Ship construction flatness inspection equipment and use method thereof

Abstract

The invention relates to shipbuilding flatness inspection equipment and a use method thereof, and belongs to the technical field of shipbuilding flatness inspection. The shipbuilding flatness inspection equipment comprises a magnetic adsorption type mobile robot platform, a high-precision vision module, an edge computing unit and a cloud data management platform; the magnetic adsorption type mobile robot platform is provided with a crawler-type walking mechanism and a self-adaptive vacuum adsorption unit. According to the invention, efficient and high-precision whole ship flatness intelligent detection can be realized through multi-sensor fusion and a self-adaptive robot platform, the system is suitable for real-time quality monitoring in ship body section construction, deck assembly and fitting-out stages, local fine scanning can be executed along a welding seam area in a ship body section detection scene, and the detection accuracy is improved. A welding deformation out-of-tolerance area is synchronously marked, a non-key plane area can be switched to a global rapid scanning mode, and in the deck assembly stage, a cloud platform can integrate data in real time and assist in carrying out an assembly dislocation correction scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of ship flatness inspection, and in particular to a ship construction flatness inspection device and a use method thereof. Background Art

[0002] Ships are a general term for various vessels. They are vehicles that can sail or anchor in waters for transportation or operations. They have different technical performance, equipment, and structural types according to different usage requirements. During ship manufacturing and maintenance, the flatness inspection of the ship's deck is a very important task, so a flatness inspection device is needed to perform this inspection.

[0003] Traditional ship flatness inspection relies on manual use of tools such as rulers and levels, which is inefficient and susceptible to subjective factors. Existing digital equipment (such as laser scanners) is highly accurate, but complex to operate, expensive, and difficult to adapt to the complex curved surface environment of the hull. Therefore, a shipbuilding flatness inspection device and its use method are proposed. Through multi-sensor fusion and an adaptive robotic platform, it can achieve efficient and high-precision intelligent flatness inspection of the entire ship. It is suitable for real-time quality monitoring during the hull block construction, deck assembly, and outfitting stages. In the hull block inspection scenario, it can perform local fine scanning along the weld area, simultaneously marking areas with out-of-tolerance weld deformation. For non-critical planar areas, it can switch to a global rapid scanning mode. During the deck assembly stage, data can be integrated in real time on a cloud platform to assist in the development of assembly misalignment correction solutions. Summary of the Invention

[0004] The present invention provides a shipbuilding flatness inspection device and a method for using the same, which solves the problems raised by the above-mentioned existing background technology. Through multi-sensor fusion and an adaptive robot platform, efficient and high-precision intelligent detection of the flatness of the entire ship can be achieved, and the device is suitable for real-time quality monitoring of the hull section construction, deck assembly and outfitting stages.

[0005] The present invention solves the above-mentioned technical problems as follows: A shipbuilding flatness inspection device includes a magnetic adsorption mobile robot platform, a magnetic adsorption mobile robot platform, a high-precision vision module, an edge computing unit, and a cloud data management platform. The magnetic adsorption mobile robot platform is provided with a crawler walking mechanism and an adaptive vacuum adsorption unit. The multi-line laser radar array is installed on the magnetic adsorption mobile robot platform. The multi-line laser radar array includes a 16-line vertical scanning radar and a 360-degree rotating horizontal radar. The high-precision vision module is installed on the magnetic adsorption mobile robot platform. The magnetic adsorption mobile robot platform includes a binocular stereo camera and an infrared thermal imager. The edge computing unit has a built-in FPGA acceleration chip and an AI inference engine.

[0006] The cloud data management platform includes a data acquisition and transmission layer, a data storage and management layer, a data processing and analysis engine, an AI model and algorithm platform, and a visualization and interaction layer. The data acquisition and transmission layer is provided with a network transmission protocol, the data storage and management layer is provided with a distributed database and a data lake architecture, the data processing and analysis engine is provided with a multimodal data fusion algorithm and a dynamic reference surface calibration module, the AI ​​model and algorithm platform is provided with a deep learning framework module and a generative adversarial network module, and the visualization and interaction layer is provided with a three-dimensional deviation heat map generation module and a compliance report generation module;

[0007] The usage includes the following steps:

[0008] S1: Adaptive Adsorption and Path Planning: The magnetic adsorption mobile robot platform can vertically adhere to the ship's surface through an adaptive vacuum adsorption unit. It uses a 16-line vertical scanning radar and a 360-degree rotating horizontal radar to obtain 3D point cloud data of the ship's surface. It builds a 3D map of the ship's surface based on the SLAM algorithm and automatically generates the optimal scanning path based on preset detection priorities, dividing the local detection grid into global key areas.

