Application of Linear CCD Technology in Steel Plate Width Measuring Instrument

　　Strip steel quality inspection Strip steel width is an important quality indicator in strip steel production. In order to ensure the quality control of strip steel production, it is necessary to monitor the strip steel width online during the strip steel rolling process to improve the product width performance indicators. The strip steel width gauge proposed in this paper uses linear array CCD image detection technology to realize non-contact dynamic width measurement of transmission strip steel and strip steel width out-of-tolerance alarm.

　　CCD technology is used to conduct real-time detection and dynamic measurement of product surface quality, which has the advantages of simple structure, non-contact, high precision, fast measurement speed, stable and reliable performance, etc. The main sensor component of the camera is CCD, which has the characteristics of high sensitivity, small distortion, long life, anti-vibration, anti-magnetic field, small size, no residual image, etc. The basic principle of CCD imaging detection is that, first, under the action of the optical system, the change of a certain characteristic of the object to be measured is converted into a change in the angle of the light beam. The light beam is irradiated on the light receiving window surface of the CD device, and photogenerated charges are generated in the pixels of the light receiving window surface irradiated by the light, and the charges are stored in the pixel unit. The pixels that are not irradiated by the light do not generate photogenerated charges. Then, under the control of the readout clock pulse in the CCD drive circuit, the charges are transferred and shifted to the output circuit, and the output circuit converts the charge amount into a voltage output. The output of the CCD device is a digital video signal, which needs to be converted into a standard signal by a user-designed data processing circuit for further processing.

　　1 Measurement scheme and measurement system

　　The width gauge currently used in China is mainly of the structure shown in the figure. It mainly includes two parts, the detection box and the light source part. The detection box is the head of the width gauge, which includes two cameras, a screw rod for connecting the two cameras, a motor for driving the screw rod, and a variable speed brake electromagnetic clutch. In addition, there are self-synchronous angle transmitters and water and air systems for measuring the reference width. In the detection box, in addition to the two rotating slit detection heads, CCD cameras are used. The lower light source box is composed of 6 40W fluorescent lamps, 3 on the left and 3 on the right. In addition, compressed air is introduced into the light source box for cooling and dust prevention. In order to prevent the light-transmitting glass on the light source box from being contaminated by iron oxide, etc., cooling water is passed through the glass. However, it can be found from the structure that its structure is complex, the control body is cumbersome, the position of the steel plate camera needs to be calibrated, and timely maintenance is required, and the real-time operability is very poor. Therefore, an improvement method can be found.

　　First, make an improvement on the light source. Use a laser tube to hit the light source below on a swinging and rotating reflector, so that the light spot hits the steel plate to form a straight line. As shown in the figure. Then make a change in the imaging. No motor is used to drive the CCD to run, but a convex lens is used to image the infrared light on the CCD. This structure is quite simple and easy to debug. For example, if a 2500-pixel CCD is selected to measure a 2.5m wide steel plate, then 1 pixel of the CCD corresponds to 1 mm of the steel plate. In actual situations, the influence of the laser beam and the influence of the steel plate transmission on the measurement can be ignored. Because the light source is relatively stable, the transmission speed of the steel plate is very slow compared to the speed of the reflector and CCD, 1.5 mm/s. In this way, a bright line is imaged on the cross section of the CCD. The number of pixels of this line multiplied by 1mm is the width of the steel plate. The measurement is very simple. Compared with the above structure, this structure is very simple. If the accuracy can be less than 0.5mm, it can be completely replaced. The accuracy of this system will be demonstrated later. As shown in the figure above, when the CCD is sensitive to light, the pixel on the CCD becomes a high-level pulse, otherwise it is a low pulse. Therefore, as long as the width of the CCD high-level pulse is measured, the size of the steel plate can be obtained. Analyzing the figure, we only need to calculate the number of pulses at the N1 moment and the number of pulses after N2. The integration time is fixed, and the number of pulses between N1 and N2 can be obtained, so as to obtain the corresponding measured size. Here, considering that the frequency of CCD is relatively high, 20M, it is difficult for MCU to process. Observe the actual output of CCD, count with CPLD, and transmit the measured data to MCU for processing when the field signal of CCD ends. This avoids cumbersome hardware design. This process is very simple. Use CPLD to count the output of DOS. When the falling edge of the field signal comes, the counter is cleared. When the rising edge of the field signal comes, the data in the counter is sent to the data line, and an interrupt is applied to the MCU at the same time. The MCU processes the data and displays the output and passes it to the lower computer.

