Optimization of underwater wet welding process parameters using neural network
- Joshua Emuejevoke Omajene^{1}Email author,
- Jukka Martikainen^{1},
- Huapeng Wu^{1} and
- Paul Kah^{1}
https://doi.org/10.1186/s40712-014-0026-3
© Omajene et al.; licensee Springer. 2014
Received: 8 September 2014
Accepted: 5 November 2014
Published: 21 November 2014
Abstract
Background
The structural integrity of welds carried out in underwater wet environment is very key to the reliability of welded structures in the offshore environment. The soundness of a weld can be predicted from the weld bead geometry.
Methods
This paper illustrates the application of artificial neural network approach in the optimization of the welding process parameter and the influence of the water environment. Neural network learning algorithm is the method used to control the welding current, voltage, contact tube-to-work distance, and speed so as to alter the influence of the water depth and water environment.
Results
The result of this work gives a clear insight of achieving proper weld bead width (W), penetration (P), and reinforcement (R).
Conclusions
An interesting implication of this work is that it will lead to a robust welding activity so as to achieve sound welds for offshore construction industries.
Keywords
Backpropagation Bead geometry Neural network Process parameter Underwater weldingBackground
Methods
Underwater welding
Underwater welding is used for the repair welding of ships and offshore engineering structures like oil drilling rigs, pipelines, and platforms. The commonly used underwater welding processes nowadays are shielded metal arc welding (SMAW) and flux cored arc welding (FCAW). The water surrounding the weld metal reduces the mechanical properties of weld done underwater due to the effect of the fast cooling rate of the weld. Heat loss by conduction from the plate surface into the moving water environment and heat loss by radiation are the major heat losses in underwater welding. Underwater welding requires a higher current for the same arc voltage to achieve a higher heat input as compared to air welding. The fast cooling rate of underwater welding results in the formation of constituents such as martensite and bainite for conventional welding of steels. These constituents lead to a high-strength, brittle material and susceptibility to hydrogen-induced cracking. The weld bead shape for underwater wet welding are more spread out and less penetrating than air welds. Underwater welding arc is constricted at increased depth or pressure. However, welding in shallow depth is more critical than that in higher depth. The unstable arc results in porosity which affects the soundness of the weld. Weld metal carbon content increases with increase in water depth. Also, manganese and silicon which are deoxidizers are increasingly lost at increased water depth (Omajene et al. 2014).
Artificial neural network
Summary of the backpropagation training algorithm
The summary of the backpropagation training algorithm is illustrated as follows (Negnevetsky 2005).
Set the weights and threshold levels of the network to uniformly random numbers distributed in small range. F_{ i } is the total number of inputs of neuron i in the network.
Step 2: Activation
- (a)Calculate the actual outputs of the neurons in the hidden layer:where n is the number of inputs of neuron j in the hidden layer and sigmoid is the sigmoid activation function.$$ {y}_j(p)=\mathrm{sigmoid}\left[{\displaystyle {\sum}_{i=0}^n{x}_i(p)}*{w}_{ij}(p)-{\theta}_j\right] $$
- (b)
Calculate the actual outputs of the neurons in the output layer.
- (c)where m is the number of inputs of neuron k in the output layer.$$ {y}_k(p)=\mathrm{sigmoid}\left[{\displaystyle {\sum}_{j=0}^m{y}_j(p)*{w}_{jk}(p)-{\theta}_k}\right] $$
Step 3: Weight training
- (a)Calculate the error gradient for the neurons in the output layer:where$$ {\delta}_k(p)={y}_k(p)*\left[1\hbox{--} {y}_k(p)\right]*{e}_k(p) $$$$ {e}_k(p)={y}_{d,k}(p)\hbox{--} {y}_k(p) $$Calculate the weight corrections$$ \Delta {w}_{jk}(p)=\alpha *{y}_j(p)*{\delta}_k(p) $$Update the weights at the output neurons:$$ {w}_{jk}\left(p+1\right)={w}_{jk}(p)+\Delta {w}_{jk}(p) $$
- (b)Calculate the error gradient for the neurons in the hidden layer:$$ {\delta}_j(p)={y}_j(p)*\left[1\hbox{--} {y}_j(p)\right]*{\displaystyle {\sum}_{k=0}^{\ell }{\delta}_k}(p)*{w}_{jk}(p) $$Calculate the weight corrections$$ \Delta {w}_{ij}(p)=\alpha *{x}_i(p)*{\delta}_j(p) $$Update the weights at the output neurons:$$ {w}_{ij}\left(p+1\right)={w}_{ij}(p)+\Delta {w}_{ij}(p) $$
Step 4: Increase iteration p by 1, go back to step 2, and repeat the process until the selected error criterion is satisfied.
