Designing Efficient Reconfigurable Control Systems Using IEC61499 and Symbolic Model Checking

Abstract

IEC 61499 provides a standardized approach for the development of distributed control systems. The standard introduces a component architecture, based on function blocks that are event-triggered components processing data and signals. However, it gives only limited support for the design of reconfigurable architectures. In particular, handling of several reconfiguration scenarios is quite heavy on this level since a scenario changes the execution model of the system due to requirements. To this end, a new IEC 61499-based model named reconfigurable function blocks (RFBs) is proposed. An RFB processes the reconfiguration events and switches directly to the suitable configuration using a hierarchical state machine model. The latter represents the reconfiguration model which reacts on changes in the environment in order to find an adequate reconfiguration scenario to be executed. Each scenario presents a particular sequence of algorithms, encapsulated in another execution control chart slave which represents the control model of an RFB. This hierarchy simplifies the design and separates the reconfiguration logic from control models. To verify its correctness and alleviate its state space explosion problem in model checking, this paper translates an RFB system automatically into a generalized model of reconfigurable timed net condition/event systems (GR-TNCES), a Petri net class that preserves the semantics of an RFB system. In this paper, along with verification of deterministic properties, we also propose to quantify and analyze some probabilistic properties. As a case study, we consider a smart-grid system, interpreting permanent faults in it as reconfiguration events, and we characterize them with the expected occurrence probability and the corresponding repair time. A tool chain ZiZo is developed to support the proposed approach. Note to Practitioners —For reconfigurable distributed control systems, two models are indispensable: a control model that defines the hardware and software behaviors and a reconfiguration model that manages unpredictable changes in the related environment for configuring accordingly the system behavior. The proposed RFB approach is based on hierarchical state chart specification within function blocks, its automatic conversion to a reconfigurable Petri net GR-TNCES, which models all possible reconfiguration scenarios and a probabilistic model checking for qualitative and quantitative analysis. The system flexibility is ensured by a decision algorithm and a reconfiguration matrix which selects dynamically the right scenario to execute. The approach is supported by ZiZo tool chain which creates and edits reconfigurable function blocks model and converts it automatically to GR-TNCES, and then to specific PRISM models. The detection of the worst cases before deployment is a major virtue of the approach that practitioners need to estimate and enhance the design process. Several properties are easily checked and estimated such as system feasibility before and after reconfiguration, deadlock detection, confluence, estimation of the reconfiguration failure, system availability, and best repair time. The developed software package can be applied in any domain requiring flexibility and failure estimation, such as new medical technologies, transportation systems, smart and microgirds, and manufacturing systems.