This article delves deep into the application of machine vision technology in detecting sand mould defects within hydrostatic automatic molding lines. It comprehensively explores the background, significance, system architecture, implementation process, and practical effects of this technology. By presenting in – depth analysis and real – world examples, it aims to provide a thorough understanding of how machine vision can enhance the quality control in the foundry industry, ultimately contributing to improved casting quality and increased production efficiency.
1. Introduction
In the foundry industry, the quality of sand moulds plays a crucial role in determining the quality of castings. Defective sand moulds can lead to a variety of casting defects such as excess material, sand adhesion, and sand holes. Traditionally, manual inspection methods have been widely used to identify sand mould defects. However, these methods are labor – intensive, time – consuming, and prone to human errors. With the continuous advancement of technology, machine vision technology has emerged as a promising solution for non – destructive online detection of sand mould defects. This technology can not only improve the detection accuracy and efficiency but also enable real – time monitoring and quality control in the production process.
1.1 The Imperative of Sand Mould Defect Detection
The increasing demand for high – precision castings in various industries, such as automotive, aerospace, and machinery manufacturing, has raised the bar for sand mould quality. Even minor defects in sand moulds can cause significant problems in the casting process, resulting in product rejects, increased production costs, and potential safety hazards. Therefore, effective sand mould defect detection is essential for ensuring the overall quality of castings and maintaining the competitiveness of foundry enterprises.
1.2 Limitations of Traditional Detection Methods
Manual inspection of sand moulds relies on the experience and visual acuity of workers. Workers need to carefully observe the surface of sand moulds to identify potential defects. This method is highly subjective, as different workers may have different standards for defect judgment. Moreover, the inspection speed is limited, and it is difficult to achieve 100% inspection in high – volume production environments. In addition, manual inspection cannot provide real – time data for process control, making it difficult to timely adjust the production process to prevent the occurrence of defects.
2. Machine Vision Technology: An Overview
Machine vision technology is a multidisciplinary field that combines computer science, optics, electronics, and image processing. It uses cameras and imaging devices to capture images of objects, and then processes and analyzes these images through algorithms to obtain information about the object’s shape, size, color, and texture. In the context of sand mould defect detection, machine vision technology can quickly and accurately identify various defects on the surface of sand moulds.
2.1 Key Components of Machine Vision Systems
| Component | Function |
|---|---|
| Cameras | Capture images of sand moulds. High – resolution cameras are often used to ensure clear image quality for defect detection. |
| Lighting Systems | Provide uniform and stable illumination to enhance the contrast of the images and make defects more visible. |
| Image Processing Software | Analyze the captured images, perform operations such as image enhancement, segmentation, and feature extraction to identify defects. |
| Hardware Platforms | Support the operation of the image processing software, including computers or dedicated image processing units with sufficient computing power. |
2.2 Image Processing Algorithms for Defect Detection
There are various image processing algorithms used in sand mould defect detection. For example, edge detection algorithms can be used to detect the boundaries of defects, and thresholding algorithms can separate defect areas from the background. In addition, more advanced deep learning – based algorithms, such as convolutional neural networks (CNNs), have shown excellent performance in defect detection. CNNs can automatically learn the features of defects from a large number of training images, improving the detection accuracy and generalization ability.
3. Design and Implementation of Sand Mould Defect Detection System
3.1 System Architecture
The sand mould defect detection system based on machine vision typically consists of an image acquisition module, an image processing module, a defect classification module, and a result output module.
| Module | Function |
|---|---|
| Image Acquisition Module | Comprises cameras and lighting devices. Cameras are installed above the sand moulds in the hydrostatic automatic molding line to capture images at specific positions and times. The lighting system ensures proper illumination for clear image capture. |
| Image Processing Module | Pre – processes the captured images, including tasks like noise reduction, image enhancement, and geometric correction. Then, it extracts features related to defects for further analysis. |
| Defect Classification Module | Uses trained models, such as CNN – based models, to classify the detected defects into different types, such as sand lumps, adhesion, and cracks. |
| Result Output Module | Displays the detection results, including the location, type, and severity of defects. It can also output data for production management, such as defect statistics and quality reports. |
3.2 Selection of Detection Equipment
- Cameras: High – resolution industrial cameras with appropriate focal lengths and field – of – view are selected. For example, cameras with a resolution of 5 million pixels or more can capture detailed images of sand mould surfaces, enabling the detection of small – scale defects.
- Lighting: LED lights are commonly used due to their stable light output, long lifespan, and adjustable color temperature. Backlighting or side – lighting methods can be chosen according to the characteristics of sand mould defects to enhance the visibility of defects.
3.3 Software Development
- Image Processing Software: Custom – developed software or commercial image processing platforms can be used. The software should be able to perform a series of image processing operations, such as filtering, segmentation, and feature extraction. For example, the OpenCV library provides a wide range of functions for image processing, which can be integrated into the software development process.
