Dongnian Jiang* and Wei Li
International Journal of Control, Automation, and Systems, vol. 20, no. 9, pp.2879-2891, 2022

Abstract : Sensor reconfigurability is the basis of the fault tolerance of a system. In order to improve the fault tolerance and reliability of sensors in the system, a method of quantitative evaluation of sensor reconfigurability based on a data-driven method is proposed. In this method, an analytic redundancy analysis of sensors in the system is carried out, and the analytic redundancy model is established. Based on this, the quantitative evaluation index of sensor reconfigurability is given by using the similarity evaluation method. Because the possible inaccuracy of the evaluation index results in uncertainties of the system, an adaptive threshold is designed to improve the evaluation accuracy. This method can quantitatively evaluate the reconfigurability of the sensor without depending on the system model, which provides a theoretical basis for improving the fault tolerance of the system at the design stage.

Keyword : Analytical redundancy, fault tolerance, Kullback-Leibler divergence, quantitative evaluation, sensor reconfigurability, sparse kernel density estimator.

Download PDF : Click this link