Fault identification method based on Hilbert-Huang transform and support vector machine

Abstract

The invention discloses a fault identification method based on Hilbert-Huang transform and a support vector machine. The fault identification method comprises the following steps: acquiring power utilization state data of a user; performing wavelet packet processing on the user power utilization state data to obtain a current signal without high-frequency interference noise; performing filtering processing on the current signal without the high-frequency noise interference signal through a wave trap to obtain a denoised and filtered current signal; using empirical mode decomposition to decompose the denoised and filtered current signal to obtain an intrinsic mode function; current signal features are extracted from the intrinsic mode function; carrying out fault identification on the extracted current signal features through a classifier; and the arc fault can be accurately identified.

Description

technical field [0001] The invention relates to the technical field of arc fault identification, in particular to a fault identification method based on Hilbert-Huang transform and support vector machine. Background technique [0002] According to the location of the arc, the fault arc can be divided into series fault arc and parallel fault arc. Parallel fault arc is mainly caused by overload and short circuit, and the effective value of current is large. Traditional protection devices such as circuit breakers, fuses and residual current protective devices in low-voltage distribution lines can provide effective protection for parallel arcs. However, when a series arc fault occurs, the circuit is still connected, and various loads are flammable and can operate. The arc current is generally less than the normal operating current of the line due to the limitation of the line load, and the arc current cannot reach the traditional protection devices such as circuit breakers or fu...

AI Technical Summary

Problems solved by technology

Currently commonly used analysis methods include short-time Fourier transform and wavelet transform, but they still use Fourier transform as the final theoretical basis, which shows the shortcomings and limitations of Fourier transform, and cannot simultaneously analyze from the frequency domain and time domain. get high-precision information

The short-time Fourier transform is not suitable for simultaneous analysis of transient signals with different rates of change and extension ranges. Although the time and frequency resolution of wavelet transform can be changed, it is suitable for local analysis of transient signals, but its basis functions and decomposition scales are different. Deterministic impact on accurate analysis of signals

Images