Time-Frequency Ridge Analysis using on distinguish regular signal from arc-fault signal

In the past decade, electronic devices are widely used in power supply system. They produce harmonic waveforms which look like chaos. Though a chaotic signal would be caused by arc-fault, it doesn't contain harmonic signal which is regular in nature. Conventional arc detection method failed to distinguish the harmonic signal from the chaotic signal, and the harmonic signal is judged as fault signal. This is because the conventional method analyzes signal solely on time or frequency domain, it losses information from the other domain. In this paper, a novel analytical methodology to distinguish regular signal from real chaos by making use of time-frequency ridges is proposed. This method is based on the knowledge that the non-fault electric signal has a highly regular time-frequency structure and its instantaneous frequency has a regular and stable representation; while the arc-fault doesn't has such properties. What's more, the method based on time-frequency ridges not only offers an effective way to represent instantaneous frequency but also is robust to noise and interference after some processing.