Results and discussion
The ANN scheme to predict the weld bead geometry in underwater wet welding is shown in Figure 1. The aim is to map a set of input patterns to a corresponding set of output patterns by learning from past examples how the input parameters and output parameters relate. A feedforward backpropagation network trained with scaled conjugate gradient (SCG) backpropagation algorithm is used. The quality of the weld can be verified when the training pattern fulfills the requirement for the accepted ranges of WPSF (penetration shape factor) = W/P and WRFF (reinforcement form factor) = W/R. The accepted ranges for a weld with good quality are a maximized penetration to width ratio and minimized undercut and reinforcement.
Design parameters
Experimental data adapted from (Shi et al. 2013 )
Serial number | Process parameters | Water depth | Bead geometry | Error = Output (W,P,R) − Target | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Target = 0 | |||||||||||
I(A) | U(V) | v(m/s) | D(m) | H(m) | W(m) | P(m) | R(m) | ΔW(m) | ΔP(m) | ΔR(m) | |
1 | 280 | 28 | 10 | 20 | 40 | 10.4 | 2.5 | 4.3 | 0.0061 | 0.0091 | 0.0016 |
2 | 320 | 32 | 6 | 20 | 20 | 12.5 | 3.8 | 8 | 0.0022 | −0.0394 | 0.0998 |
3 | 300 | 32 | 10 | 22 | 60 | 10.4 | 3 | 4 | −0.3280 | −1.1397 | −0.5471 |
4 | 340 | 28 | 6 | 22 | 0.1 | 13.9 | 3.5 | 3 | 1.6136 | −0.1071 | −0.7470 |
5 | 280 | 30 | 6 | 24 | 60 | 12.9 | 3.7 | 6.1 | 0.0223 | −0.0104 | 0.0083 |
6 | 320 | 26 | 10 | 24 | 0.1 | 11.6 | 1.8 | 2 | 0.0093 | 0.0650 | 0.0269 |
7 | 300 | 26 | 6 | 18 | 40 | 12 | 2.9 | 5 | −0.0177 | −0.0126 | 0.0080 |
8 | 340 | 30 | 10 | 18 | 20 | 9.4 | 4.2 | 4.3 | −0.0193 | 0.1365 | 0.0038 |
9 | 280 | 26 | 12 | 22 | 20 | 8.9 | 1.7 | 4.5 | −0.3662 | −1.0633 | −0.2616 |
10 | 320 | 30 | 8 | 22 | 40 | 11.8 | 3.3 | 4.8 | −0.0323 | 0.0312 | 0.0007 |
11 | 300 | 30 | 12 | 20 | 0.1 | 12.8 | 1.7 | 1.9 | −0.0363 | −0.1342 | −0.0876 |
12 | 340 | 26 | 8 | 20 | 60 | 9.5 | 3.4 | 4.8 | 0.0169 | −0.0132 | 0.0008 |
13 | 280 | 32 | 8 | 18 | 0.1 | 12.5 | 2 | 2 | 0.0089 | 0.0151 | 0.0553 |
14 | 320 | 28 | 12 | 18 | 60 | 7.9 | 2.7 | 4.9 | −0.8149 | −0.9025 | 1.8429 |
15 | 300 | 28 | 8 | 24 | 20 | 10.1 | 3.1 | 4.9 | −0.0211 | 0.0055 | −0.0067 |
16 | 340 | 32 | 12 | 24 | 40 | 10 | 3 | 4 | 0.0011 | −0.0140 | 0.0213 |
Program algorithm
Program algorithm
Program algorithm | |
---|---|
load matlab.mat | |
% inputs | |
I=DataProject(:,1); | D=DataProject(:,4); |
U=DataProject(:,2); | H=DataProject(:,5); |
v=DataProject(:,3); | F=[I U v D H]; |
%outputs | |
W=DataProject(:,6); | |
P=DataProject(:,7); | G=[W P R]; |
R=DataProject(:,8); | |
% training | |
p=F(1:12,:); | t=G(1:12,:); |
% testing | |
x=F(13:16,:); | Z=[x y]; |
y=G(13:16,:); | |
% form the network | |
net=feedforwardnet([40],'trainscg'); | net.trainParam.max_fail=2000; |
net.trainParam.goal=0; % error goal | net.trainParam.lr=0.001; |
net.trainParam.epochs=3000; % maximum iterations | net.trainParam.mc=0.9; |
net.trainParam.show=25; % showing intervals | |
% Network initialization | |
net.initFcn='initlay'; | [net,tr]=train(net,p',t'); % training the net |
net.layers{1}.initFcn='initnw'; | view(net) |
