Sensor selection for fault diagnosis in uncertain systems
(English)Manuscript (preprint) (Other academic)
The goal of this work is to find the cheapest set of sensors such that a designed diagnosis system can achieve required fault detectability and isolability performance. Algorithms have been developed that find sets of sensors that makes faults detectable and isolable under ideal circumstances. However, due to model uncertainties and measurement noise, different sets of sensors result in different achievable diagnosability performance in practice. Here, a quantitative measure of diagnosability performance, called distinguishability, is used to quantify diagnosability performance given a set of sensors. The sensor selection problem is then formulated, using distinguishability, to assure that the set of sensors fulfills required performance specifications also when model uncertainties and measurement noise are taken into consideration. However, the algorithms that can be employed for finding the optimal solution are intractible, and it is demonstrated why it is hard to find optimal solutions to the sensor selection problem without exhaustive search. Therefore, the use of a greedy stochastic search algorithm is proposed to solve the sensor selection problem. A case study is used to show how the greedy stochastic search is able to find sets of sensors close to the global optimum in short computational time.
Electrical Engineering, Electronic Engineering, Information Engineering Computer Engineering
IdentifiersURN: urn:nbn:se:liu:diva-117176OAI: oai:DiVA.org:liu-117176DiVA: diva2:806672