TiB_2-Cu复合材料SHS工艺的人工神经网络优化  

Optimization of TiB_2-Cu composite self-propagating high-temperature synthesis(SHS) process parameters by artificial neural network

作　　者：徐强[1] 张幸红[1] 赫晓东[1] 韩杰才[1] 王雪峰[2] 

机构地区：[1]哈尔滨工业大学复合材料研究所,黑龙江哈尔滨150001 [2]哈尔滨工业大学数学系,黑龙江哈尔滨150001

出　　处：《哈尔滨工业大学学报》2003年第10期1199-1201,共3页Journal of Harbin Institute of Technology

基　　金：教育部博士点基金资助项目(20020213037);航天创新科技基金资助项目资助项目.

摘　　要：利用3×5×1的3层BP神经网络模型对SHS法制备TiB2-Cu基复合材料过程中的工艺参数进行优化.将SHS工艺的3个主要参数———延迟时间、高压压力和高温保压时间作为人工神经网络的输入,合成产物的相对密度作为网络的输出,利用单参数动态搜索算法对SHS工艺参数进行了优化.结果表明训练样本和检验样本的网络实际输出值与相应的试验值均非常接近,TiB2-Cu基复合材料的SHS最佳工艺参数是延迟时间为6 8s,高压压力为360MPa,高温保压时间为9 2s.With the three main SHS parameters, time delay, high pressure and high pressure holding time, used as the inputs to the neural network, and the relative density of TiB2-Cu composite as the output from the neural network, the SHS process parameters of TiB2-Cu composite have been optimized by single parameter dynamic search algorithm (SPDS). Test results show that the outputs of both training and check-out samples from the neural network are very close to the corresponding test results, and the optimal SHS process parameters of TiB2-Cu composite are time delay of 6.8 s, high pressure of 360 MPa, and high pressure holding time of 9.2 s.

关 键 词：硼化钛-铜复合材料 SHS 工艺优化 人工神经网络 自蔓延高温燃烧合成 模型 

分 类 号：TB333[一般工业技术—材料科学与工程]