Eigenvalue extraction of current signal during spot welding based on space-phase reconstruction |
| |
Authors: | Liu Pengfei Shah Ping Qin Hede Luo Zhen |
| |
Affiliation: | 1. School of Materials Science and Engineering,Tianjin University,Tianjing 300072 2. School of Mcchatronics Engineering of Xin xiang,Xinxiang,453000 |
| |
Abstract: | Nonlinear methods are used to analyze current signal of spot welding and the minimum embedding dimension, correlation dimension, the optimal time delay and the largest Lyapunov exponent of current signal time series are calculated in this paper. The chaotic character of current signal time series is discovered. Then a chaotic neural network is built and used to predict the future current signal. Means of residual error out of the network are used us eigenvalue of current signal during spot welding. It is shown that spatter can greatly affect the means of residual error of spot welding after analysis, the mean values of output errors of signal contaminated by spatter noise are more than O.08,but the mean values of output errors of the signal with no spatter noise are less than O.04,so mean of residual errors can be employed as the character of spatter. |
| |
Keywords: | spot welding chaotic time series spatter the largest Lyapunov exponent chaotic neural network |
本文献已被 维普 万方数据 等数据库收录! |
|