• 대한기계학회논문집
• /
• 제10권2호
• /
• pp.241-246
• /
• 1986

본 연구에서는 산업 현장에 쉽게 적용할 수 있도록 측정과 설치가 용이한 용접 모재간에 가해지는 전압(weld voltage)파형에서 용접 변수를 구하고, 실험으로 구한 용접 강도와의 상관 관계로부터, 점용접 품질 감시체제를 개발하고자 한다. 전압 파형에서 용접 변수를 구하기 위하여 추계 모델링 기법을 적용하고, 구해진 모 델로부터 분산 해석(dispersion analysis)을 통하여 용접 변수를 찾으며, 동시에 전극 변위 곡선을 실험적으로 구하여서, 분산 해석 방법의 신뢰도를 검사 하고자 한다.

• 대한용접접합학회:학술대회논문집
• /
• 대한용접접합학회 2002년도 Proceedings of the International Welding/Joining Conference-Korea
• /
• pp.595-599
• /
• 2002

Edge welding of thin sheets is very difficult because of the fit-up problem and small weld area In laser welding, joint fit-up and penetration are critical for sound weld quality, which can be monitored by appropriate methods. Among the various monitoring systems, visual monitoring method is attractive because various kinds of weld pool information can be extracted directly. In this study, a vision sensor was adopted for the weld pool monitoring in pulsed Nd:YAG laser edge welding to monitor whether the penetration is enough and the joint fit-up is within the requirement. Pulsed Nd:YAG laser provides a series of periodic laser pulses, while the shape and brightness of the weld pool change temporally even in one pulse duration. The shutter-triggered and non-interlaced CCD camera was used to acquire a temporally changed weld pool image at the moment representing the weld status well. The information for quality monitoring can be extracted from the monitored weld pool image by an image processing algorithm. Weld pool image contains not only the information about the joint fit-up, but the penetration. The information about the joint fit-up can be extracted from the weld pool shape, and that about a penetration from the brightness. Weld pool parameters that represent the characteristics of the weld pool were selected based on the geometrical appearance and brightness profile. In order to achieve accurate prediction of the weld penetration, which is nonlinear model, neural network with the selected weld pool parameters was applied.

• Journal of Welding and Joining
• /
• 제12권1호
• /
• pp.59-72
• /
• 1994

This paper presents an on-line quality monitoring and control method to obtain a uniform weld quality in gas metal arc welding (GMAW) processes. The geometrical parameters of the weld pool such as the top bead width and the penetration depth plus half back width are utilized to assess the integrity of the weld quality. Since a good quality weld is characterized by a relatively high depth-to-width ratio in its dimensions, the second geometrical parameter is regulated to a desired one. The monitoring variables are the surface temperatures measured at various points on the top surface of the weldment which are strongly related to the formation of the weld pool The relationship between the measured temperatures and the weld pool size is implemented on the multilayer perceptrons which are powerful for realization of complex mapping characteristics through training by samples. For on-line quality monitoring and control, it is prerequisite to estimate the weld pool sizes in the region of transient states. For this purpose, the time history of the surface temperatures is used as the input to the neural estimator. The control purpose is to obtain a uniform weld quality. In this research, the weld pool size is directly regulated to a desired one. The proposed controller is composed of a neural pool size estimator, a neural feedforward controller and a conventional feedback controller. The pool size estimator predicts the weld pool size under growing. The feedforward controller compensates for the nonlinear characteristics of the welding process. A series of simulation studies shows that the proposed control method improves the overall system response in the presence of changes in torch travel speed during GMA welding and guarantees the uniform weld quality.

• 한국기계가공학회지
• /
• 제21권6호
• /
• pp.89-97
• /
• 2022

Ultrasonic metal welding has been widely used for joining lithium-ion battery tabs. Weld quality monitoring has been an important issue in lithium-ion battery manufacturing. This study focuses on the weld quality monitoring in ultrasonic metal welding with the longitudinal-torsional vibration mode horn developed newly. As the quality of ultrasonic welding depends on welding parameters like pressure, time, and amplitude, the suitable values of these parameters were selected for experimentation. The welds were tested via tensile testing machine and weld strengths were investigated. The dataset collected for performance test was used to train the multi-layer perceptron neural network. The three layer neural network was used for the study and the optimum number of neurons in the first and second hidden layers were selected based on performances of each models. The best models were selected for the horn and then tested to see their performances on an unseen dataset. The neural network models for the longitudinal-torsional mode horn attained test accuracy of 90%. This result implies that proposed models has potential for the weld quality monitoring.

