Water and Ice Detection in Marine Icing by Capacitive Sensor Array and the Artificial Neural Network Model

Abstract

This work describes a capacitive sensor array for the measurement of marine ice accretion. Currently, all commercial icing detectors only track the ice phase disregarding the water phase. This approach is acceptable in many applications, mostly in the atmospheric icing domain however in the case of marine icing, the water phase creates a significant component of the measurement signal. The water phase cannot be disregarded and requires a novel methodology that takes into account the water-ice conditions simultaneously. A neural network approach is used to implement the signal-to-measurand mapping. Compared to the previous least-squares based mapping, the neural network method is more accurate and therefore becomes a preferred method of signal processing in capacitive array based marine icing sensing.