Intelligent inspection method and system for apparent diseases of railway housing construction equipment
By combining drones with artificial intelligence, intelligent inspection of surface defects in railway building equipment has been achieved, solving the problems of low efficiency, high safety risks, and low level of intelligence in existing technologies. This forms a complete business closed loop and is applicable to various inspection scenarios for railway facilities.
Patent Information
- Application Number
- CN202511835025.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-01-23
AI Technical Summary
Existing methods for detecting surface defects in railway building equipment suffer from problems such as low efficiency, high safety risks, strong subjectivity, and low level of intelligence, especially in complex and dangerous scenarios where efficient and accurate defect detection is difficult to achieve.
By combining drone technology with artificial intelligence, image data is collected by drones, and disease identification and classification are performed using intelligent detection terminals. Combined with 3D modeling and geolocation, standardized detection reports are generated, forming a complete business loop.
It has achieved unmanned inspection, reduced operational risks, improved detection efficiency and accuracy, and formed a standardized operation and maintenance management process, which is applicable to the detection of defects in various railway facilities.
Smart Images
Figure CN121386785A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of railway facility apparent disease detection, in particular to a railway housing facility apparent disease intelligent inspection method and system. BACKGROUND
[0002] At present, the apparent disease detection means of railway houses and buildings mainly relies on manual detection, specifically through manual close-range observation or remote observation with the help of a telescope. However, such traditional detection means has many drawbacks in practical application, especially when applied to complex scenes such as high-rise building facades, high-rise roofs, narrow spaces, dangerous houses and buildings to be repaired, the problems are more prominent: Firstly, information collection is difficult. Many high-rise buildings, large station facilities and the like are difficult for staff to reach the location of the disease for close-range observation due to their special location and height limitation, resulting in difficulty in disease observation and easy omission; Secondly, the subjectivity of judgment is strong. The existing detection means mainly rely on manual naked eye combined with simple tools such as telescopes for subjective interpretation, and the application of intelligent technology means is scarce, and the adaptability and response ability are insufficient in the face of complex detection environment, and the detection result is easily affected by subjective factors such as experience and responsibility of the detection personnel; Thirdly, the detection efficiency is low. Since the detection work highly depends on manual operation, the disease in high-altitude, complex and dangerous scenes cannot be observed closely, not only the detection accuracy is difficult to guarantee, but also the disease information cannot be efficiently collected and standardized recorded, and the overall detection efficiency is low; Fourthly, the operation safety risk is high. Most of the railway housing facilities are distributed in complex structures and dangerous environments such as high-rise buildings and large stations, and when the staff conducts close-range disease observation, they face high safety risks of falling and collision.
[0003] With the rapid development of unmanned aerial vehicle technology and artificial intelligence technology, it has been widely applied in many fields such as electric power, agriculture and transportation, but the existing technology has the following shortcomings: first, no special detection scheme is designed for the structural characteristics and disease types of railway housing facilities (such as station buildings, auxiliary houses and platform canopies); second, there is lack of modeling and positioning means for multi-source data fusion, and the detection accuracy cannot meet the requirements of railway operation and maintenance; third, no report generation and operation transfer mechanism is formed to adapt to the railway industry standard, which cannot directly serve the railway housing facility operation and maintenance management. Therefore, it is urgent to develop an apparent disease detection technology that can realize unmanned and intelligent operation, cover the business closed loop of "inspection-detection-operation", and be specifically adapted to the characteristics of railway housing facilities, so as to improve the detection efficiency and accuracy and reduce the operation risk. SUMMARY
[0004] The present application aims to solve the technical problems of high risk, low efficiency, strong subjectivity and low intelligent degree of the existing artificial detection means of apparent diseases of railway building equipment, and proposes an intelligent inspection method and system for apparent diseases of railway building equipment, which aims to realize unmanned and intelligent detection process through the application of unmanned aerial vehicle technology and artificial intelligence technology, form a complete business closed loop, improve detection efficiency and accuracy, and reduce operation safety risk.
