Automobile part X-ray defect intelligent detection system based on 5G industrial internet

Abstract

The invention discloses an intelligent X-ray defect detection system for automobile parts based on the 5G industrial internet, and belongs to the crossing field of intelligent manufacturing and the industrial internet. The hardware part comprises a six-axis collaborative robot, a sorting manipulator, an X-ray imaging module and a high-precision rotary conveying table. The software part comprises a microfocus X-ray imaging system, a 5G edge computing gateway and a cloud defect analysis platform; the six-axis robot cooperates with the high-precision rotary conveying table to achieve 360-degree dead-corner-free scanning, high-definition imaging of the internal structure of each casting is ensured, and real-time data is uploaded to the cloud defect analysis platform through the 5G edge computing gateway for defect judgment; and the automatic sorting manipulator performs efficient sorting according to an identification result. According to the invention, X-ray imaging, 5G communication, machine vision and robot control technologies are integrated, and multi-stage data processing and intelligent algorithm fusion are carried out, so that the system significantly improves the detection efficiency and precision, effectively reduces the labor cost, guarantees the product quality, and promotes the intelligent upgrading of the automobile manufacturing industry.

Description

Technical Field

[0001] The present invention belongs to the intersection of intelligent manufacturing and industrial Internet, and specifically relates to an intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet, which is particularly suitable for fully automatic quality inspection of key castings such as automobile gearbox housings and steering knuckles. Background Art

[0002] In the automotive industry, lightweight castings made of aluminum and magnesium alloys are widely used in key components such as engine blocks, transmission housings, and chassis structures. Due to the complexity of the casting process, defects such as pores, shrinkage, inclusions, and cracks are prone to forming within the castings, directly impacting the mechanical performance and safety of the components. Therefore, X-ray defect detection technology is often used during the production process to ensure product quality control.

[0003] Traditional X-ray defect detection technology primarily relies on manual image evaluation, which presents the following technical bottlenecks: ① Inefficient detection: Traditional X-ray inspection equipment utilizes fixed-station, single-angle imaging, requiring manual adjustment of the casting's orientation. This results in slow inspection speeds (typically 3-5 minutes per piece), making it incompatible with automated production line cycles. ② Low intelligence: Defect identification relies on manual experience and lacks standardized algorithm support. This leads to a high miss rate (approximately 15%-20%) for minor defects, such as pores <0.1mm, and the inability to correlate inspection results with process parameters in real time. ③ Poor data interoperability: Inspection data is stored in isolation on local industrial computers and is not integrated with MES (Manufacturing Execution Systems) and QMS (Quality Management Systems). This results in coarse-grained quality traceability and delayed rectification. ④ Network transmission latency: Existing industrial Wi-Fi or wired networks cannot meet the real-time transmission requirements of high-definition X-ray images (single image >50MB), resulting in high latency fluctuations (>100ms), impacting the real-time performance of cloud-based collaborative analysis.

[0004] Therefore, the traditional X-ray defect detection technology has the above-mentioned technical problems such as low efficiency, high missed detection rate, and difficulty in traceability, which makes it difficult to meet the needs of modern intelligent manufacturing for high precision, high efficiency and full-process quality control. Summary of the Invention

[0005] The purpose of this invention is to propose an intelligent X-ray defect detection system for automotive parts based on the 5G Industrial Internet, which integrates X-ray imaging, 5G communication, machine vision and robot control technologies. Through the integration of multi-level data processing and intelligent algorithms, the system significantly improves detection efficiency and accuracy, effectively reduces labor costs, ensures product quality, and promotes the intelligent upgrade of the automotive manufacturing industry.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] An intelligent X-ray defect detection system for automotive parts based on the 5G Industrial Internet includes a hardware part and a supporting software part. The hardware part includes a six-axis collaborative robot, a sorting manipulator, an X-ray imaging module and a high-precision rotary conveyor table; the software part is integrated in the software control area, including a micro-focus X-ray imaging system, a 5G edge computing gateway, and a cloud defect analysis platform; the six-axis robot is equipped with an X-ray imaging module, and cooperates with the high-precision rotary conveyor table to achieve 360° no-dead-angle scanning, ensuring high-definition imaging of the internal structure of each casting, and uploading real-time data to the cloud defect analysis platform through the 5G edge computing gateway for defect judgment; the automatic sorting manipulator performs efficient sorting based on the recognition results.

