Data-Driven Behaviour Model Recovery Method for Finite-State Transition Model

The IEC 61499 standard provides a component-based system-level modeling language for future-proof automation systems. Automatic generation and migration of function block finite-state transition models encounter problems of the original black-box system and state explosion, especially for the FB network containing multiple types of input/output events and data in interfaces. This paper proposes a recovery method that automatically generates finite-state transition models of the FB network from unlabeled data. Type identification, input-output pairing, and correlation analysis are applied before state transition model generation so as to identify system structure and reduce state space. The experiment is implemented in a simulation system of traffic lights control.