Application of On-Line Signal Recovery to Feedwater Control System

An online signal recovery from a high noise environment is applied to a feedwater control system in nuclear steam generators, which is based on the wavelet denoising method. The wavelet denosing method proposed by Mallat, et al., has shown a superior performance for a signal extraction under a severe noise effect, but this method requires massive computations and a manual inspection of the trajectory of the modulus maxima. This paper proposes a systematic approach for finding the modulus maxima of a signal component under a dyadic decomposition, which is much cheaper for the computation than the other existing methods. From this method, the signal is extracted from a highly fluctuating fluid-flowing signal induced by a measurement uncertainty in a steam generator. The feedwater controller based on this method can control the water level safely through a water level control simulation