We live in a "distributed world" made by countless "nodes", being them cities, computers, people, etc., connected by a dense web of transportation, communication, or social ties. The term "network", describing such a collection of nodes and links, nowadays has become of common use thanks to our extensive reliance on "connections of interdependent systems" for our everyday life, for building complex technical systems, infrastructures and so on. In an increasingly "smarter" planet, it is expected that systems are safe, reliable, available 24/7, and possibly at low-cost maintenance. In this connection, monitoring and fault diagnosis are of customary importance to ensure high levels of safety, performance, reliability, dependability, and availability. In fact, faults and malfunctions can result, just referring to industrial plants, in off-specification production, increased operating costs, chance of line shutdown, danger for humans, detrimental environmental impact, and so on. Faults and malfunctions should be detected promptly and their source and severity should be diagnosed so that corrective actions can be possibly taken.
This lecture deals with a on-line approximation-based distributed fault diagnosis approach for large-scale nonlinear systems, by exploiting a "divide et impera" approach in which the overall diagnosis problem is decomposed into smaller subproblems, simpler enough to be solved within the existing computation and communication architectures. The distributed detection, isolation and identification task is broken down and assigned to a network of "Local Diagnostic Units", each having a "different/local view" on the system: they are allowed to communicate with each other and also to cooperate on the diagnosis of system components that may be shared thus yielding a global diagnosis decision. In the lecture, issues and perspectives will be addressed as well in a paradigmatic industrial context of safety-critical process control systems.