[0005] To solve the above technical problems, the technical scheme adopted by the present application comprises the following steps: Step S1, planning an inspection task; Step S2, outputting a three-dimensional model based on the complete image information of the building equipment to be detected; Step S3, on the basis of the completed three-dimensional model of the room equipment, combining the inspection task requirements, planning an inspection route, and generating a route task; Step S4, downloading and issuing the generated route task to the flight control unit of the unmanned aerial vehicle, controlling the unmanned aerial vehicle to perform an inspection operation according to the route task, completing the collection of building equipment image data, and storing the collected image data in real time to the storage unit of the unmanned aerial vehicle platform; Step S5, transmitting the image data stored in the storage unit of the unmanned aerial vehicle platform to the intelligent detection terminal, and the intelligent detection terminal reads, identifies and classifies the collected image data, and classifies and identifies the disease image data; Step S6, for the image data containing diseases identified by the intelligent detection terminal, a camera photography model is established to measure the physical size parameters of the diseases; Step S7, combining geographic space information, on the one hand, the disease location is positioned in two dimensions through the map, and on the other hand, the disease is three-dimensionally calibrated on the three-dimensional model of the building equipment constructed in the early stage; Step S8, generating an inspection detection report containing image data identification results, disease physical size parameters, map positioning results, three-dimensional calibration results and maintenance suggestions.
[0006] Further, the inspection task in step S1 includes the building equipment object to be inspected, the inspection period, the inspection equipment, the task content and the execution time plan.
[0007] Further, the image information includes one or more of BIM design drawings, field survey data, and image data collected or taken by aerial photography in the early stage.
[0008] Further, the route task in step S3 includes the flight route of the unmanned aerial vehicle, the point to be collected information, and the image collection method and collection quantity of each point.
[0009] Further, the unmanned aerial vehicle is equipped with a high-precision camera module, a distance measuring module and a positioning and navigation module.
[0010] Further, the image data identified by the intelligent detection terminal in step S5 includes disease image data and non-disease image data.
[0011] Further, the disease categories in the disease image data identified in step S5 include at least cracks, peeling and water leakage.
[0012] A system for implementing the above-mentioned intelligent inspection method for apparent diseases of railway housing facilities, comprising: An unmanned aerial vehicle inspection module for performing intelligent inspection tasks, including a task planning unit, a three-dimensional modeling unit, a flight path planning unit, a flight control unit and an information acquisition unit; The task planning unit is used to generate an inspection task; the three-dimensional modeling unit is used to construct a three-dimensional model of the housing facilities; the flight path planning unit is used to generate a flight path task based on the three-dimensional model; the flight control unit is used to control the unmanned aerial vehicle to automatically perform the flight path task; and the information acquisition unit is used to acquire image data. An intelligent detection module for processing the image data acquired during inspection, including an intelligent identification unit, a disease measurement unit, a positioning and calibration unit and a report generation unit. The intelligent identification unit is used to implement classification identification of image data and disease classification identification of disease image data; the disease measurement unit is used to measure the physical size of the disease; the positioning and calibration unit is used to fuse geographic spatial information and three-dimensional modeling data to achieve two-dimensional positioning and three-dimensional calibration of the disease; and the report generation unit is used to automatically generate an inspection detection report in accordance with the railway industry standard; and the data storage and interaction module is used to store inspection task information, three-dimensional model data, acquired image data, detection results and reports, and to realize data interaction between the unmanned aerial vehicle inspection module and the intelligent detection module.
[0013] The present application has the following beneficial effects compared with the prior art: 1. unmanned inspection is realized, the operation risk is reduced, and the inspection efficiency is improved: the unmanned aerial vehicle is used to replace the traditional manual inspection, which can easily cope with complex, dangerous and difficult-to-reach work scenes, and the working personnel need not be close to the dangerous area, which greatly reduces the operation safety risk; at the same time, the unmanned aerial vehicle can automatically complete the inspection task according to the preset flight path, without real-time manual control, which effectively improves the inspection efficiency and shortens the inspection cycle.
