激光冲击区表面质量人工神经网络的研究.pdfVIP

第 25 卷 　第 1 期 激 　　光 　　技 　　术 Vol . 25 ,No . 1 200 1 年 2 月 LA SER TECHN OL O GY February ,200 1 激光冲击区表面质量的人工神经网络研究 於自岚 　高传玉 　曾丹勇 　杨继昌　张永康 (江苏理工大学机械工程学院 ,镇江 ,2 120 13) 摘要 : 大量的实验表明 ,经激光冲击处理后 ,材料受冲击区的表面质量与材料的疲劳寿命有 着明显的关系 。因此 ,表面质量是判断激光冲击强化效果的重要手段 。将人工神经网络技术用于 激光冲击处理后试件的表面质量分析 ,建立了激光参数与激光冲击处理后试件的表面质量之间的 联系 ,并用其实现了对冲击处理后的试件表面质量的预测 。研究及实验表明 ,该方法不仅具有准 确及稳定性好等特点 ,而且这种预测能力在实际应用中还具有不断提高的智能特性 。 关键词 : 激光冲击处理 　表面质量 　神经网络 Study of the surface qual ities of laser shockprocessing zones using an artif icial neural net work Y u Zil an , Gao Ch uany u , Zeng D any ong , Yang J ichang , Zhang Yongk ang (College of Mechanical Engineering ,J iangsu University of Science and Technolohy ,Zhenjiang ,2 120 13) Abstract : A lot of exp eriment s have shown t hat t here is an obvious relation between surface ( ) qualities of sp ecimen after laser shockprocessing L SP and it s fatigue life . Consequently ,t he L SP effect s can be evaluated by surface qualities in L SP areas. In t his p ap er ,an artificial neural network (ANN ) is utilized to study t he surface qualities of sp ecimen after L SP. Based on t he data obtained in t he exp eriment ,an ANN is established . The trained ANN could acquire t he relations between surface qualities and laser p armeters. From t he verification of aluminium alloy 2024T62 , it is proved t hat t he neural network can successfully predict t he surface quality grades of sp ecimen after L SP ,and easily determine t he laser p arameters under different production conditions. The research and exp erimental result s show