• 한국레이저가공학회지
• /
• 제3권2호
• /
• pp.53-61
• /
• 2000

On the laser weld production line, a slight alteration of the welding condition produces many defects. The defects are monitored in real time, in order to prevent continuous occurrence of defects, reduce the loss of material, and guarantee good quality. The measurement system is produced by using three photo-diodes for detection of the plasma and spatter signal in CO$_2$ laser welding. For high speed CO$_2$ laser welding, laser tailored welded blanks for example, on-line weld quality monitoring system was developed by using fuzzy multi-feature pattern recognition. Weld qualities were classified optimal heat input, a little low heat input, low heat input, and focus misalignment, and final weld quality were classified good and bad.

• International Journal of Korean Welding Society
• /
• 제1권1호
• /
• pp.7-12
• /
• 2001

Real time monitoring has become critical as the use of laser welding increases. Plasma and spatter are measured and used as the signal for estimating weld quality. The estimating algorithm was made using the fuzzy pattern recognition with the area of data that is beyond the tolerance boundary. Also, an algorithm that detects the spatter and the localized defect was created in order to kd the partially produced pit and the sudden loss of weld penetration. These algorithms were used in quality monitoring of the $CO_2$ laser tailored blank weld. Statistical program that can display the laser weld quality result and the signal transition was made for the first stage of the remote control system.

• Journal of Welding and Joining
• /
• 제1권2호
• /
• pp.69-75
• /
• 1983

Both in-process quality control and reliability of the weld is one of the major concerns is applying friction welding. No reliable nondestructive monitoring method is available at present to determine the weld quality particularly in process of production. So that, this paper presents an experimental examination and quantitative analysis for the effects of initial acoustic emission(AE) counts on the weld strength relating to the rotating speed as a new approach which attempts finally to develop an on-line quality monitoring system design for friction welds using AE techniques. As one of the important results, it was well confirmed that the initial AE counts occurring during plastic deformation period of welding were quantitatively correlated with reliability at 95% confidence level to the joint strength of welds, tube-to-tube (SM 20 C to STS 304) and then an AE technique using the initial AE counts can be reliably applied to in-process strength monitoring of the weld.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1996년도 추계학술대회 논문집
• /
• pp.474-478
• /
• 1996

Process monitoring in laser welding is essential for automation and quality control of products. Various signals from laser welding, such as plasma, sound, optical signals, etc., are utilized for monitoring the process and detecting abnormal weld conditions. In this study, both W light from plasma formed above the weld pool and IR signal from the melting pool are detected with photodiodes and PC-based A/D board, and analyzed to give a guidance about the weld quality. Experimental results show the possibility of using the signals for predicting and evaluating the weld qualify and adapting into the system for on-line process monitoring.

• 대한용접접합학회:학술대회논문집
• /
• 대한용접접합학회 2002년도 Proceedings of the International Welding/Joining Conference-Korea
• /
• pp.307-313
• /
• 2002

This paper describes a weld pool monitoring technique, which is based on the weld pool image analysis. The proposed image analysis algorithm uses machine vision techniques to extract geometrical information from the weld pool image such as maximum weld pool width, gap width and misalignment between the joint longitudinal axis and the welding wire. These can be related to the welding parameters (welding voltage and current, wire feed speed and standoff) to produce control actions necessary to ensure that the required weld quality will be achieved. The experiments have shown that the algorithm is able to produce good estimates of the weld pool geometry; however, the adjustment of the camera parameters affects the image quality and, consequently, has a great influence over the estimation.

• 한국레이저가공학회지
• /
• 제2권3호
• /
• pp.19-31
• /
• 1999

This paper introduces the monitoring scheme of laser welding quality using neural network. The developed monitoring scheme detects light signal emitting from plasma formed above the weld pool with optic sensor and DSP-based signal processor, and analyzes to give a guidance about the weld quality. It can automatically detect defects of laser weld and further give an information about what kind of defects it is, specially partial penetration and porosity among the interior defects. Those could be detected only by naked eyes or X-ray after welding, which needs more processes and costs in mass production. The monitoring scheme extracts four feature vectors from signal processing results of optical measuring data. In order to classify pattern for extracted feature vectors and to decide defects, it uses single-layer neural network with perceptron learning. The monitoring result using only the first feature vector shows confidence rate in recognition of 90%($\pm$5) and decides whether normal status or defects status in real time.