- Deep Learning – Based Defect Classification: When using deep learning algorithms, a large number of labeled sand mould images are required for training. The training process involves adjusting the parameters of the neural network to minimize the error between the predicted and actual defect labels. Popular deep – learning frameworks like TensorFlow or PyTorch can be used for model training and deployment.
4. 静压自动造型线改造 for Machine Vision – Based Detection
4.1 Data Interaction
In order to integrate the sand mould defect detection system with the hydrostatic automatic molding line, data interaction is crucial. The Snap7 open – source communication library can be used to establish communication between the detection system and the PLC (Programmable Logic Controller) of the molding line.
| PLC Data Block | Function |
|---|---|
| DB2006 | Used for the sand mould defect detection system to read data. It contains information such as the current sand mould attributes, production data of the molding line, detection control signals, and equipment heartbeat signals. |
| DB2007 | Used for the sand mould defect detection system to write data. It includes feedback on the system status, detection results, and equipment heartbeat signals. |
4.2 Control Implementation
The control process of the sand mould defect detection system in the hydrostatic automatic molding line is as follows:
| Step | Operation |
|---|---|
| 1 | When the sand mould defect detection function is enabled, after each sand box is pushed in the molding section, the system queries the information of the specified sand box. |
| 2 | It judges the status of the sand box. If the sand box is in good condition and not empty, the system locks the molding section and starts the sand mould defect detection. Otherwise, it skips the detection. |
| 3 | After the detection is completed, according to the results returned by the sand mould defect detection system, if the sand mould is defective, it needs to be marked as a defective type and then the molding section is unlocked. If the sand mould is good, the molding section is directly unlocked. |
5. Practical Application and Results
5.1 Deployment in Foundry Production Lines
The sand mould defect detection system based on machine vision has been successfully deployed in multiple hydrostatic automatic molding lines in enterprises, such as the 4 KW and 4 HWS molding lines in Weichai (Weifang) Material Forming & Manufacturing Center Co., Ltd. The system has been operating stably and has covered a wide range of casting products, including sand moulds for engine blocks, cylinder heads, crankcases, flywheels, hydraulic valves, gear chambers, and rear covers.
5.2 Detection Accuracy and Efficiency
Through continuous optimization of the system and training of the model, the detection accuracy and efficiency of the sand mould defect detection system have been significantly improved. The system can detect various types of defects with high accuracy, and the detection speed can meet the requirements of high – volume production. For example, the detection accuracy of common defects such as sand lumps and adhesion can reach over 95%, and the detection time for each sand mould is within a few seconds.
5.3 Impact on Production Quality and Cost
The application of the sand mould defect detection system has had a positive impact on production quality and cost. By detecting and removing defective sand moulds in a timely manner, the occurrence of casting defects has been effectively reduced, resulting in an improvement in the overall quality of castings. At the same time, the reduction of defective products has also led to a decrease in production costs, including the cost of raw materials, rework, and waste disposal.
| Indicator | Before System Implementation | After System Implementation |
|---|---|---|
| Casting Defect Rate | High (e.g., 10 – 15%) | Low (e.g., 3 – 5%) |
| Production Cost per Unit | High | Reduced by approximately 10 – 20% |
6. Challenges and Future Outlook
6.1 Challenges in Machine Vision – Based Detection
- Complex Defect Patterns: Sand mould defects can have complex shapes and patterns, making it difficult for the detection system to accurately identify and classify all types of defects. Some defects may be similar in appearance but have different causes and impacts on casting quality.
- Illumination Variations: In the production environment, changes in lighting conditions, such as fluctuations in ambient light or dirt on the lighting equipment, can affect the quality of the captured images and lead to inaccurate defect detection.
- Data Overfitting: When using deep – learning algorithms, there is a risk of overfitting, especially when the training data is limited or not representative enough. Overfitting can cause the model to perform well on the training data but poorly on new, unseen data.
6.2 Future Research Directions
- Advanced Defect Classification Algorithms: Research on more advanced defect classification algorithms, such as combining multiple deep – learning models or using transfer learning techniques, to improve the ability to handle complex defect patterns.
- Adaptive Illumination Control: Develop adaptive lighting control systems that can automatically adjust the lighting intensity and color according to the production environment, ensuring stable and reliable image capture for defect detection.
- Big Data and Cloud – Based Solutions: Leverage big data and cloud computing technologies to store and analyze a large amount of sand mould defect data. This can help in continuous improvement of the detection system, as well as in predicting potential defects and optimizing the production process.
7. Conclusion
The application of machine vision technology in the detection of sand mould defects in hydrostatic automatic molding lines represents a significant advancement in the foundry industry. It has effectively addressed the limitations of traditional manual inspection methods, improving the accuracy and efficiency of defect detection, and enhancing the overall quality of castings. Although there are still challenges to be overcome, with continuous research and technological innovation, machine vision – based sand mould defect detection systems will become more intelligent, accurate, and reliable, contributing to the development and competitiveness of the foundry industry in the future.
