Temporal causal networks for simulation and diagnosis

Abstract

The paper is devoted to presentation of temporal causal networks (TCN) for representing and dealing with causal dependencies propagation over time. A temporal causal network is a causal network incorporating explicit representation of intervals of time during which its symptoms/nodes are valid, not valid, and unknown. The temporal knowledge is represented with use of the so called characteristic functions formalism, which allows for classical binary valued logic specification and can be easily extended towards multivalued or fuzzy logic. The presented approach makes use of specific time constraints propagation algorithm for determining causality propagation over the network in time. This is done by evaluating the characteristic functions used as means for representing time constraints for network nodes. The main application includes simulation, monitoring and elements of diagnostic reasoning for dynamic systems with explicit time representation.