Impact Damage Identification for Composite Material Based on Transmissibility Function and OS-ELM Algorithm

Abstract

A method is proposed based on the transmissibility function and the Online Sequence Extreme Learning Machine (OS-ELM) algorithm, which is applied to the impact damage of composite materials. First of all, the transmissibility functions of the undamaged signals and the damage signals at different points are calculated. Secondly, the difference between them is taken as the damage index. Finally, principal component analysis (PCA) is used to reduce the noise feature. And then, input to the online sequence limit learning neural network classification to identify damage and confirm the damage location. Taking the amplitude of the transmissibility function instead of the acceleration response as the signal analysis for structural damage identification cannot be influenced by the excitation amplitude. The OS-ELM algorithm is based on the ELM (Extreme Learning Machine) algorithm, increased training speed also increases the recognition accuracy. Experiment in the epoxy board shows that the method can effectively identify the structural damage accurately.

Civil Engineering Structural Damage Identification by Integrating Benchmark Numerical Model and PCNN Network

“…This method has good robustness and generalization in structural damage identification, and is less affected by noise data interference . Sun Y [6] et al proposed a combination of transfer function and sequential Extreme learning machine to detect impact damage of composite materials, and input the structure with principal component analysis for feature noise reduction into the neural network to determine the damage location. This method has good accuracy in structural damage identification .…”

“…In formula (6), I represents the length of the sliding window, z represents the number of window data points, f represents the sampling frequency, and 1 f represents the Fundamental frequency of the original vibration response. Standardize feature values of different magnitudes, that is, handle outliers with a zero mean, and the expression is Eq.…”