[0014] 2. Introduce artificial intelligence technology to improve detection accuracy and reduce subjective dependence: Through artificial intelligence deep learning technology, intelligent classification and identification of diseases are realized, which eliminates the dependence on artificial subjective interpretation and avoids the problems of missed detection and false detection caused by insufficient experience of detection personnel, differences in responsibility, etc., significantly improving the accuracy and consistency of disease detection; at the same time, it can realize the rapid processing of large batches of images, solving the problem of low efficiency of expert interpretation.
[0015] 3. Form a business closed loop to improve operation and management level: The present application integrates intelligent inspection and intelligent detection of two core links, realizes the intelligentization of the whole process from inspection task planning, information collection, disease identification, measurement, positioning, to report generation and operation work order circulation, forms a complete business closed loop of "inspection-detection-operation", provides standardized and standardized technical support for the operation and management of railway housing facilities, and effectively improves the operation efficiency and management level.
[0016] 4. Wide applicability and high promotion value: The detection method proposed by the present application is not only suitable for the apparent disease detection of railway housing facilities, but also can be flexibly adapted to the apparent disease detection scene of other railway facilities such as railway bridges, tunnels and slopes, has wide applicability and high promotion value, and can provide a general technical solution for intelligent detection of the railway industry. BRIEF DESCRIPTION OF DRAWINGS
[0017] The present application will be further described below in conjunction with the drawings: Figure 1 is a structural block diagram of the system of the present application; Figure 2 is a detection result schematic diagram of roof damage of an embodiment of the present application; Figure 3 is a detection result schematic diagram of roof damage of an embodiment of the present application; Figure 4 is a detection result schematic diagram of crack and peeling of an embodiment of the present application Figure 5 is a detection result schematic diagram of crack and peeling of an embodiment of the present application Figure 1 ; Figure 6 is a detection result schematic diagram of crack and peeling of an embodiment of the present application Figure 2 ; Figure 7 is a detection result schematic diagram of crack and peeling of an embodiment of the present application Figure 3 . DETAILED DESCRIPTION
[0018] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate relative or positional relationships, which are relative or positional relationships for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined with "first", "second" and the like can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.
[0019] In the description of the present application, it should be noted that, unless otherwise specified and limited, the terms "mounting", "connection", "connection" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be connected inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.
[0020] As shown in Figures 1 to 7 The present application provides a railway housing construction equipment apparent disease intelligent inspection method, comprising the following steps: Step S1, combined with the railway housing construction equipment account information, determine the priority level of the inspection object, match the differentiated inspection cycle (key facilities once a week, general facilities once a month), clear the load configuration requirements and task content (photograph or video) of the inspection equipment, develop a standardized inspection task containing emergency response mechanism, ensure the pertinence and flexibility of the inspection.
[0021] Step S2, based on the complete image information of the measured housing construction equipment, output a three-dimensional model.
[0022] In one embodiment, the complete image information of the measured housing construction equipment is fused with BIM design drawings, field survey data, image data collected or taken by aerial photography, and a three-dimensional model of the railway housing construction equipment with semantic information is constructed by using multi-source data fusion technology.
[0023] Step S3: Based on the completed 3D model of the building equipment, and in accordance with the inspection task requirements, plan the optimal inspection route, generate detailed route tasks, and clarify the flight path of the UAV, the points where information needs to be collected, and the image acquisition method (such as shooting angle and resolution) and the number of images collected at each point, so as to ensure that the collected images can comprehensively and clearly reflect the surface condition of the building equipment.
