Self-x integrated sensor circuits immune to measurement noise in the presence of input perturbation by using robust optimization

Abstract

This paper aims to propose a new technique to extend the performance of the reconfigurable self-x sensory system for industry 4.0 to efficiently obtain robust solution even in the presence of uncertainty both in the input and output stage. Variance measure is employed to handle the uncertainty in the input stage or search space. As far as measurement or objective space, uncertainty is concerned archive-based method applied, and it does not demand any additional computational resources. The traditional evolutionary algorithm, i.e., particle swarm optimizer (PSO), has been modified by expanding its selection process with the proposed solutions. The performance of the extended algorithm is undertaken to study on three benchmarking functions in the presence of uncertainties. The extrinsic evaluation of the proposed algorithm is also performed on the Miller operational amplifier, which is a fundamental part of sensory systems for industry 4.0. Drift due to fabrication process tolerances and aging effects of the transistors is modeled as input uncertainty of the operational amplifier, and imperfect observer (sensor or analog to digital converter) is modeled as output uncertainty. The application confirms the worthiness of proposed uncertainty handling algorithm for industry 4.0.