RESEARCH ON DIAGNOSING FATIGUE DAMAGE BASED ON WAVELET-NEURAL NETWORK
The paper makes approaches on the analysis and treatment of the acoustic emission (AE) signals and the diagnosis of fatigue damage by combining wavelet analysis with neural network loose. It makes recognition calculations in Matlab6.5 by determining the main factors of the wavelet-neural network reasonably and taking the lower crossbeam of the vibrating screen as an object of study, picking up separately four groups of AE signal data related to accruing fatigue flaw and another four groups related to the ground noise of testing ground when the lower crossbeam is working well. The results have shown that it is an effective fault diagnosismethod of fatigue damage for metal structures that AE signals are processed and characteristic vectors are extracted by the use of the wavelet packet energy method and the fault pattern recognition is done by the use of a BP neural network.
Wavelet-Neural Network Fault Diagnosis Fatigue Damage
Li Zigui Yan Bijuan
Taiyuan university of science and technology, Taiyuan ,030024,China Taiyuan university of science and technology, Taiyuan ,030024,China.
国际会议
北京
英文
927-933
2008-10-27（万方平台首次上网日期，不代表论文的发表时间）