Reliability Modeling Method Based on State Truncation and Error Bound Analysis

Markov model is one of the most widely used methods for system reliability and availability analysis. However, the large state space is a great challenge in constructing Markov model for complex system. This paper provides a simple, but accurate method to build a simplified Markov model (layer-based model). Theoretical analysis shows that the error bound in the layered model can be determined only by failure rate and repair rate. Thus, by performing the error bound analysis on model formulated by adding states layer by layer, we will get the number of layers needed in the model to meet the targeting error requirement. Then, the reliability parameter of the system can be solved on the simplified model. This method not only greatly reduces the time to build the model, but also simplifies the solving process due to the much less states.