[0024] Step S4: Download the generated flight path task and send it to the flight control unit of the UAV. Control the UAV to perform take-off, inspection and return operations according to the flight path task. The whole process does not require real-time manual operation, realizing unmanned inspection. The drone is equipped with a high-precision camera module, a ranging module, and a positioning and navigation module. During the inspection operation, it takes pictures and videos of the target objects and key inspection areas specified in the inspection task to complete the collection of image data of building equipment, and stores the collected image data in real time to the storage unit of the drone platform.
[0025] Step S5: Transmit the image data stored in the storage unit of the UAV platform to an intelligent detection terminal built based on artificial intelligence deep learning technology. The intelligent detection terminal identifies and classifies the collected image data. The image data identified by the intelligent detection terminal includes diseased image data (image data containing disease) and non-diseased image data (image data without disease). It also performs disease classification and identification on the diseased image data, which can identify typical diseases such as cracks, peeling, and water leakage, reducing the reliance on manual judgment and analysis and greatly improving the efficiency of disease identification.
[0026] The intelligent detection terminal has been trained with a large number of railway building equipment defect samples and can automatically identify a variety of common surface defects such as cracks, peeling, water leakage, and corrosion.
[0027] Step S6: For the image data containing the disease identified by the intelligent detection terminal, first use the checkerboard method to complete the camera calibration and image distortion correction, then establish a camera photography model that integrates laser ranging data, construct a precise correspondence between pixel size and actual physical size, and combine artificial intelligence deep learning methods to realize the automatic measurement of physical size parameters such as disease length, width, and area.
[0028] Step S7: Combining geospatial information, on the one hand, the location of the disease is located in two dimensions using a map, and on the other hand, the disease is 3D-marked on the 3D model of the building and equipment constructed in the early stage, so as to intuitively and accurately mark the specific location of the disease, which provides convenience for the later maintenance personnel to quickly locate the disease.
[0029] Step S8: Integrate the intelligent identification results, physical size parameters of the defects, map positioning results, and three-dimensional calibration results, and generate a standardized inspection and testing report according to the preset format. This report can serve as the basis for business work order circulation, realize standardized operation and maintenance task circulation, and facilitate the orderly development of subsequent operation and maintenance and support work.
[0030] The present invention will be further described in detail below with reference to specific embodiments.
[0031] Taking the inspection of the roof of a railway passenger station as an example, the specific steps include: Step S1: For the roof, facade and surrounding ancillary buildings and equipment of a railway passenger station, the inspection cycle is determined to be once a month. The inspection equipment is a DJI M4E RTK drone equipped with a D-RTK2 high-precision positioning module. The inspection content includes surface defects such as cracks, peeling, water leakage, and corrosion. The execution time is planned to be Tuesday to Thursday of the first week of each month.
[0032] Step S2: Obtain multi-angle image data of the passenger station building and equipment through aerial photography in the early stage, and use ContextCapture modeling software to perform three-dimensional modeling to generate a high-precision three-dimensional model of the building and equipment, so as to clearly present the structural layout and key parts of the building and equipment.
[0033] Step S3: Based on the generated 3D model of the building equipment, use UAV ground station software to plan the inspection route. For the roof of the station building, a grid-like route is adopted, with a flight altitude of 5m above the roof and an image overlap of 80%. For the facade, a circular route is adopted, with the flight altitude matching the building height. A collection point is set every 30°, and 3 images of different resolutions are collected at each point.
[0034] Step S4: Download the planned flight path task to the flight control unit of the DJI M4E RTK drone. The flight control unit controls the drone to take off automatically at the specified time, perform the inspection task according to the preset route, and provide real-time feedback on the flight status during the inspection. After the inspection is completed, the drone will automatically return to the take-off and landing point.
[0035] During the inspection process, the drone takes pictures of preset points with its camera and records the latitude, longitude, altitude and other positioning information of each collection point through the RTK block. The collected images and positioning information are stored in the drone's SD card.