[0009] S2: Synchronous multimodal data acquisition: A multi-line lidar array scans the surface geometry, a high-precision vision module captures texture images, and an IMU records the robot's posture data;

[0010] S3: Real-time edge processing: The edge computing unit has a built-in multimodal data fusion algorithm that fuses laser point clouds, visual images, and IMU inertial data. It uses Kalman filtering to eliminate motion errors, calculate surface curvature, slope, and deviation from the design model, and generate a surface flatness model with sub-millimeter accuracy.

[0011] S4: AI defect diagnosis: Uses the pre-trained ResNet-3D network to classify abnormal areas and output defect types and repair suggestions.

[0012] S5: Global data integration: The cloud-based data management platform receives real-time data from mobile robots, lidars, and vision modules through edge computing units, including 3D point clouds, image textures, and IMU poses. It stitches multi-robot inspection data to generate a 3D flatness heat map and compliance report for the entire ship.

[0013] On the basis of the above technical solution, the present invention can also be improved as follows.

[0014] Furthermore, the multi-line lidar array adopts wavelength diversity technology to synchronously emit 905nm and 1550nm laser beams, which are used for high-precision close-range detection and long-range anti-interference scanning respectively.

[0015] Furthermore, the cloud platform integrates a deep learning anomaly detection model based on convolutional neural networks and point cloud Transformer architecture to automatically identify uneven welds, plate warping and assembly misalignment defects, and mark the risk level.

[0016] Furthermore, in step S3, a dynamic reference surface calibration technology is used to dynamically adjust the local flatness evaluation benchmark according to the theoretical curvature of the hull line shape to eliminate misjudgment of large curvature surfaces.

[0017] Furthermore, the cloud data management platform adopts 5G or Industrial Internet of Things (IIoT) protocols (such as MQTT, OPC UA) to achieve low-latency data transmission and ensure the real-time and integrity of data.

[0018] Furthermore, the distributed database uses a time series database (such as InfluxDB) to store sensor data collected at a high frequency, and combines it with a relational database (such as PostgreSQL) to manage structured detection reports and metadata. The data lake architecture supports the original storage of unstructured data (such as point clouds and images) and implements batch processing through Hadoop or Spark.

[0019] Furthermore, the multimodal data fusion algorithm of the data processing and analysis engine uses Kalman filtering or particle filtering technology to fuse lidar, visual images and inertial data, eliminate motion errors and generate surface models with sub-millimeter accuracy. The dynamic reference surface calibration module dynamically adjusts the local flatness evaluation benchmark according to the theoretical curvature of the hull to avoid misjudgment of large curvature surfaces.

[0020] Furthermore, the AI ​​model is integrated with the algorithm platform's deep learning framework, TensorFlow or PyTorch, and supports pre-trained ResNet-3D and point cloud Transformer models for defect classification such as uneven welds and plate warping. The generative adversarial network module optimizes the model's ability to recognize small sample defects through generative adversarial training, thereby improving generalization.

[0021] Furthermore, the three-dimensional deviation heat map generation module of the visualization and interaction layer renders the global three-dimensional model based on the WebGL or Unity engine, and intuitively displays the flatness deviation distribution through color gradients. The compliance report generation module automatically matches the ISO 484 and DNV GL-ST-0111 standards and generates a test report in PDF or Excel format.

[0022] The beneficial effects of the present invention are as follows: the present invention provides a shipbuilding flatness inspection device and a method for using the same, which have the following advantages:

[0023] 1. Compared with the traditional monitoring method that relies on manually operated rulers and laser scanners, this application uses magnetic adsorption robots for autonomous navigation, covering the hull surface (including vertical surfaces) without human intervention, thereby effectively improving detection efficiency. Through the collaborative operation of multiple robots, large sections can be scanned in parallel, which can effectively shorten the inspection period;

[0024] 2. Multi-sensor fusion is used to balance accuracy and coverage. Compared with traditional single devices such as laser scanners, which cannot simultaneously achieve high local precision, the multi-line lidar + vision module uses 1550nm laser to achieve 0.1mm weld detection at close range and 905nm laser to complete hull linear scanning at long distance. Dynamic reference surface calibration automatically adapts to the hull curvature, avoiding the misjudgment of large curved surfaces caused by traditional fixed reference surfaces.

