炭素电极损伤的敲击响应粗糙集分析检测方法  

Detection Mechanism Analysis of Coin-Tapping Response Signals of Carbon Electrodes

作　　者：曹晓琳[1] 王双维[2] 魏志刚[3] 沈晓波[2] 高志华[2] 侯英哲[2] 

机构地区：[1]吉林大学汽车工程学院,长春130025 [2]东北师范大学物理学院,长春130024 [3]中钢集团吉林炭素股份有限公司,吉林市132002

出　　处：《噪声与振动控制》2010年第4期138-143,共6页Noise and Vibration Control

摘　　要：利用小波分解和粗糙集研究敲击法炭素电极检测技术中的敲击后响应信号,得到了炭素电极损伤的敲击响应粗糙集分析检测方法。文章利用db9小波对敲击后响应信号进行8层分解,对得到的9个小波通道利用粗糙集删除各冗余属性通道,根据删除后知识系统的描述精度确定相应属性的重要度。研究结果表明其中D5、D6、D7、D8四个通道为鉴别正品与次品的有效通道。在研究中提出了二元属性重要度分析的方法,以将重要度为零的可简约属性成对删除,提高属性简约效率,找到知识表达系统的核。并通过机理分析证明,正品与次品敲击信号差异主要来自对产品特征频率基频和低次谐波的共振,从理论上解释了碳素电极敲击法的检测原理。Wavelet and rough sets are used to analyze the detection mechanism of coin-tapping response signals from carbon electrodes.After 8 layers decomposition of the response signals using db9 wavelet basic function,9 wavelet channels are obtained.Then,rough sets are used to delete the redundant character channels in the 9 wavelet channels.The results show that channels D5,D6,D7 and D8 are the effective character channels to distinguish defective electrodes from quality electrodes.Furthermore,research on detection mechanism shows that the difference of coin-tapping response signals between defective and quality electrodes is distinguished from the difference of resonance phenomena in base frequency and low order harmonic wave.In addition,an effective method for binary attribute importance analysis,which can delete redundant attributes with null importance in pairs,is proposed.And the core of knowledge expression system can be found with this method.

关 键 词：振动与波 炭素电极 敲击 检测机理 小波 粗糙集 

分 类 号：TH89[机械工程—仪器科学与技术]