[0008] This invention integrates multi-angle 3D imaging of castings, millisecond-level cloud storage, deep learning defect recognition, and automated sorting through a three-tiered "end-edge-cloud" architecture. The system utilizes URLLC (Ultra-Reliable Low Latency Communication) network slicing technology to ensure transmission latency of less than 2ms. The ADR (Automatic Defect Recognition) system, based on an improved YOLOv5 algorithm, achieves a detection accuracy of 99.6%, significantly improving the rejection rate by 300% compared to traditional methods and significantly improving the rejection rate compared to manual film inspection and evaluation. Compared to existing technologies, this invention offers the following benefits:

[0009] 1. This invention innovatively achieves efficient coordination of detection, analysis, and storage functions. Through the high-speed, low-latency characteristics of the 5G network, it ensures the real-time and reliability of image data upload to the cloud, effectively solving the problems of insufficient bandwidth and data loss that may exist in traditional storage methods.

[0010] 2. The cloud defect analysis platform uses advanced machine learning algorithms to accurately identify various defect characteristics and establish a comprehensive quality database, providing strong support for product quality traceability and improvement. The collaborative operation of the six-axis robot and the sorting manipulator significantly improves the automation level of the production line, reduces manual intervention, and significantly reduces labor costs and production errors.

[0011] 3. The overall system design fully considers the actual needs of industrial scenarios, ensures operational flexibility in hardware configuration, and realizes intelligent analysis in software functions. This makes the solution not only have excellent detection performance, but also have good scalability and adaptability, and can meet the quality inspection needs of industrial production of different scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A schematic diagram of the on-site hardware assembly structure of the intelligent X-ray defect detection system for automotive parts based on the 5G Industrial Internet provided in an embodiment of the present invention, which details the physical layout and connection relationship of the main functional modules of the system.

[0013] Figure 2 for Figure 1 The top view of the hardware assembly diagram clearly shows the relative positions and spatial distribution of the various components from a top perspective.

[0014] Figure 3 for Figure 1 、 2 The hardware assembly diagram shown is a schematic diagram of the six-axis collaborative robot, focusing on the robot's joint structure and degrees of freedom of movement.

[0015] Figure 4 for Figure 1 、 2 The structural diagram of the software control area in the hardware assembly diagram shown above systematically demonstrates the functional division and operation logic of the control interface.

[0016] Figure 5 for Figure 1 、 2 The hardware assembly diagram shows the structure of the sorting robot and the schematic diagram of the working area, which details the gripping mechanism and execution components of the robot and demonstrates the working range and operation details of the sorting area.

[0017] Figure 6 This is a network architecture diagram of the terminal layer, edge layer, cloud layer, and cloud data center.

[0018] Figure 7 This is a system network architecture diagram, based on a hybrid model of local deployment + cloud collaboration, covering the entire link from terminal devices to cloud data centers.

[0019] Figure 1 、 2 The meanings of the accompanying symbols are:

[0020] 1. Six-axis collaborative robot; 2. Sorting robot; 3. X-ray imaging module; 4. High-precision rotary conveyor; 5. Software control area. DETAILED DESCRIPTION

[0021] The present invention is further described below in conjunction with embodiments and drawings.