[0025] 3. Compared with traditional defect judgment dominated by manual experience, which is prone to missing hidden problems, the AI ​​multi-model collaborative ResNet-3D is used to identify geometric deformation and infrared thermal imaging to detect stress anomalies. It can provide real-time repair suggestions, generate 3D thermal maps in the cloud and mark repair priorities, guiding workers to accurately rework.

[0026] 4. Such shipbuilding flatness inspection equipment can achieve efficient and high-precision intelligent flatness inspection of the entire ship through multi-sensor fusion and adaptive robot platform. It is suitable for real-time quality monitoring of hull section construction, deck assembly and outfitting stages. In the hull section inspection scenario, it can perform local fine scanning along the weld area, and simultaneously mark the areas with excessive welding deformation. It can switch to global fast scanning mode for non-critical plane areas. During the deck assembly stage, the cloud platform can integrate data in real time to assist in the assembly misalignment correction plan.

[0027] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and to implement it according to the contents of the description, the following preferred embodiments of the present invention are described in detail with reference to the accompanying drawings. The specific implementation methods of the present invention are given in detail by the following embodiments and the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0029] Figure 1 A flow chart of a shipbuilding flatness inspection device and a method of using the same is provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following is combined with Figure 1The principles and features of the present invention are described, and the examples given are only for the purpose of explaining the present invention and are not intended to limit the scope of the present invention. The following paragraphs describe the present invention in more detail by way of example with reference to the accompanying drawings. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are in a very simplified form and are not in exact proportions, and are only used for the purpose of conveniently and clearly assisting in illustrating the embodiments of the present invention.

[0031] It should be noted that when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may also be a central component. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may also be a central component. When a component is considered to be "set on" another component, it may be directly set on the other component or there may also be a central component. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0033] like Figure 1 As shown, the present invention provides a shipbuilding flatness inspection device, including a magnetic adsorption mobile robot platform, a magnetic adsorption mobile robot platform, a high-precision vision module, an edge computing unit, and a cloud data management platform. The magnetic adsorption mobile robot platform is provided with a crawler walking mechanism and an adaptive vacuum adsorption unit. A multi-line laser radar array is installed on the magnetic adsorption mobile robot platform. The multi-line laser radar array includes a 16-line vertical scanning radar and a 360-degree rotating horizontal radar. The high-precision vision module is installed on the magnetic adsorption mobile robot platform. The magnetic adsorption mobile robot platform includes a binocular stereo camera and an infrared thermal imager. The edge computing unit has a built-in FPGA acceleration chip and an AI inference engine.

[0034] The cloud data management platform includes a data collection and transmission layer, a data storage and management layer, a data processing and analysis engine, an AI model and algorithm platform, and a visualization and interaction layer. The data collection and transmission layer is equipped with a network transmission protocol, the data storage and management layer is equipped with a distributed database and data lake architecture, the data processing and analysis engine is equipped with a multimodal data fusion algorithm and a dynamic datum calibration module, the AI ​​model and algorithm platform is equipped with a deep learning framework module and a generative adversarial network module, and the visualization and interaction layer is equipped with a three-dimensional deviation heat map generation module and a compliance report generation module.

[0035] Preferably, the multi-line lidar array uses wavelength diversity technology to synchronously emit 905nm and 1550nm laser beams, which are used for high-precision close-range detection and long-range anti-interference scanning, respectively.

[0036] Preferably, the cloud platform integrates a deep learning anomaly detection model based on convolutional neural networks and point cloud Transformer architecture to automatically identify uneven welds, plate warping, and assembly misalignment defects, and mark the risk level.

[0037] Preferably, the cloud data management platform uses 5G or Industrial Internet of Things (IIoT) protocols (such as MQTT and OPC UA) to achieve low-latency data transmission and ensure data real-time and integrity.

[0038] Preferably, the distributed database uses a time series database (such as InfluxDB) to store sensor data collected at a high frequency, combined with a relational database (such as PostgreSQL) to manage structured detection reports and metadata. The data lake architecture supports the original storage of unstructured data (such as point clouds and images) and implements batch processing through Hadoop or Spark.

[0039] Preferably, the multimodal data fusion algorithm of the data processing and analysis engine uses Kalman filtering or particle filtering technology to fuse lidar, visual images and inertial data, eliminate motion errors and generate surface models with sub-millimeter accuracy. The dynamic reference surface calibration module dynamically adjusts the local flatness evaluation benchmark according to the theoretical curvature of the hull to avoid misjudgment of large curvature surfaces.

[0040] Preferably, the deep learning framework of the AI ​​model and algorithm platform is integrated with TensorFlow or PyTorch, and supports pre-trained ResNet-3D and point cloud Transformer models for defect classification such as uneven welds and plate warping. The generative adversarial network module optimizes the model's ability to recognize small sample defects through generative adversarial training, thereby improving generalization.

