The Optimal Levenberg-Marquardt Algorithm to Predict Welding Quality Using Mahalanobis Distance Theory

2014 ◽  
Vol 936 ◽  
pp. 1873-1877
Author(s):  
Ill Soo Kim ◽  
Qian Qian Wu ◽  
Ji Hye Lee ◽  
Jong Pyo Lee ◽  
Min Ho Park ◽  
...  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Mahalanobis Distance ◽  
Gma Welding ◽  
Welding Process ◽  
Welding Parameters ◽  
Welding Quality ◽  
Marquardt Algorithm ◽  
Levenberg Marquardt ◽  
On Line

With the development of computational technology, neural network has attracted the more and more attentions to reveal the relationships between the process parameters and welding geometry. However, the Gas Metal Arc (GMA) welding is complex and of multiple interactions so that mathematical model for welding parameters has not been achieved. Neural networks have been noted as being particularly advantageous for modeling systems which contain noisy, fuzzy and uncertain elements, while a sufficient algorithm is employed. In this study, Levenberg-Marquardt algorithm was employed into GMA welding process. Mahalanobis Distance (MD) was measured to determine the on-line welding quality to avoid joint failure as welding quality. To get an optimal neural network, cases with different configurations were carried out. The Root of the Mean sum of Squared (RMS) error was adopted to evaluate the accuracy of the prediction by neural networks with LM algorithm. The results presented that the proposed algorithm had the superiority of high accuracy that can be used in the on-line welding process.

A Study on Development of the Optimal Neural Network in GMA Welding Process

2014 ◽  
Vol 936 ◽  
pp. 1759-1763
Author(s):  
Qian Qian Wu ◽  
Ji Hye Lee ◽  
Jong Pyo Lee ◽  
Min Ho Park ◽  
Young Su Kim ◽  
...  
Keyword(s):  
Neural Network ◽  
Process Parameters ◽  
Gma Welding ◽  
High Reliability ◽  
Welding Process ◽  
Welding Current ◽  
Welding Parameters ◽  
Welding Quality ◽  
Optimal Parameters ◽  
On Line

Gas Metal Arc (GMA) welding is considered as a multi-parameter process that it’s hard to find optimal parameters for good welding. To overcome the problem, an artificial neural network based on the backpropagation algorithm was built to realize the relationships between process parameters and welding quality as output parameter. In this study, Mahalanobis Distance (MD) was employed to evaluate the availability of a given welding parameters which was proved to performance well in multivariate statistics. Input parameters such as welding current and arc voltage were chosen due to their significant influence on the welding quality. To improve the precision of given parameters’ evaluation, neural networks with different configurations were verified. The analyses on the measured and predicted MD by the proposed neural network were conducted. The proposed neural network based on the error backpropogation algorithm was proved to have high reliability to evaluate process parameters, which further makes it available in on-line monitoring system.

Control of the weld quality using welding parameters in a robotic welding process

2018 ◽  
Vol 1 (87) ◽  
pp. 32-40 ◽  
Author(s):  
M.-H. Park ◽  
B.-J. Jin ◽  
T.-J. Yun ◽  
J.-S. Son ◽  
C.-G. Kim ◽  
...  
Keyword(s):  
Neural Network ◽  
Arc Welding ◽  
Gma Welding ◽  
Welding Process ◽  
Welding Parameters ◽  
Automatic Welding ◽  
Welding Quality ◽  
Arc Welding Process ◽  
Intelligent Models ◽  
Welding System

