Reliability Optimization of Hybrid Systems Driven by Constraint Importance Measure Considering Different Cost Functions

Abstract

The requirements of high reliability for hybrid systems are urgent for engineers to maximize the system reliability under the limited cost budget. The cost constraint importance measure (CIM) is an important tool to achieve the local optimal solution by considering the relationship between constraint conditions and objective functions in the optimization problem. To better consider the contribution of the CIM, this paper considers three different cost function forms, including power type, trigonometric type, and exponential type. Combining the global search ability of the arithmetic optimization algorithm (AOA) with the local search ability of the CIM, a CIM-based arithmetic optimization algorithm (CIAOA) is developed to analyze the contribution of the CIM. Through the numerical experiments, the optimal system reliability and convergence generation of the CIAOA and AOA under different cost function forms are regarded as the indexes to analyze algorithm performance. The experimental results show that the average system reliability improvement percentages under power type, trigonometric type, and exponential cost constraint are 8.07%, 0.14%, and 0.53%, respectively, while the average convergence improvement percentages under three cost forms are 37.30%, 0.08%, and 1.66%, respectively. Therefore, the CIAOA performs the best under power cost constraints. Finally, a numerical example of a hybrid power vehicle system is introduced to analyze the contribution of the CIM under different cost functions by considering the reliability improvement rate in the optimal solution and the ranking of the CIM. The higher prioritization components in the two rankings are similar, which shows that the component with higher a CIM is selected to improve its reliability.

Using a Node–Child Matrix to Address the Quickest Path Problem in Multistate Flow Networks under Transmission Cost Constraints

“…While deterministic (non-stochastic) flow networks have undeniably been instrumental in understanding and optimizing various systems, the practical reality is that many real-world systems exhibit dynamic characteristics, necessitating the adoption of a more nuanced approach. Multistate (stochastic) flow networks (MFNs) have gained prominence as a result of their ability to model complex systems in which fluctuations, failures, maintenance, and other dynamic factors play a significant role [14][15][16][17][18][19][20]. Within an MFN, arcs and nodes can exist in various potential states that are influenced by traffic conditions, maintenance activities, failures, or other underlying causes.…”