基于弱周期多通道信号正态重复率理论的CO_2弧焊监测  

CO_2 arc welding detection based on theory of weak periodic and multi-channel signals with normal repetition rate

作　　者：高理文[1,2] 薛家祥[1] 张文[1] 林放[1] 陈小峰[1] 

机构地区：[1]华南理工大学机械与汽车工程学院,广州510640 [2]广州中医药大学信息技术学院,广州510006

出　　处：《焊接学报》2011年第11期29-32,114,共4页Transactions of The China Welding Institution

基　　金：国家自然科学基金资助项目(50875088);广东省科技计划资助项目(2010B010700001);番禺区科技计划资助项目(2010-Z-22-1);黄埔区科技计划资助项目(1021)

摘　　要：受到弧焊稳定性自动化监测需求的驱使和启发,创新性地提出了弱周期多通道信号正态重复率理论.该理论通过科学的数学推理,推导出描述周期间信号重复程度的特征值计算方程.为了克服实际采样时信号离散性带来的问题,突破性地把一个信号点细分为以正态分布散落在多维信号空间的无数微点,从而合理地计算方程中的重要参量——广义实面积,最终准确求解方程.采集了104次CO2弧焊焊接过程的电信号作为样本,以概率神经网络为决策方法,进行了训练和预测试验.结果表明,依照该理论所得的特征值与CO2弧焊的稳定性具有较大的相关性,能实现自动化的监测.Aiming at the automatic detection on arc welding stability,this paper proposed the theory of weak periodic and multi-channel signals with normal repetition rate.Through a scientific mathematical deduction,an equation has been put forward,which can sufficiently demonstrate the dynamic characteristics of the repetition rate of periodic signal.To overcome the problems brought in by scattered signals during actual sampling,this paper has introduced a method by subdividing a signal point into countless micro points which normally distribute in a multi-dimensional signal space.Therefore,an important parameter of the equation,the generalized actual area could be reasonably calculated,through which the whole equation could be accurately solved.104 times of arc welding electric signals were conducted as samples.Based on the probabilistic neural network(PNN),the samples has been trained and tested,which show that the dynamic characteristics drawn through the above-mentioned theory are closely related to the stability of CO2 arc welding,and thereby the automatic detection can be realized.

关 键 词：CO2弧焊 稳定性 多通道信号 正态分布 

分 类 号：TG409[金属学及工艺—焊接]