New tools for distributed diagnosis and preventive maintenance decision support
Résumé
This paper aims at presenting a new approach for supporting preventive maintenance decisions for the availability of Grid Systems (GS). One common limit in large Grid systems is the lack of decision-support information for availability assessment under real-time operating conditions with availability requirements. Grid Systems' global availability can change unexpectedly when the number of components increases under some hardware or software failures and maintenance policies with resource constraints. To meet this challenge, an advanced assessment model of the Grid’s components availability is proposed beyond the classical exponential model. A real-time approach based on two major tools implemented in two phases is then developed to deal with such operating conditions. The first one concerns new information collecting strategy on the components by deploying networked Product Embedded Information Devices (PEID) and agents on the Grid components, using a dynamic matrix model and a set of circulant operators. These operators are new promising tools in collecting and dispatching components’ health information and data. The second phase consists of evaluating the global availability of the system using the collected information. Thus, for a required global availability for the system, it is possible to estimate in real-time the health (effectiveness level)) of each component on the Grid. , , , , , . ,