[0022] See also Figure 1-5 As shown, the present invention provides an intelligent X-ray defect detection system for automotive parts based on the 5G Industrial Internet. The system mainly includes hardware and supporting software. The hardware includes a six-axis collaborative robot 1, a sorting manipulator 2, an X-ray imaging module 3, and a high-precision rotary conveyor 4. The software includes a microfocus X-ray imaging system, a 5G edge computing gateway, a cloud defect analysis platform, etc., which are integrated into the software control area 5. The composition and function of each component are described in detail below:

[0023] Six-axis collaborative robot 1: A KUKA KR6 R900 robot is used. The X-ray imaging module 3 on the top of its arm consists of a microfocus X-ray imaging system (200kV / 5μm focus) and a high-resolution flat-panel detector (3072×3072 pixels). Figure 3 (as shown), it enables 360° scanning of castings without blind spots, ensuring high-definition imaging of internal structures and real-time data upload to the 5G edge computing gateway. The control software for the X-ray imaging module 3 is integrated into the software control area 5. The six-axis robot 1 has six degrees of freedom, enabling precise, all-round, multi-angle scanning of products through multi-axis coordinated rotation.

[0024] High-precision rotary conveyor 4: Using servo drive technology, it can continuously and stably automatically convey the product to be inspected to the detection position of the X-ray imaging module 3. Its repeated positioning accuracy is ±0.005mm. It cooperates with the robot to achieve automatic multi-angle positioning of the casting and optimize scanning efficiency.

[0025] 5G Edge Computing Gateway: Supports the TSN (Time-Sensitive Network) protocol and achieves millisecond-level data transmission and processing through 5G+ Industrial Internet technology. It has a built-in AI acceleration chip to optimize image preprocessing and feature extraction.

[0026] Cloud defect analysis platform: deployed as the core algorithm module in software control area 5, and 5G cloud storage program as the data management module is also installed in software control area 5. The cloud defect analysis platform integrates deep learning algorithms, 3D reconstruction modules and blockchain quality traceability functions to monitor the quality of castings throughout the entire process, realize intelligent defect identification and reliable data storage, and ensure data security and traceability ( Figure 4 shown).

[0027] Automatic sorting robot 2: By receiving the defective product information sent in real time by the cloud defect analysis platform, it accurately grasps and performs efficient sorting on the identified defective products (with defects such as pores, looseness, cracks, etc.), with a sorting error rate of less than 0.5% ( Figure 5 shown).

[0028] This invention provides an intelligent X-ray defect inspection system for automotive parts based on the 5G Industrial Internet. This system builds a three-level "end-edge-cloud" collaborative system, integrating a six-axis robot, a high-precision turntable, an X-ray imaging module, a 5G edge computing gateway, and a cloud defect analysis platform to achieve a fully automated closed-loop process from scanning to sorting. The system operates as follows: A six-axis robot (KUKA KR6 R900), equipped with a microfocus X-ray imaging system (200kV / 5μm focal spot) and a high-resolution flat-panel detector (3072×3072 pixels), collaborates with a high-precision turntable to achieve 360° scanning without blind spots, ensuring high-definition imaging of the internal structure of each casting. The data is uploaded to the 5G edge computing gateway in real time. The servo-driven turntable (with a repeatability of ±0.005mm) enables an automatic sorting robot to efficiently sort parts based on the recognition results. The overall system response time is less than 1 second, effectively improving production efficiency and product quality, reducing labor costs, and realizing an intelligent and automated production process. At the same time, the cloud defect analysis platform, including a blockchain quality traceability module, monitors the quality of castings throughout the entire process to ensure data security and traceability.

[0029] In one embodiment, the cloud defect analysis platform includes: ① an intelligent defect recognition module that uses a convolutional neural network to automatically detect defects in X-ray images; ② a three-dimensional reconstruction module that generates a spatial distribution model of internal defects in castings; and ③ a blockchain evidence storage unit that records detection data, sorting results, and operation logs, supporting tamper-proof traceability.

[0030] Among them, the intelligent defect recognition module has the following features: ① Improved YOLOv5's ADR system: integrating local binary patterns (LBP), deep convolution features and 3D point cloud registration, with a detection accuracy of 99.6% (an increase of 300% compared to traditional methods); ② Adaptive threshold algorithm: δ=α×σ+β×μ (σ is the variance, μ is the mean), which can dynamically adjust the recognition accuracy and combine transfer learning to improve the model's generalization ability to ensure accurate recognition of complex defects.

