Bayesian-Network-Based Reliability Analysis of PLC Systems

Reliability analysis is an important part of safety critical programmable logic controller (PLC) systems. The complexity of PLC system reliability analysis arises in handling the complex relations among the hardware components and the embedded software. Different embedded software types will lead to different arrangements of hardware executions and different system reliability quantities. In this paper, we propose a novel probabilistic model, called the hybrid relation model (HRM), for the reliability analysis of PLC systems. Its construction is based upon the execution logic of the embedded software and the distribution of the hardware components. We prove the constructed HRM to be a Bayesian network (BN) that captures the execution logic of the embedded software. Then, we map the hardware components to the corresponding HRM nodes and embed the failure probabilities of the hardware components into the well-defined conditional probability distribution tables of the HRM nodes. With the computational mechanism of the BN, the HRM handles the failure probabilities of the hardware components as well as the complex relations caused by the execution logic of the embedded software. Experiment results demonstrate the accuracy of our model.