Research on On-line Inspection System of Stamping Parts Based on Sensors
At present, the stamping parts used in air conditioners have been transformed from backward manual operations to unit-type automatic line production. Although automated production has been achieved and the goal of reducing staff and increasing efficiency has been achieved, the automation line has been caused by abnormal operations such as personnel misoperation and equipment failure. It’s not uncommon for the body to miss procedures in the process of producing composite parts. According to the analysis of the defective part structure and the working principle of the equipment by the personnel at the production site, the automated production line is improved by means of mechanical sensors, visual inspection technology, etc., to realize the real-time intelligent detection of the assembly during the production process, and the intelligent inspection equipment and the automated line program Linkage and interlocking can realize automatic shutdown and alarm of production equipment while identifying part defects, preventing abnormal parts from flowing out. This article takes the chassis components of the air conditioner as an example, and focuses on the application of the sensor to realize the online detection function on the automation line.
Air conditioner chassis component parts introduction
The air conditioner chassis assembly is formed by riveting 1 chassis, 2 footings (16 soldering points), 2 stoppers (4 soldering points), and 3 bolts (3 soldering points), as shown in Figure 1. The chassis component is a component of the external unit of the air conditioner. It is a key structural part of the external unit of the household air conditioner. It mainly plays an important role in fixing the compressor, condenser and after-sales installation (Figure 2). The component is strictly prohibited from leaking process quality abnormal.
Application of sensor detection in automatic riveting production line
Figure 1 Structure diagram of chassis components
Figure 2 Use of the chassis
In the production process, the chassis components use TOX riveting automatic production line to complete the stop riveting, bolt spot welding and footing riveting processes. However, when the individual parts are manually discharged into the automatic production line, there will be leakage. At the same time, the parts are not placed in place due to equipment vibration and tool wear during the riveting process, and the parts fall during the operation of the line, which directly leads to the leakage of the parts. The quality of the process is abnormal, and the personnel of placing and loading the vehicle are not in place to make the abnormal parts flow into the next process.
After analyzing the part structure and simulation matching, the team adopted a mechanical detection device (Figure 3) based on its characteristics, namely a contact pressure sensing device, to detect various parts of the chassis components, and used pressure sensors for modeling and design. The model design refers to the chassis The component shape is made of profiling tooling, and the corresponding number of springs and pressure sensors are added to the corresponding position of each part of the chassis component. After the chassis component is detected, the signal will be fed back to the robot for logical calculation. If there is an abnormality in the missing process, the robot will The chassis components are removed and the inspection of the chassis components is completed.
Application of sensor vision inspection in spraying production
The riveted chassis components need to be sprayed, and the parts should be hung on a special hanger for cleaning during the spraying process. After the chassis is hung on the hanger, the contact method is point contact. The parts are free to shake and rotate at 180°. The quality of spraying is not easy to control. Special personnel must be arranged for online inspection. This not only causes waste of personnel, but also has serious defects and missed inspections. Happening. Traditional visual inspection is caused by false alarms due to chassis tilt detection, no alarms when hanging over the hanger, false alarms when hanging chassis of different sizes at the same time, and online inspection of spray lines is difficult, and it is not easy to implement using traditional detection methods.
Figure 4 Online photoelectric visual inspection of spray coating
After the team conducted on-site investigation and comparison of the abnormal conditions of the spraying line, it was determined to use the photoelectric sensor and the intelligent visual detection scheme of deep learning to solve it. The detected abnormal conditions were alarmed and the line was stopped in conjunction with the spraying line to remind the operation The personnel handle it on site to solve the problem of missing procedures.
The basic solution to solve the problem of the spraying line body shaking and the parts shaking on the hanger and the defective parts cannot be identified is to add a photoelectric alarm sensor to the spraying line body, install two cameras in the fixed area, and install the light source to supplement the light in the operating area in time , To meet the camera’s photo requirements, as shown in Figure 4. Its working principle is: when the line body runs to the designated detection area, the hanger rod touches the photoelectric sensor, and triggers the signal of the camera to take pictures. By editing the calculation method of the camera controller software, the camera detects the position of the base angle of the chassis to solve the problem. For the problem of shaking and unstable rotation of the chassis, adjust the installation angle of the cameras on both sides to 60°. When the chassis enters the detection range, the left and right cameras start detection at the same time. When a non-conforming product is detected, an alarm will be sent to stop the line.
During the hanging process, the chassis vibrates and rotates at a large angle. The detection of photos taken by the cameras on both sides can solve the problem of false alarms and non-alarms caused by incomplete photos, as shown in Figure 5.
Solve the problem of false alarms caused by simultaneous hanging of large and small chassis and non-alarm when the hanging equipment is empty. With the cooperation of photoelectric sensors, cameras and controller programming, the controller identification and detection process is re-edited, and photos of the large and small chassis states are collected on site. Perform simulation training on the controller system to make the controller recognize the size of the chassis, and then detect the chassis footing, as shown in Figure 6.
Figure 5 Chassis tilt state detection
Figure 6 Different types of chassis detection
For the case of an empty chassis, according to the editing of the controller system process and photo acquisition training, after the chassis hanger enters the detection range of the camera, first identify whether there is a chassis in the four positions of the hanger, and then detect the identified chassis , Solves the problem of no alarm when the hanger is empty, as shown in Figure 7.
Figure 7 Identification of empty hanger
The team uses mechanical sensors, photoelectric sensors, deep learning vision detection technology and automated lines to form a self-learning system that can continuously enrich the knowledge base in practice. The system has the ability to search and understand environmental information, analyze, judge and plan its own behavior The ability to improve inspection reliability, ensure product quality, and upgrade automated production to intelligent production.