Hybrid, recursive, nested monitoring of control systems using Petri nets and particle filters

In this paper we propose a recursive and nested hybrid monitoring and diagnosis architecture for systems with Recursive Nested Behaviour-based Control structure. Such systems consist of behavioural levels, which use several models on different levels of abstractions. In this architecture scheme, the monitors of each subsystem use recursively the output of the monitors of the next lower level in order to get an estimate of the global status of the system at each time and having the advantage of low dimensionality for each level. The proposed monitoring structure is based on hybrid Petri Nets and particle filters. The advantages of the approach are illustrated by an example for a Heating Control System.