net=init(net);% initialize the net (weights and biases initialized) | |
% simulating the network with training inputs for testing | |
f=net(x'); | f' |
% compare results/target | |
Error=f'-y |
Validation performance
Regression analysis
Controller for underwater wet welding process
Conclusions
The optimization of the parameters that affect weld bead geometry during underwater welding can be done by artificial neural network training algorithm. In this study, the regression analysis show that the target follows closely the output as R is at least 96% for training, testing, and validation. The trained neural network with satisfactory results can be used as a black box in the control system of the welding process. The effective optimization of the welding process parameter in underwater wet welding has the ability of welding with an optimized heat input and optimized arc length which will guarantee arc stability. The use of optimized process parameters enables the achievement of an optimized weld bead geometry which is a key factor in the soundness of welds. The control process for underwater welding as suggested in this paper requires further research so as to fully apply the NN optimization process.
Declarations
Authors’ Affiliations
References
- Al-Faruk, A, Hasib, A, Ahmed, N, Kumar Das, U. (2010). Prediction of weld bead geometry and penetration in electric arc welding using artificial neural networks. International Journal of Mechanical & Mechatronics Engineering, 10(4), 19–24.Google Scholar
- AWS. (2010). “Underwater welding code”. USA: AWS.Google Scholar
- Brown, RT, & Masubuchi, K. (1975). “Fundamental Research on Underwater Welding”. In Welding research supplement (pp. 178–188).Google Scholar
- Fahlman, SE. (1988). An empirical study of learning speed in backpropagation. USA: Carnegie Mellon University.Google Scholar
- Johnson, RL. (1997). The Effect of Water Temperature on Underbead Cracking of Underwater Wet Weldments. California: Naval Postgraduate School.Google Scholar
- Juang, SC, Tarng, YS, & Lii, HR. (1998). A comparison between the backpropagation and counterpropagation networks in the modeling of TIG welding process. Journal of Material ocessing Technology, 75, 54–62.View ArticleGoogle Scholar
- Liu, S, Olson, DL, & Ibarra, S. (1993). Underwater Welding. ASM Handbook, 6, 1010–1015.Google Scholar
- Nagendra, SM, & Khare, M. (2006). Artificial neural network approach for modelling nitrogen dioxide dispersion from vehicular exhaust emissions. Ecological Modelling, 190(1–2), 99–115.View ArticleGoogle Scholar
- Negnevetsky, M. (2005). Artificial Intelligence. UK: Addison Wesley.Google Scholar
- Omajene, JE, Martikainen, J, Kah, P, & Pirinen, M. (2014). Fundamental difficulties associated with underwater wet welding. International Journal of Engineering Research and Applications, 4(6), 26–31.Google Scholar
- Shi, Y, Zheng, Z, & Huang, J. (2013). “Sensitivity model for prediction of bead geometry in underwater wet flux cored arc welding”. In Transaction of nonferrous metals society of China (pp. 1977–1984).Google Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.