically updated to generate the mother machine's movement path sequence and the corresponding target detection task package for each sub-machine. According to the movement path sequence, the master machine is controlled to move to the corresponding task point, and the deployment and docking of the sub-machines are completed in sequence. Simultaneously, each sub-machine is controlled to perform multi-dimensional inspection of the heated surface pipe wall in the corresponding gridded inspection area, and the inspection data transmitted back by the sub-machines is received and the data is bound to the spatial coordinates. Based on the detection data returned by all slave units, complete the full coverage verification of the boiler heating surface detection area, generate supplementary detection tasks for uncovered areas, and repeatedly execute the operations of moving the main unit, deploying slave units for detection and retrieval until all detection tasks are completed.

2. The boiler heating surface mother-daughter type detection method according to claim 1, characterized in that, The steps of acquiring the three-dimensional structural model of the boiler's internal heating surface and the preset detection task instructions, dividing the three-dimensional structural model into a pipe row area grid, and generating an initial task allocation set for the corresponding gridded detection area include: Extract the tube bank structural parameters of the water-cooled wall, superheater, reheater, and economizer from the three-dimensional structural model of the boiler heating surface, including tube diameter, tube spacing, number of rows, bending angle, and spatial layout coordinates; Based on the detection accuracy requirements, coverage and operation cycle in the preset detection task instructions, the tube array structure is divided into grid areas, taking a single heated surface tube as the smallest unit, to generate multiple non-overlapping grid detection areas. For each gridded detection area, match the corresponding submachine detection operation parameters, including detection range, detection type, operation path and data acquisition requirements. Combine the movable range of the master machine and the operation radius of the submachine to generate the initial task allocation set containing multiple sets of submachine-detection area pairing relationships.

3. The boiler heating surface mother-daughter type detection method according to claim 1, characterized in that, The control machine, equipped with at least one set of slave units, enters the boiler furnace. Through a global sensing device mounted on the machine, it collects real-time global environmental data of the boiler's tube banks, performs real-time calibration of the three-dimensional structural model, and simultaneously completes the steps of global positioning and closed-loop attitude control of the machine within the furnace, including: The machine is controlled to move along a preset initial path inside the furnace. Through the lidar and binocular vision device mounted on the machine, three-dimensional point cloud data and image data of the tube bank structure inside the furnace are collected in real time to generate a real-time point cloud map of the furnace environment. The real-time environmental point cloud map is matched and the pre-stored three-dimensional structural model of the boiler heating surface is registered with feature points to identify the deviation between the model and the actual structure, and the pipe row coordinates and spatial structure of the three-dimensional structural model are calibrated in real time. Based on the registered coordinate system, and combined with the fusion data from the inertial measurement unit, wheeled odometer, and visual odometer on the mother machine, the real-time eight-degree-of-freedom pose parameters of the mother machine are calculated. Based on the deviation between the pose parameters and the preset path, the pose closed-loop control of the mother machine is completed.

4. The boiler heating surface mother-daughter type detection method according to claim 1, characterized in that, The steps of dynamically updating the initial task allocation set based on the calibrated 3D structural model and real-time positioning results to generate the mother machine's movement path sequence and the corresponding target detection task packages for each sub-machine include: Based on the calibrated 3D structural model, the actual obstacles, irregular tube structures and impassable areas inside the furnace are identified, and the boundaries of the gridded detection areas in the initial task allocation set are adjusted and merged / split. Based on the real-time positioning results of the mother machine, the passable range, the rated operating capacity of each submachine, and the remaining power, the submachine tasks are redistributed in the adjusted gridded detection area, and the pairing relationship between submachine and detection area is updated. Based on the updated pairing relationship, with the optimization objectives of minimizing the total movement path of the mother machine and minimizing the number of deployment and retrieval times of the daughter machines, the mother machine movement path sequence is planned and generated. At the same time, for each daughter machine, a target detection task package is generated, which includes a single-pipe climbing path, inter-pipe transition rules, and detection parameter configuration for the corresponding gridded detection area.