[0041] Preferably, the three-dimensional deviation heat map generation module of the visualization and interaction layer renders the global three-dimensional model based on the WebGL or Unity engine, and intuitively displays the flatness deviation distribution through color gradients. The compliance report generation module automatically matches the ISO484 and DNV GL-ST-0111 standards and generates a test report in PDF or Excel format.

[0042] The specific working principle and method of use of the present invention are as follows:

[0043] S1: Adaptive adsorption and path planning: The magnetic adsorption mobile robot platform can vertically adhere to the surface of the ship through an adaptive vacuum adsorption unit. It can obtain three-dimensional point cloud data of the ship surface through a 16-line vertical scanning radar and a 360-degree rotating horizontal radar. It builds a three-dimensional map of the hull based on the SLAM algorithm and automatically generates the optimal scanning path based on the preset detection priority. It divides the local detection grid and the global key area. The dynamic path planning algorithm is based on the global planning of SLAM, uses LiDAR-SLAM to build a three-dimensional map of the hull, and combines AI algorithms to generate the optimal detection path. Based on the global planning of SLAM, DWA (dynamic window method) is used to adjust the robot's motion trajectory in real time to avoid collisions with obstacles.

[0044] S2: Synchronous multimodal data acquisition: A multi-line lidar array scans the surface geometry, a high-precision vision module captures texture images, and an IMU records the robot's pose data. The laser point cloud (geometry), visual image (texture and stress distribution), and IMU pose data (motion compensation) are input into the multimodal data fusion algorithm. The multi-frame point cloud is aligned to a unified coordinate system using the ICP (Iterative Closest Point) algorithm to complete point cloud registration. SLAM (Simultaneous Localization and Mapping) technology is used to map image feature points to point clouds for image-point cloud association. Kalman filtering is used to eliminate jitter errors caused by robot motion for dynamic error correction.

[0045] S3: Real-time edge processing: The edge computing unit has a built-in multimodal data fusion algorithm that fuses laser point clouds, visual images, and IMU inertial data. Kalman filtering eliminates motion errors, calculates surface curvature, slope, and deviation from the design model, and generates a surface flatness model with sub-millimeter accuracy. Dynamic reference surface calibration technology is used to dynamically adjust the local flatness evaluation benchmark based on the theoretical curvature of the hull line, eliminating misjudgments of surfaces with large curvature.

[0046] S4: AI Defect Diagnosis: This algorithm uses a pre-trained ResNet-3D network to classify abnormal areas and output defect types and repair suggestions. The AI ​​defect diagnosis algorithm's model architecture includes a ResNet-3D network designed for 3D point cloud data, which extracts multi-scale geometric features through a residual structure; a PointNet++ module that processes point clouds with non-uniform density and identifies local defects (such as pits and bumps); and a Transformer module that captures long-range dependencies for global analysis of assembly misalignment. The training strategy uses transfer learning, pre-training the model based on standard ship component datasets (such as ISO standard models). Online learning allows for real-time model updates to adapt to process differences across different shipyards.

[0047] S5: Global data integration: The cloud-based data management platform receives real-time data from mobile robots, lidars, and vision modules through edge computing units, including 3D point clouds, image textures, and IMU poses. It stitches multi-robot inspection data to generate a 3D flatness heat map and compliance report for the entire ship.

[0048] It should be noted that, in this document, relational terms such as first and second are used solely to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations. Matters not described in detail in this specification are well known to those skilled in the art.

[0049] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any ordinary technician in this industry can smoothly implement the present invention as shown in the drawings and described above. However, any equivalent changes, modifications and evolutions made by technicians familiar with this profession without departing from the scope of the technical solution of the present invention using the technical content disclosed above are all equivalent embodiments of the present invention. At the same time, any equivalent changes, modifications and evolutions made to the above embodiments based on the essential technology of the present invention are still within the scope of protection of the technical solution of the present invention.

