CAJ | 학술논문

将人工神经网络引入镀锌钢板点焊的质量预测中，将实验得到的大量点焊规范参数与相应点焊接头质量的试验数据提供给神经网络学习，通过其非线性映射的泛化能力自动抽取所学习知识的特征，描述点焊规范参数空间与焊点接头质量空间的映射关系.结果表明，学习后的神经网络的预测曲线能够合理地预测焊接电流对镀锌钢板点焊的焊点熔核直径与拉剪强度的影响规律，恰当地反映点焊的基本原理，说明人工神经网络在镀锌板点焊性能估测中的应用是可行的.
장인공신경망락인입도자강판점한적질량예측중，장실험득도적대량점한규범삼수여상응점한접두질량적시험수거제공급신경망락학습，통과기비선성영사적범화능력자동추취소학습지식적특정，묘술점한규범삼수공간여한점접두질량공간적영사관계.결과표명，학습후적신경망락적예측곡선능구합리지예측한접전류대도자강판점한적한점용핵직경여랍전강도적영향규률，흡당지반영점한적기본원리，설명인공신경망락재도자판점한성능고측중적응용시가행적.
Artificial neural networks (ANN) were used in quality predicting in the spot welding of zinccoated steel sheets. Lots of experimental data about welding parameters and corresponding spot weld quality were provided to the ANN. Through the generalization of nonlinear mapping, the characteristics of the learnt knowledge were summarized and the mapping relationship between welding parameters and welding quality was described by ANN after studying. The results show that the influence of welding currents on nugget diameters or shear loads of spot welds can be predicted by ANN, and the predicted curves made by ANN conform to the principles of resistance spot welding. So the application of ANN in quality predicting of zinc-coated steel sheets is feasible.