[0036] Step S5: After the inspection is completed, the image data stored on the SD card is exported to the intelligent detection terminal. The intelligent detection terminal reads the image data stored on the SD card and performs batch recognition and classification of the collected image data, including disease classification and identification of the disease image data. The recognition results are as follows: There were 3 crack-type defects, with confidence levels of 0.78, 0.88, and 0.85, respectively. The first crack is 73 pixels wide, 67 pixels long, and has an area of 4891 pixels. 2 ; The second crack is 58 pixels wide, 136 pixels long, and has an area of 7888 pixels. 2 ; The third crack is 80 pixels wide, 115 pixels long, and has an area of 9200 pixels. 2 .
[0037] There were 3 instances of falloff disease, with confidence levels of 0.93, 0.92, and 0.96, respectively. The first spall has a width of 94 pixels, a length of 139 pixels, and an area of 13066 pixels. 2 ; The second spall has a width of 183 pixels, a length of 187 pixels, and an area of 34,221 pixels. 2 ; The second spalling point has a width of 17 pixels, a length of 50 pixels, and an area of 850 pixels. 2 .
[0038] There is one roof-related defect with a confidence level of 0.81.
[0039] The width of the damaged roof is 816 pixels, the length is 423 pixels, and the area is 345,168 pixels. 2 .
[0040] Step S6: For the identified damage images, use AIDD software to build a camera photography model. Based on the conversion relationship between pixel size in the model and actual physical size, obtain the actual length, width and area of cracks, peeling and roof damage.
[0041] Step S7: Combine the positioning information collected by the drone to perform two-dimensional positioning of cracks, peeling, and roof damage in Baidu Maps, and mark the specific latitude and longitude of the damage; at the same time, link the location of the damage to the three-dimensional model of the building equipment constructed in the early stage to realize the three-dimensional calibration of the damage and intuitively display the specific location of the damage in the building structure.
[0042] Step S8: Integrate the intelligent recognition results, physical size parameters of the defects, map positioning results, and 3D calibration results. In accordance with the relevant standards for operation and maintenance management of railway building equipment, generate an inspection and testing report that includes image data recognition results, physical size parameters of defects, map positioning results, 3D calibration results, and maintenance suggestions. Push this report to the passenger station operation and maintenance management platform as the basis for operation and maintenance work order circulation, realize standardized operation and maintenance task circulation, and facilitate the orderly development of subsequent operation and maintenance and support work.
[0043] In this embodiment, the comprehensive inspection and testing of the railway passenger station building and equipment was completed in just one day using the above method. A total of 7 apparent defects were identified, and a standardized inspection report was generated. Compared with traditional manual inspection, the efficiency was increased by 80%, while the accuracy of the inspection was improved and the safety risks of workers working at heights were avoided.
[0044] The present invention provides a system for implementing the above-described intelligent inspection method for surface defects in railway building equipment, comprising: The UAV inspection module is used to perform intelligent inspection tasks, including a task planning unit, a 3D modeling unit, a flight path planning unit, a flight control unit, and an information acquisition unit. The task planning unit is used to generate inspection tasks; the 3D modeling unit is used to construct 3D models of building equipment; the flight path planning unit is used to generate flight path tasks based on the 3D models; the flight control unit is used to control the UAV to automatically execute flight path tasks; and the information acquisition unit is used to acquire image data. The intelligent detection module is used to process the image data collected during inspections, including an intelligent recognition unit, a disease measurement unit, a positioning and calibration unit, and a report generation unit. The intelligent recognition unit is used to classify and recognize image data and disease image data; the disease measurement unit is used to measure the physical size of the disease; the positioning and calibration unit is used to integrate geospatial information and 3D modeling data to achieve 2D positioning and 3D calibration of the disease; the report generation unit is used to automatically generate inspection and testing reports that conform to railway industry standards; the data storage and interaction module is used to store inspection task information, 3D model data, collected image data, test results and reports, and at the same time realizes data interaction between the UAV inspection module and the intelligent detection module.
