Flatness detection research based on the string vibration method

In order to better identify the flatness pattern, this paper proposes a non-contacted flatness detection called string vibration method. This method first use the principle of string vibration to design the detection scheme, and then use the wavelet packet decomposition to extract the effective shape characteristics, finally use the RBF neural network with strong classification ability to come true the shape pattern classification. By building shape detection platform, getting a lot of experimental data, the relationship between wavelet node vector and shape pattern was established. The plate-shaped pattern recognition rate reached 90%. The method improves the shortcomings of FFT in vibration signal feature extraction, the idea of string vibration method provides a new solution for the flatness detection.