[0031] Among them, the blockchain evidence storage unit realizes the following functions: ①Associating full life cycle inspection data according to product QR code; ②Generte digital certificates for third parties to verify product quality compliance; ③Automatically trigger quality warnings and push rectification suggestions when defect patterns match historical cases.

[0032] In one embodiment, the automatic sorting robot is equipped with a defect classification unit that identifies the types of defects such as pores, porosity, and cracks based on a deep learning algorithm, and a sorting execution unit that sorts the castings into qualified products, reworked products, or scrap areas according to the classification results.

[0033] In one embodiment, the 5G edge computing gateway implements the following functions: real-time transmission of X-ray scanning image data, synchronous control of the motion instructions of the six-axis robot, turntable and sorting robot, network delay compensation, and ensuring that the system response time is less than 1 second.

[0034] In one implementation, 5G transmission signals utilize: ① Dynamic dual-channel transmission: Raw data (12Gbps) and feature data (300Mbps) are transmitted separately; ② Dynamic coding strategy: Rateless coding is enabled when RSSI is less than -90dBm, ensuring stable and efficient data transmission. Furthermore, the edge computing gateway has a built-in AI accelerator chip that supports real-time image preprocessing and feature extraction, optimizing network load and increasing data processing speed.

[0035] In one embodiment, see Figure 6 As shown, the micro-focus X-ray imaging system adopts advanced imaging technology to obtain X-ray photos of the product in different directions by scanning the product at multiple angles, ensuring the comprehensiveness of the inspection; through the enterprise's dedicated 5G network, the generated high-definition X-ray images are uploaded to the cloud server in real time for synchronous storage to ensure data security; the cloud defect analysis platform uses deep learning algorithms to intelligently analyze the images generated by the micro-focus X-ray imaging system, accurately identify and classify product defects, and transmit the defect information in real time to the sorting robot to perform automatic sorting operations.

[0036] In one embodiment, the intelligent X-ray defect detection system for automotive parts based on the 5G industrial Internet also includes: 1. a process parameter adaptive module that dynamically adjusts the casting process parameter thresholds based on defect statistics; 2. an audible and visual alarm device that triggers on-site warnings when critical defects are detected; and 3. a mobile terminal interface that supports quality inspectors in remotely reviewing defect images and making sorting decisions.

[0037] In one embodiment, see Figure 7 As shown, a six-axis robot and a servo-driven rotary table collaborate to perform multi-angle X-ray scanning of castings. Edge computing nodes process scanned images in real time to extract defect features. Data is synchronized to a cloud analysis platform via a 5G industrial gateway for intelligent judgment. Based on the judgment results, an automatic sorting robot is controlled to perform sorting actions. The inspection data and sorting results are written to a blockchain database to generate a quality traceability chain. The intelligent judgment steps include comparing defect size with process standard thresholds, matching historical defect pattern knowledge graphs, and automatically generating rework process parameters when the defect is repairable and distributing them to the production system.

[0038] In one embodiment, the system can be seamlessly integrated with existing production lines, receive production data in real time through an API interface, and dynamically adjust scanning strategies to ensure that detection accuracy and production efficiency are improved simultaneously, realizing intelligent and automated quality management.

[0039] In one implementation, each process achieves real-time data synchronization via a 5G network, ensuring seamless information flow. Edge computing nodes rapidly preprocess image data, while a cloud platform performs deep learning analysis, generating defect reports that are fed back to the production line. Robots then precisely sort according to instructions, creating a closed-loop quality control system that improves overall production efficiency and product quality.

[0040] In one implementation, the robot monitors sorting accuracy in real time during the automated sorting process, dynamically adjusting its parameters to ensure accurate sorting. The system records data from each sorting operation, which is then used to continuously optimize the algorithm, improve sorting efficiency and accuracy, and further enhance the stability and reliability of the quality control system.

[0041] In one implementation, at the start of the casting process, the casting operator inputs the casting model and batch information via a terminal device. The system automatically matches the optimal scanning parameters to maximize inspection accuracy and efficiency. The operator monitors the scanning process in real time and intervenes promptly if any system anomalies occur, ensuring production continuity and safety.