[0045] Regarding the specific structure of this invention, it should be noted that the connection relationships between the various component modules used in this invention are definite and achievable. Except as specifically described in the embodiments, their specific connection relationships can bring about corresponding technical effects and solve the technical problems proposed by this invention without relying on the execution of corresponding software programs. The models of the components, modules, and specific components appearing in this invention, the connection methods between them, and the conventional usage methods and expected technical effects brought about by the above technical features, unless specifically described, are all publicly disclosed content in patents, journal articles, technical manuals, technical dictionaries, and textbooks that can be obtained by those skilled in the art before the application date, or belong to conventional technology, common knowledge, and other existing technologies in this field. There is no need to elaborate, which makes the technical solution provided in this case clear, complete, and achievable, and can reproduce or obtain corresponding physical products based on this technical means.
[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. An intelligent inspection method for surface defects in railway building equipment, characterized in that: Includes the following steps: Step S1: Plan the inspection tasks; Step S2: Output a 3D model based on the complete image information of the building equipment to be tested; Step S3: Based on the completed 3D model of the room equipment, and in accordance with the inspection task requirements, plan the inspection route and generate the route task. Step S4: Download the generated flight path task and send it to the flight control unit of the UAV. Control the UAV to perform inspection operations according to the flight path task, complete the collection of building equipment image data, and store the collected image data in real time to the storage unit of the UAV platform. Step S5: Transmit the image data stored in the storage unit of the UAV platform to the intelligent detection terminal. The intelligent detection terminal reads, identifies and classifies the collected image data, and classifies and identifies the disease images. Step S6: Based on the image data containing the disease identified by the intelligent detection terminal, establish a camera photography model to measure the physical size parameters of the disease; Step S7: Combining geospatial information, on the one hand, the location of the disease is located in two dimensions using a map, and on the other hand, the disease is calibrated in three dimensions on the three-dimensional model of the building and equipment constructed in the early stage. Step S8: Generate an inspection report that includes image data recognition results, physical size parameters of the disease, map positioning results, three-dimensional calibration results, and maintenance suggestions.
2. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: The inspection task in step S1 includes the building equipment to be inspected, the inspection cycle, the inspection equipment, the task content, and the execution time plan.
3. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: The image information includes one or more of the following: BIM design drawings, on-site survey data, and image data collected in advance or taken by aerial photography.
4. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: Step S3 includes the flight path of the UAV, the locations where information needs to be collected, and the image collection method and quantity for each location.
5. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: The drone is equipped with a high-precision camera module, a ranging module, and a positioning and navigation module.
6. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: The image data identified by the intelligent detection terminal in step S5 includes diseased image data and non-diseased image data.
7. The intelligent inspection method for surface defects of railway building equipment according to claim 1, characterized in that: The disease categories identified in step S5 include at least cracks, peeling, and water leakage in the disease image data.
8. A system for implementing the intelligent inspection method for surface defects of railway building equipment as described in any one of claims 1-7, characterized in that, include: The UAV inspection module is used to perform intelligent inspection tasks, including a task planning unit, a 3D modeling unit, a flight path planning unit, a flight control unit, and an information acquisition unit. The task planning unit is used to generate inspection tasks; The 3D modeling unit is used to construct 3D models of building equipment; the flight route planning unit is used to generate flight route tasks based on the 3D model. The flight control unit is used to control the UAV to automatically execute flight path tasks; The information acquisition unit is used to acquire image data; The intelligent detection module is used to process the image data collected during inspections, including an intelligent recognition unit, a disease measurement unit, a positioning and calibration unit, and a report generation unit. The intelligent recognition unit is used to classify and recognize image data and disease image data; the disease measurement unit is used to measure the physical size of the disease; the positioning and calibration unit is used to integrate geospatial information and 3D modeling data to achieve 2D positioning and 3D calibration of the disease; the report generation unit is used to automatically generate inspection and testing reports that conform to railway industry standards; the data storage and interaction module is used to store inspection task information, 3D model data, collected image data, test results and reports, and at the same time realizes data interaction between the UAV inspection module and the intelligent detection module.