|Table of Contents|

Nonlinear Combination Prediction of Mechanical Properties of CarbonSteel Electrode Deposited Metal Based on Neural Network


Research Field:
Publishing date:


Nonlinear Combination Prediction of Mechanical Properties of CarbonSteel Electrode Deposited Metal Based on Neural Network
HUANG JunXU Yue-lan
School of Materials Science and Engineering,NUST,Nanjing 210094,China
genetic algorithm neural network carbon electrodes deposited metals nonlinearcombination prediction
TG115. 28
To improve the prediction accuracy of the elongation percentage and the impacting work ofthe carbon steel electrode deposited metal,a nonlinear combination predicting neural network modelis build based on predicted data acquired from the single predicting models of back propagation(BP),radial basis fundion(RBF) and adaptive fuzzy neural network(AINN). By using the geneticalgorithm to optimize the connection weight of BP network,the model predicting property is improvedeffectively. Fifty-five samples are selected from experiments to train and verify the model. Resultsshow that predicted average relative errors of the elongation percentage and the impacting workdecrease to 3. 15% and 2. 67% separately,both of them are far below 5% and satisfy the practicaldemands. Compared with the single predicting model,the accuracy and generalization ability of theneural network combination model based on the genetic algorithm(GA)are greatly improved.


[1]Sukhomay Pal,Surjya K P,Arun K S. Artificial neuralnetwork modeling of weld joint strength prediction of apulsed metal inert gas welding process using arc signals[J]. Journal of Materials Processing Technology,2008,202(6):464-474.
[2] 魷scar Mart侏n, Pilar De Tiedra, Manuel L佼pez, et al.Quality prediction of resistance spot welding joints of304 austenitic stainless steel [ J ]. Materials andDesign,2009(30):68-77.
[3] Hasan Okuyucu,Adem Kurt,Erol Arcaklioglu. Artificialneural network application to the friction stir welding ofaluminum plates[J]. Materials and Design,2007(28):78-84.
[4] 董俊慧,张艳飞,唐正魁. 利用模糊神经网络技术预测焊接接头力学性能[ J]. 焊接学报,2008,29(7):29-33.Dong Junhui,Zhang Yanfei,Tang Zhengkui. Predictionof welding joint mechanical properties by using fuzzyneural network[J]. Transactions of the China WeldingInstitution,2008,29(7):29-33.
[5] Botlani Esfahani M,Toroghinejad M R,Key Yeganeh AR. Modeling the yield strength of hot strip low carbonsteels by artificial neural network[ J]. Materials andDesign,2009,30(8):3653-3658.
[6] Huang Jun,Xu Yuelan. Combination predicting neuralnetwork model for E4303 electrode mechanicalproperties [ A ]. The Proceedings of the SixthInternational Conference on Natural Computation[C].Yantai: Institute of Electrical and ElectronicsEngineers,2010:1610-1613.
[7] Sumantra Mandal,Sivaprasad P V,Venugopal S,et al.Artificial neural network modeling to evaluate andpredict the deformation behavior of stainless steel typeAISI 304L during hot torsion [ J ]. Applied SoftComputing,2009,6(9):237-244.
[8] 王硕,张有富,金菊良. 基于BP-AGA 的非线性组合预测方法研究[J]. 中国工程科学,2005,7(4):83-87.Wang Shuo, Zhang Youfu, Jin Juliang. Research onnonlinear combination forecasting approach based onBP-AGA[J]. Engineering Science,2005,7(4):83-87.
[9] 崔吉峰,乞建勋,杨尚东. 基于粒子群改进BP 神经网络的组合预测模型及其应用[J]. 中南大学学报,2009,40(1):190-194.Cui Jifeng, Qi Jianxun, Yang Shangdong. Combinedforecasting model based on BP improved by PSO and itsapplication [ J]. Journal of Central South University(Science and Technology),2009,40(1):190-194.
[10] 董志波,魏艳红,占小红,等. 遗传算法与神经网络结合优化焊接接头力学性能预测模型[J]. 焊接学报,2007,28(12):69-72.Dong Zhibo, Wei Yanhong, Zhan Xiaohong, et al.Combination of genetic algorithm and neural networkoptimizing welding joint mechanical propertiesprediction model [ J ]. Transactions of the ChinaWelding Institution,2007,28(12):69-72.


Last Update: 2012-11-26