An efficient power set mapping space blocking algorithm for sensor selection in uncertain systems with quantified diagnosability requirements

Rui Sun

doi:10.1007/s10489-022-03290-0

An efficient power set mapping space blocking algorithm for sensor selection in uncertain systems with quantified diagnosability requirements

2022-05-14

2

PRE

AI

DOI：10.1007/s10489-022-03290-0

原文链接

原文求助

摘要

En 中文

For uncertain systems, the impacts of model uncertainties and measurement noise are considered during the modeling process. The distinguishability that is derived from the Kullback-Leibler divergence quantifies fault diagnosability for uncertain systems. Selecting a set of sensors that fulfill the quantified diagnosability requirements with the lowest cost is a significant step of fault detection and isolation in uncertain systems. In this paper, we propose a two-form power set mapping space blocking algorithm(PMSBA) for this problem, which includes three main strategies: stochastic search that is motivated by an inclusion relation strategy (SSMIRS) ,the BILP-based search space blocking strategy (BSSBS) and the best from multiple selections as measured by overlapping space strategy(BMSMOSS). SSMIRS, which is motivated by inclusion relations, is an efficient strategy for searching for local optimal solutions. BSSBS is used to block the spatialce area that is unnecessary to explore. BMSMOSS considers the overlapping space between each blocking, and can only be applied in the complete form of PMSBA. By modifying the parameter N, PMSBA can be transformed from an incomplete algorithm to a complete algorithm. The incomplete form of PMSBA is suitable for large-scale systems. The complete form is suitable mainly for small-scale or medium-scale systems and can find the optimal solution. To evaluate the performance of PMSBA, experiments are performed on four uncertain system instances, namely,two theoretical use cases and two practical use cases. The experimental results show that, compared with a state-of-the-art algorithm, namely, the incomplete form of PMSBA obtains superior solutions and performs better in terms of efficiency. The complete form of PMSBA is the first complete algorithm to be proposed for sensor selection in uncertain systems. It provides great effectiveness advantages over the depth-first search algorithm for most feasible solution spaces.

Keyword:

Model-based diagnosis

Fault diagnosis

Sensor placement

Stochastic search

Fault diagnosability analysis

An efficient power set mapping space blocking algorithm for sensor selection in uncertain systems with quantified diagnosability requirements

摘要

期刊

机构

引用论文