Study of Various Conjugate Gradient Based ANN Training Methods for Designing Intelligent Manhole Gas Detection System

Abstract

Human fatality occurs due to presence of excessive proportion of toxic gases such as Ammonia (NH3), Carbon Dioxide (CO2), Carbon Monoxide (CO), Hydrogen Sulfide (H2S), Methane (CH4), and Nitrogen Oxide (NOx) in manholes. To ensure safety of the workers and the environment as well, we are motivated to develop an intelligent sensory system to serve the purpose of predetermination of the aforementioned gases. To design such intelligent sensory system, we are using Soft Computing tools like Artificial Neural Network (ANN) and resort to use Conjugate Gradient (CG) method to offer training to the ANN. In present article, we offer study on CG based ANN training algorithm used in design of an intelligent sensory system for sensing gas components of manhole gas mixture. We offer exhaustive discussion on performance of different variants of CG. We report two new variants of CG which are found to perform better than most of the existing variants of CG applied for the said purpose.