[0042] In one implementation, a human viewer remotely views scanned images via a specialized terminal, annotates defect locations in real time, and verifies AI-powered recognition results to ensure accurate defect identification. The system records the viewer's operational data and uses it to optimize the AI ​​algorithm, improving overall inspection accuracy and efficiency. The viewer and AI work together to provide dual assurance for defect identification, ensuring that every casting meets quality standards. The system updates the defect database in real time, continuously optimizes the algorithm model, and improves inspection accuracy, helping companies achieve intelligent production management and ensuring stable and reliable product quality.

Claims

1. An intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet, characterized in that: It includes hardware and supporting software. The hardware part includes a six-axis collaborative robot, a sorting robot, an X-ray imaging module and a high-precision rotary conveyor table; the software part is integrated in the software control area, including a micro-focus X-ray imaging system, a 5G edge computing gateway, and a cloud defect analysis platform; the six-axis robot is equipped with an X-ray imaging module, and cooperates with the high-precision rotary conveyor table to achieve 360° no-dead-angle scanning, ensuring high-definition imaging of the internal structure of each casting, and uploading real-time data to the cloud defect analysis platform through the 5G edge computing gateway for defect judgment; the automatic sorting robot performs efficient sorting based on the recognition results.

2. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 1 is characterized in that: The top of the six-axis collaborative robot's arm is equipped with an X-ray imaging module, which consists of a microfocus X-ray imaging system (200kV / 5μm focus) and a high-resolution flat-panel detector (3072×3072 pixels).

3. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 2 is characterized in that: The six-axis collaborative robot has six degrees of freedom of movement and can perform all-round and multi-angle precise scanning of products through multi-axis coordinated rotation.

4. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 1 is characterized in that: The high-precision rotary conveyor table continuously and stably transports the products to be inspected to the detection position of the X-ray imaging module, and cooperates with the robot to achieve automatic multi-angle positioning of the casting and optimize scanning efficiency.

5. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 1 is characterized in that: The 5G edge computing gateway supports the TSN (time-sensitive network) protocol, achieves millisecond-level data transmission and processing through 5G+ industrial Internet technology, and has a built-in AI acceleration chip to optimize image preprocessing and feature extraction.

6. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 1, characterized in that: The cloud defect analysis platform monitors the quality of castings throughout the entire process, enabling intelligent defect identification and reliable data storage, ensuring data security and traceability. It includes: Intelligent defect recognition module, which uses convolutional neural networks to automatically detect defects in X-ray images; 3D reconstruction module to generate the spatial distribution model of internal defects in castings; The blockchain evidence storage unit records test data, sorting results and operation logs, and supports tamper-proof traceability.

7. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 6, characterized in that: The defect intelligent identification module has the following features: Improve YOLOv5's ADR system: integrating local binary patterns (LBP), deep convolutional features, and 3D point cloud registration; Adaptive threshold algorithm: δ = α × σ + β × μ (σ is the variance, μ is the mean), which can dynamically adjust recognition accuracy and combine with transfer learning to improve model generalization ability, ensuring accurate identification of complex defects.

8. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 6, characterized in that: The blockchain evidence storage unit implements the following functions: Associate the full life cycle testing data by product QR code; Generate digital certificates for third-party verification of product quality compliance; Automatically trigger quality warnings and push correction suggestions when defect patterns match historical cases.

9. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet as claimed in claim 1, characterized in that: The automatic sorting robot receives defective product information sent in real time by the cloud defect analysis platform, and accurately grasps and performs efficient sorting on the identified defective products.

10. The intelligent X-ray defect detection system for automotive parts based on 5G industrial Internet according to any one of claims 1 to 9, characterized in that: Also includes: Process parameter adaptive module dynamically adjusts casting process parameter thresholds based on defect statistics; Sound and light alarm device, triggering on-site warning when critical defects are detected; The mobile terminal interface supports quality inspectors to remotely review defect images and make sorting decisions.

Images