MACHINERY FAULT DIAGNOSIS BASED ON CHAOTIC OSCILLATOR AND APPROXIMATE ENTROPY

摘要

Weak periodic signals can be detected by identifying the transformation of the chaotic oscillator from the chaotic state to the large-scale periodic state when a weak external periodic signal is applied. Approximate entropy and its two-dimensional expression which was proposed by our group have been proved to be an effective measure of the states of chaotic oscillator. This paper discusses and summarizes machinery fault diagnosis based on chaotic oscillator and approximate entropy and its engineering application. In practical engineering measurement, recorded data is generally a mixture of signal and noise, and interested weak signal is usually submerged in heavy noise. For example, the fault characteristic signals of rolling bearings are in low frequency band, and the useful signals are often buried in heavy noise and difficult to be detected. By using the characters of chaotic oscillator being sensitive to weak periodic signals, and approximate entropy being available in recognizing the change of chaotic oscillator, the fault feature could be extracted. Satisfactory results have been achieved when using the presented method to diagnose the fault of rolling bearings.