Air-discharge Decomposition Gases Detected by Sensor Array with Contrast Learning

Abstract

Before the insulation failure of the air switch cabinet, its partial discharge will decompose the air and form characteristic decomposition products including carbon monoxide (CO) and nitrogen dioxide (NO2). Because of its cross sensitivity, a single semiconductor gas sensor can hardly selectively distinguish these characteristic gases. Sensor array can effectively identify the gas composition, but to reduce the cost and sensor volume, the array needs to be optimized to reduce the number of sensors. In this paper, In this paper, an unsupervised algorithm based on contrast learning is employed to optimize semiconductor gas sensor array, to maintaining gas recognition accuracy while reducing the number of sensors.