基于LSTM神经网络算法的大型筒体原材料钛元素定性分析  

Qualitative Analysis of Titanium Element in Large Cylinder Raw Material Based on LSTM Neural Network Algorithm

作　　者：莫堃 张沛 官雪梅 光海杰 熊章伍 MO Kun;ZHANG Pei;GUAN Xue-mei;GUANG Hai-jie;XIONG Zhang-wu(China Dongfang Electric Group Co.,Ltd.,Chengdu Sichuan 610036)

机构地区：[1]中国东方电气集团有限公司,四川成都610036

出　　处：《数字技术与应用》2021年第8期93-95,98,共4页Digital Technology & Application

基　　金：国家重点研发计划(2017YFF0107300)。

摘　　要：钛(Ti)元素可防止不锈钢产生晶间腐蚀,但含钛不锈钢成品的抛光性能差,高精度表面加工难度高。而采用不锈钢做原材料的大型筒体需要高精度的表面加工,且加工过程需反复对尺寸进行测量,对测量仪器性能和测量效率要求较高。因此需要对大型筒体不锈钢原材料进行钛元素定性分析。本文采用人工神经网络(LSTM)算法定性判别不锈钢中钛元素,将光谱分析仪实际测得谱图数据作为训练矩阵,利用Matlab软件构建LSTM模型,建立相干元素的峰值信息和钛元素浓度的关系,得到钛元素的定性判别结果。Titanium(Ti)element can prevent stainless steel from intergranular corrosion,however the polishing performance of titanium-containing stainless steel products is poor,and high-precision surface processing is difficult.The large cylinder made of stainless steel needs high precision surface processing,and the machining process needs to measure the size repeatedly,it requires high performance and efficiency of measuring instruments.Therefore,it is necessary to carry out the qualitative analysis of titanium element in the raw material of large cylinder stainless steel.In this paper,LSTM neural network algorithm is used to qualitatively identify titanium in stainless steel,the spectrum data measured spectrum analyzer is used as the training matrix,and the LSTM neural network model is constructed by using Matlab software to establish the relationship between the peak information of coherent elements and the concentration of titanium,the qualitative results of titanium were obtained.

关 键 词：钛 不锈钢 大型筒体 光谱 算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]