A fault detection method based on wavelet theory for satellite power systems

Considering that the temperature sensors of satellite power systems are sensitive to the noise, a fault detection method based on the theory of wavelet entropy and the wavelet multi-scale products is purposed to reduce the influence of the noise. Wavelet entropy is applied to amplify the signal-noise ratio by finding the optimal de-noising threshold. Besides, by using of the multi-scale products, the proportion coefficients are further magnified. Then the fault time could be determined by the modulus maximum. Finally, simulation for a class of satellite power systems is conducted, verifying the effectiveness of the proposed method.