Purpose: Since the welding automations have widely been required for industries and engineering, the development of the predicted model has become more important for the increased demands for the automatic welding systems where a poor welding quality becomes apparent if the welding parameters are not controlled. The automated welding system must be modelling and controlling the changes in weld characteristics and produced the output that is in some way related to the change being detected as welding quality. To be acceptable a weld quality must be positioned accurately with respect to the joints, have good appearance with sufficient penetration and reduce low porosity and inclusion content. Design/methodology/approach: To achieve the objectives, two intelligent models involving the use of a neural network algorithm in arc welding process with the help of a numerical analysis program MATLAB have been developed. Findings: The results represented that welding quality was fully capable of quantifying and qualifying the welding faults. Research limitations/implications: Welding parameters in the arc welding process should be well established and categorized for development of the automatic welding system. Furthermore, typical characteristics of welding quality are the bead geometry, composition, microstructure and appearance. However, an intelligent algorithm that predicts the optimal bead geometry and accomplishes the desired mechanical properties of the weldment in the robotic GMA (Gas Metal Arc) welding should be required. The developed algorithm should expand a wide range of material thicknesses and be applicable in all welding position for arc welding process. Furthermore, the model must be available in the form of mathematical equations for the automatic welding system. Practical implications: The neural network models which called BP (Back Propagation) and LM (Levenberg-Marquardt) neural networks to predict optimal welding parameters on the required bead reinforcement area in lab joint in the robotic GMA welding process have been developed. Experimental results have been employed to find the optimal algorithm to predict bead reinforcement area by BP and LM neural networks in lab joint in the robotic GMA welding. The developed intelligent models can be estimated the optimal welding parameters on the desired bead reinforcement area and weld criteria, establish guidelines and criteria for the most effective joint design for the robotic arc welding process. Originality/value: In this study, intelligent models, which employed the neural network algorithms, one of AI (Artificial Intelligence) technologies have been developed to study the effects of welding parameters on bead reinforcement area and to predict the optimal bead reinforcement area for lab joint in the robotic GMA welding process. BP (Back Propagation) and LM (Levenberg-Marquardt) neural network algorithm have been used to develop the intelligent model.

A Study on Welding Quality of Robotic Arc Welding Process Using Mahalanobis Distance Method

2013 ◽  
Vol 773-774 ◽  
pp. 759-765 ◽  
Author(s):  
Reenal Ritesh Chand ◽  
Ill Soo Kim ◽  
Ji Hye Lee ◽  
Jong Pyo Lee ◽  
Ji Yeon Shim ◽  
...  
Keyword(s):  
Process Parameters ◽  
Mahalanobis Distance ◽  
Arc Welding ◽  
Gma Welding ◽  
Welding Process ◽  
Experimental Results ◽  
Welding Parameters ◽  
Bead Geometry ◽  
Welding Quality ◽  
Arc Welding Process

In robotic GMA (Gas Metal Arc) welding process, heat and mass inputs are coupled and transferred by the weld arc and molten base material to the weld pool. The amount and distribution of the input energy are basically controlled by the obvious and careful choices of welding process parameters in order to accomplish the optimal bead geometry and the desired mechanical properties of the quality weldment. To make effective use of automated and robotic GMA welding, it is imperative to predict online faults for bead geometry and welding quality with respect to welding parameters, applicable to all welding positions and covering a wide range of material thickness. To successfully accomplish this objective, two sets of experiment were performed with different welding parameters; the welded samples from SM 490A steel flats adopting the bead-on-plate technique were employed in the experiment. The experimental results of current and voltage waveforms were used to predict the magnitude of bead geometry and welding quality, and to establish the relationships between weld process parameters and online welding faults. MD (Mahalanobis Distance) technique is employed for investigating and modeling of GMA welding process and significance test techniques were applied for the interpretation of the experimental data. Statistical models developed from experimental results which can be used to control the welding process parameters in order to achieve the desired bead geometry based on weld quality criteria.

A study on welding quality for the automatic vertical-position welding process based on Mahalanobis Distance method

2018 ◽  
Vol 1 (91) ◽  
pp. 31-40
Author(s):  
B.-J. Jin ◽  
M.-H. Park ◽  
T.-J. Yun ◽  
J.-Y. Shim ◽  
B.-Y. Kang ◽  
...  
Keyword(s):  
Mahalanobis Distance ◽  
Welding Process ◽  
Welding Current ◽  
Welding Parameters ◽  
Vertical Position ◽  
Welding Quality ◽  
Distance Method ◽  
Arc Voltage ◽  
On Line ◽  
Arc Welding Process