5. The boiler heating surface mother-daughter type detection method according to claim 1, characterized in that, The steps of controlling the mother machine to move to the corresponding task location according to the movement path sequence, sequentially deploying and docking the submachines, synchronously controlling each submachine to perform multi-dimensional inspection of the heated surface pipe wall within the corresponding gridded inspection area, and receiving the inspection data transmitted back by the submachines and binding the data with spatial coordinates include: Control the mother machine to move to the target task point in the movement path sequence, complete the mother machine attitude lock and docking platform leveling, drive the delivery and recovery mechanism to perform the daughter machine delivery operation, and attach the daughter machine to the target heated surface tube wall; The corresponding target inspection task package is sent to the submachine, and the submachine is controlled to perform single pipe climbing, pipe transition and obstacle avoidance operations along the operation path in the task package within the gridded inspection area. Simultaneously, the ultrasonic thickness measuring device, eddy current detection device and visual inspection device on the submachine are controlled to perform multi-dimensional inspection operations of pipe wall thickness, internal defects and surface cracks. The system receives the detection data and real-time pose data transmitted back from the slave unit in real time, and binds the detection data with the spatial coordinates and pipe segment number of the corresponding heated surface tube to generate a detection dataset with spatial positioning tags. After the slave unit completes its work in the corresponding gridded detection area, it is controlled to return to the docking point, and the mother machine's deployment and retrieval mechanism is driven to complete the docking and retrieval of the slave unit.

6. The boiler heating surface mother-daughter type detection method according to claim 1, characterized in that, The steps of completing the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generating supplementary detection tasks for uncovered areas, and repeatedly executing the operations of moving the main unit, deploying slave units for detection, and retrieving them until all detection tasks are completed include: Based on the detection dataset with spatial positioning labels, all pipe section areas that have been inspected are marked in the calibrated 3D structural model of the boiler heating surface, and a detection coverage heat map is generated. The detection coverage heat map is compared with the target coverage area in the preset detection task instruction to identify the uncovered pipe section area or the invalid detection data, and mark it as the area to be retested; For the area to be retested, a retesting task package is generated, the master machine's movement path sequence and the slave machine's operation task are updated, and the master machine and slave machine are controlled to repeatedly execute the corresponding retesting operation until the detection coverage meets the preset task requirements and all detection tasks are completed.

7. The boiler heating surface mother-daughter type detection method according to claim 6, characterized in that, The method further includes: Defect features are extracted from the detection dataset bound to spatial coordinates to identify abnormal data such as pipe wall thinning, internal defects, and surface cracks. The spatial coordinates, pipe segment numbers, and defect parameters corresponding to the abnormal data are then extracted. In the calibrated 3D structural model of the boiler heating surface, the defect points are visually marked, and a boiler heating surface inspection report containing the defect location, defect type, and defect level is generated.

8. A boiler heating surface mother-daughter type detection system, characterized in that, The system includes: The planning module is used to obtain the three-dimensional structural model of the heating surface inside the boiler and the preset detection task instructions, divide the three-dimensional structural model into a pipe row area grid, and generate an initial task allocation set for the corresponding grid detection area. The positioning module is used to control the mother machine to carry at least one set of daughter machines into the boiler furnace. It collects global environmental data of the tube bank in the furnace in real time through the global sensing device on the mother machine, performs real-time calibration of the three-dimensional structural model, and completes the global positioning and posture closed-loop control of the mother machine in the furnace. The update module is used to dynamically update the initial task allocation set based on the calibrated 3D structural model and real-time positioning results, and generate the mother machine's movement path sequence and the corresponding target detection task package for each submachine. The operation module is used to control the mother machine to move to the corresponding task point according to the movement path sequence, and to complete the deployment and docking and retrieval of the sub-machines in sequence. It also controls each sub-machine to perform multi-dimensional inspection of the heated surface pipe wall in the corresponding gridded inspection area, and receives the inspection data transmitted back by the sub-machines and completes the binding of the data with spatial coordinates. The verification module is used to complete the full coverage verification of the boiler heating surface detection area based on the detection data returned by all slave units, generate supplementary test tasks for uncovered areas, and repeatedly execute the operation of moving the main unit, deploying slave units for detection and retrieval until all detection tasks are completed.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the boiler heating surface mother-daughter type detection method as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the boiler heating surface mother-daughter type detection method as described in any one of claims 1 to 7.

Images