Claims

1. A shipbuilding flatness inspection device, comprising a magnetic adsorption mobile robot platform, a magnetic adsorption mobile robot platform, a high-precision vision module, an edge computing unit, and a cloud data management platform, characterized in that: The magnetic adsorption mobile robot platform is provided with a crawler walking mechanism and an adaptive vacuum adsorption unit. The multi-line laser radar array is installed on the magnetic adsorption mobile robot platform. The multi-line laser radar array includes a 16-line vertical scanning radar and a 360-degree rotating horizontal radar. The high-precision vision module is installed on the magnetic adsorption mobile robot platform. The magnetic adsorption mobile robot platform includes a binocular stereo camera and an infrared thermal imager. The edge computing unit has a built-in FPGA acceleration chip and an AI inference engine. The cloud data management platform includes a data acquisition and transmission layer, a data storage and management layer, a data processing and analysis engine, an AI model and algorithm platform, and a visualization and interaction layer. The data acquisition and transmission layer is provided with a network transmission protocol, the data storage and management layer is provided with a distributed database and a data lake architecture, the data processing and analysis engine is provided with a multimodal data fusion algorithm and a dynamic reference surface calibration module, the AI ​​model and algorithm platform is provided with a deep learning framework module and a generative adversarial network module, and the visualization and interaction layer is provided with a three-dimensional deviation heat map generation module and a compliance report generation module; The usage includes the following steps: S1: Adaptive Adsorption and Path Planning: The magnetic adsorption mobile robot platform can vertically adhere to the ship's surface through an adaptive vacuum adsorption unit. It uses a 16-line vertical scanning radar and a 360-degree rotating horizontal radar to obtain 3D point cloud data of the ship's surface. It builds a 3D map of the ship's surface based on the SLAM algorithm and automatically generates the optimal scanning path based on preset detection priorities, dividing the local detection grid into global key areas. S2: Synchronous multimodal data acquisition: A multi-line lidar array scans the surface geometry, a high-precision vision module captures texture images, and an IMU records the robot's posture data; S3: Real-time edge processing: The edge computing unit has a built-in multimodal data fusion algorithm that fuses laser point clouds, visual images, and IMU inertial data. It uses Kalman filtering to eliminate motion errors, calculate surface curvature, slope, and deviation from the design model, and generate a surface flatness model with sub-millimeter accuracy. S4: AI defect diagnosis: Uses the pre-trained ResNet-3D network to classify abnormal areas and output defect types and repair suggestions. S5: Global data integration: The cloud-based data management platform receives real-time data from mobile robots, lidars, and vision modules through edge computing units, including 3D point clouds, image textures, and IMU poses. It stitches multi-robot inspection data to generate a 3D flatness heat map and compliance report for the entire ship.

2. A shipbuilding flatness inspection device according to claim 1, characterized in that: The multi-line laser radar array adopts wavelength diversity technology to synchronously emit 905nm and 1550nm laser beams.

3. A shipbuilding flatness inspection device according to claim 1, characterized in that: The cloud platform integrates a deep learning anomaly detection model based on convolutional neural networks and point cloud Transformer architecture.

4. A shipbuilding flatness inspection device according to claim 1, characterized in that: In step S3, a dynamic reference plane calibration technology is used to dynamically adjust the local flatness evaluation benchmark according to the theoretical curvature of the hull line.

5. The shipbuilding flatness inspection equipment according to claim 1, characterized in that: The cloud data management platform adopts 5G or Industrial Internet of Things (IIoT) protocols (such as MQTT, OPC UA).

6. The shipbuilding flatness inspection equipment according to claim 1, characterized in that: The distributed database uses a time-series database (such as InfluxDB) to store sensor data collected at a high frequency, and combines it with a relational database (such as PostgreSQL) to manage structured detection reports and metadata. The data lake architecture supports the original storage of unstructured data (such as point clouds and images) and implements batch processing through Hadoop or Spark.

7. The shipbuilding flatness inspection equipment according to claim 1, characterized in that: The multimodal data fusion algorithm of the data processing and analysis engine uses Kalman filtering or particle filtering technology to fuse lidar, visual images and inertial data, eliminate motion errors and generate a surface model with sub-millimeter accuracy. The dynamic reference surface calibration module dynamically adjusts the local flatness evaluation benchmark according to the theoretical curvature of the hull.

8. The shipbuilding flatness inspection equipment according to claim 1, characterized in that: The AI ​​model and algorithm platform's deep learning framework integrates TensorFlow or PyTorch, supports pre-trained ResNet-3D and point cloud Transformer models for defect classification, and the generative adversarial network module optimizes the model's ability to recognize small sample defects through generative adversarial training, thereby improving generalization.

9. The shipbuilding flatness inspection equipment according to claim 1, characterized in that: The 3D deviation heat map generation module of the visualization and interaction layer renders the global 3D model based on the WebGL or Unity engine, and intuitively displays the flatness deviation distribution through color gradients. The compliance report generation module automatically matches the ISO 484 and DNV GL-ST-0111 standards and generates inspection reports in PDF or Excel format.

Images