VoI-Based Optimal Sensors Positioning and the Sub-Modularity Issue

Abstract

The problem of sensor positioning for condition monitoring of Systems, Structures and Components (SSCs) has been recently proposed to be addressed by Value of Information (VoI) optimization. VoI is a metric that quantifies the benefit of taking a measurement prior to adopting it. This metric lacks the characteristics of sub-modularity, i.e. the benefit of adding a measurement to a small set of measurement is higher than adding it to a bigger set. This causes the VoI optimization to not guarantee optimal results when the problem is solved by greedy optimization algorithms. In this work, the sub-modularity issue is considered with reference to the thickness gauge sensor positioning on a Steam Generator (SG) of Nuclear Power Plant (NPP), and ways forward to overcome the sub-modularity issue are suggested.