Purpose: The welding quality and reducing production cost could be achieved by developing the automatic on-line welding quality monitoring system. However, investigation of welding fault to quantify the welding quality on the horizontal-position welding has been concentrated. Therefore, MD (Mahalanobis Distance) method on the vertical-position welding process by analysing the transform arc voltage and welding current gained from the on-line monitoring system has been applied. Design/methodology/approach: The transformed welding current and arc voltage data were taken from the experiment whereby the data number was 2500 data/s. The prediction of Contact Tip to Work Distance (CTWD) to gain best welding quality using the waveform variations were then taken from the experimental results. MD was employed to quantify the welding quality by analysing the transformed arc voltage and welding current. Finally, the optimal CTWD setting has verified the developed algorithms through additional experiments. Two kinds of experiments has been carried out by changing welding parameters artificially to verify the sensitivity and feasibility of WQ (Welding Quality) based on the concepts of MD and normal distribution. Findings: The results represented that WQ was fully capable of quantifying and qualifying the welding faults for automatic vertical-position welding process. Research limitations/implications: The arc welding process on the vertical-position compared to a horizontal-position welding is much more difficult because the metal transfer is influenced by the gravity force. To solve the problem, a new algorithm to monitor and control the welding fault during the arc welding process has been developed. Furthermore, optimization of welding parameters for the vertical-position welding process was really difficult to use the developed algorithms because they are only useful in selecting stored data and not for evaluating the effect of the variation of welding parameters on the weld ability. Practical implications: The developed algorithm could be achieved the highest welding quality at 15mm CTWD setting which the welding quality is 99.50% for the start section and 99.68% at the middle section. Originality/value: This paper proposed a new algorithm which employed the concepts of MD (Mahalanobis Distance) and normal distribution to describe a good quality welding.

scholarly journals Adaptive Levenberg–Marquardt Algorithm: A New Optimization Strategy for Levenberg–Marquardt Neural Networks

Mathematics ◽  
2021 ◽  
Vol 9 (17) ◽  
pp. 2176
Author(s):  
Zhiqi Yan ◽  
Shisheng Zhong ◽  
Lin Lin ◽  
Zhiquan Cui
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Frequency Noise ◽  
Optimization Strategy ◽  
Engineering Data ◽  
Marquardt Algorithm ◽  
Highly Nonlinear ◽  
Rectified Linear Unit ◽  
Levenberg Marquardt ◽  
Lm Algorithm

Engineering data are often highly nonlinear and contain high-frequency noise, so the Levenberg–Marquardt (LM) algorithm may not converge when a neural network optimized by the algorithm is trained with engineering data. In this work, we analyzed the reasons for the LM neural network’s poor convergence commonly associated with the LM algorithm. Specifically, the effects of different activation functions such as Sigmoid, Tanh, Rectified Linear Unit (RELU) and Parametric Rectified Linear Unit (PRLU) were evaluated on the general performance of LM neural networks, and special values of LM neural network parameters were found that could make the LM algorithm converge poorly. We proposed an adaptive LM (AdaLM) algorithm to solve the problem of the LM algorithm. The algorithm coordinates the descent direction and the descent step by the iteration number, which can prevent falling into the local minimum value and avoid the influence of the parameter state of LM neural networks. We compared the AdaLM algorithm with the traditional LM algorithm and its variants in terms of accuracy and speed in the context of testing common datasets and aero-engine data, and the results verified the effectiveness of the AdaLM algorithm.

scholarly journals The atmospheric model of neural networks based on the improved Levenberg-Marquardt algorithm

2021 ◽  
Vol 30 (1) ◽  
pp. 24-35
Author(s):  
Wenhui Cui ◽  
Wei Qu ◽  
Min Jiang ◽  
Gang Yao
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Influencing Factor ◽  
Atmospheric Model ◽  
Average Error ◽  
Atmospheric Models ◽  
Neural Network Models ◽  
Model Based ◽  
Marquardt Algorithm ◽  
Levenberg Marquardt

