Noise Reduction Technology of Random Vibration Signal Based on Singular Entropy Theory

Structural random vibration effective information extraction is one of the most important issues to study the dynamic characteristics of engineering structures. Combined with singular entropy theory, an effective method of noise reduction for random vibration signal is put forward. Construct analog signal with varying degrees of noise, and the noise is reduced based on the wavelet method, IIR numeric filter, and singular entropy technique separately, and also the SNR index is calculated separately after noise reduction. Study shows that this method can determine the proper order of singular spectrum of the random vibration signal, and ensure the integrity of information contained in the signal after noise reduction and the maximum SNR improvement to a large extent. This is essential to extract the accurate signal feature and reliable analysis.