One or two circular shapes? A binary detection for electrical capacitance tomography sensors

Abstract

AbstractImage reconstruction using engine electrical capacitance tomography sensors rises several drawbacks. Recently, the authors proposed a novel methodology dedicated to a fast detection of the configuration: size and position, of a single circular form using basic algebraic operations. However, this technique is efficient when a single form is present in the sensor cross‐section under interest. The present work proposes 2 different techniques that determine if the sensor cross‐section contains 1 or 2 circular shapes. Both methods are applied to simulated measurements. They operate much differently, and each comes with its own pros and cons. The first method considers that the sensor signal presents an axis of symmetry if a single circular object lies in the sensor. Otherwise, the symmetry is broken. It achieves a recognition rate of 73.9%. In the second method, machine‐learning techniques are used to perform a binary supervised classification. Promising rates of recognition over 99% are obtained. The present study also reveals that a 4‐electrode sensor leads to the best recognition rates with both methods. This result was also established by the authors in a different framework when dealing with physical limitations on spatial resolution in electrical capacitance tomography sensors.