Abstract Traditional atmospheric models are based on the analysis and fitting of various factors influencing the space atmosphere density. Neural network models do not specifically analyze the polynomials of each influencing factor in the atmospheric model, but use large data sets for network construction. Two traditional atmospheric model algorithms are analyzed, the main factors affecting the atmospheric model are identified, and an atmospheric model based on neural networks containing various influencing factors is proposed. According to the simulation error, the Levenberg-Marquardt algorithm is used to iteratively realize the rapid network weight correction, and the optimal neural network atmospheric model is obtained. The space atmosphere is simulated and calculated with an atmospheric model based on neural networks, and its average error rate is lower than that of traditional atmospheric models such as the DTM2013 model and the MSIS00 model. At the same time, the calculation complexity of the atmospheric model based on the neural networks is significantly simplified than that of the traditional atmospheric model.

Prediction Welding Quality in Multi-Pass Welding Process Using Mahalanobis Distance Method

2013 ◽  
Vol 291-294 ◽  
pp. 2688-2693 ◽  
Author(s):  
Reenal Ritesh Chand ◽  
Il Soo Kim ◽  
Ji Hye Lee ◽  
Ji Sun Kim
Keyword(s):  
Mahalanobis Distance ◽  
Welding Process ◽  
Welding Parameters ◽  
Welding Quality ◽  
Qualitative And Quantitative Analysis ◽  
Distance Method ◽  
Distribution Method ◽  
Mahalanobis Distances ◽  
Qualitative And Quantitative

The welding quality in multi-pass welding is mainly dependent on the pre-heating from pervious pass or root-pass welding. In this study, a Mahalanobis Distance and normal distribution method is illustrated and employed to determine whether welding faults have occurred after each pass welding and also to quantify welding quality percentage. To successfully accomplish this objective, sets of multi-pass welding experiment were performed with different welding parameters in each pass; the welded samples of SS400 steel flats adopting the bead-on-plate technique were employed in the experiment. The result of current and voltage for each pass is obtained through the real time mentoring systems. In order to verify the effect of the performance and weld quality of the different weld-pass, Mahalanobis distances for voltage and current values were calculated and used for qualitative and quantitative analysis with comparison to values obtained from the root-pass as reference welds. The results of the experiment and statistical analysis have demonstrated that the weld faults after each weld pass is feasible.

A Fuzzy Rule-Based Method for Seeking Stable Arc Condition under Short-Circuiting Mode of GMA Welding Process

1992 ◽  
Vol 206 (2) ◽  
pp. 117-125
Author(s):  
Y J Won ◽  
H S Cho
Keyword(s):  
Gma Welding ◽  
Welding Process ◽  
Fuzzy Rule ◽  
Welding Parameters ◽  
Experimental Conditions ◽  
Rule Based ◽  
Arc Stability ◽  
Linguistic Rules ◽  
On Line ◽  
Searching Method

An unstable arc during welding induces unsound bead, thus resulting in poor mechanical properties at the welded joints. One of the important factors affecting arc stability is the welding voltage/current combination. In practice, a proper combination is usually determined by performing off-line experiments, and no extensive work has been reported on developing on-line searching methods. The paper proposes a systematic on-line method for searching the proper voltage/current combination to provide a stable arc condition in the short-circuiting metal transfer mode. Because the relationship between such welding parameters and the resulting arc stability is excessively complex and non-linear, it is difficult to determine a proper voltage/current combination based on mathematical modelling. To overcome this difficulty, a fuzzy rule-based method is proposed so that the complexity and non-linearity of arc behaviour is effectively represented by adopting fuzzy linguistic rules. The fuzzy rules here are constituted with two fuzzy variables, namely Mita's arc stability index and its derivative with respect to voltage. Then, under experimental conditions, fuzzy inferencing is performed to adjust recursively the voltage/current combination in such a way as to give a stable arc condition. The performance of the proposed searching method is also evaluated through a series of experiments. The experimental results show that the proposed searching method can effectively estimate the proper voltage/current combination to give a stable arc condition.

Sign in / Sign up

Export Citation Format

Share Document