　　2 Edge segmentation

　　The video digital signal output by CCD contains image background information and image information, as shown in the light shielding pixel area and signal output area in the figure. Due to the change in light intensity between the object under test and the background, it is reflected in the image spectrum corresponding to the CCD video signal, and there will be obvious level changes at the boundary. However, how to accurately locate the boundary specifically requires separating the background and image information in the CCD video signal into binary level information. The threshold method is usually used. If the threshold is too high, too many target points will be mistakenly classified as background; otherwise, the opposite situation will occur. This will inevitably affect the specific size of the segmented target. Therefore, determining the threshold is the core of the problem. Analysis of the output waveform of the CCD video signal shows that the image boundary is at the point where the waveform curve has the largest rate of change. For this reason, the point corresponding to the maximum rate of change of the curve can be found by differentiation. This method is called the differentiation method. The circuit principle block diagram of the differentiation method is shown in the figure. The amplitude modulated pulse signal output by the CCD video is converted into a continuous video signal after a sampling and holding circuit or a low-pass filter, as shown in the figure. The continuous video signal is differentiated by differential circuit I, and its output is the rate of change of the video signal. The maximum value of the signal voltage corresponds to the point where the video signal boundary filter area changes the most (points A and A in the figure). Differentiation I generates a negative pulse on the falling edge of the video signal and a positive pulse on the rising edge, as shown in the second waveform of the fourth figure. The two pulse signals with opposite levels output by differential I are sent to the absolute value circuit, and the signal output by differential I circuit is converted into a pulse signal of the same level through the absolute value circuit, as shown in the third waveform of the figure. The amplitude point of the signal corresponds to the boundary feature point. The pulse of the same level is sent to differential circuit II for differentiation again, and the zero-crossing signal corresponding to the absolute maximum value is obtained. The zero-crossing signal passes through the zero trigger again, and outputs two pulse signals with falling edges corresponding to the zero-crossing point. By using the falling edges of these two pulses to trigger a trigger, the initial boundary feature pulse of the video signal, that is, the binary signal, can be obtained. Its pulse width is the width between the graph AA.

　　The strip image is separated from the background in this way, and the accuracy is higher than the previous one. Its theoretical accuracy can be improved to 1/7 pixel, that is, 1/7mm. Such accuracy is sufficient in practical applications. The actual measurement can be guaranteed to 0.2mm.
 

　　3 Usage and Conclusion

　　In the actual application of steel plants, some of the measured data are shown in the following table:

　　As can be seen from the figure, its accuracy meets the actual needs. The transformation of the width gauge is basically successful. Whether in terms of reliability, detection accuracy, failure rate, etc., it meets the production process requirements and reaches the design requirements. The user response is good.

　　The author's innovation: There is a great innovation in the structure. The structure is quite simple, no repeated calibration is required, the operation is simple, and the cost is quite economical, which is one-third of the original cost. The measurement principle is also relatively simple compared with the signal acquisition system.

　　references

　　[1] Wang Qingyou. CCD Application Technology, Tianjin: Tianjin University Press, 2000, 120-127, 177-180

　　[2] Chen Zhuan. Analysis and design of CCD photoelectric size detection device, Journal of Changchun Institute of Optics and Precision Mechanics, 1992, 15(1): 47-52

　　[3] Zhang Zhuo. Research on dynamic angle measurement and its error detection technology: [Master's degree thesis], Harbin: Harbin Institute of Technology. 1996

　　[4] Xiao Zexin. Engineering Optical Design, Beijing: Publishing House of Electronics Industry, 2003, 7-45

　　[5] Zhang Li, Yang Yongming. UART design based on CPLD. [J] Microcomputer Information, 2002, 18(2): 60-61

　　Author profile: He Sheyang, male (1976-), Han nationality, from Lingbao, Henan, master's degree, teacher at Henan University of Science and Technology, main research directions: detection and control, instrumentation, etc.

　　Correspondence address: (P.O. Box 65, School of Mechanical and Electrical Engineering, Henan University of Science and Technology, Luoyang, Henan 471003) He Sheyang

　　(Received: 2007.2.13) (Revised: 2007.3.15)