With the development of machine learning, data-driven mechanical fault diagnosis methods have been widely used in the field of PHM. Due to the limitation of the amount of fault data, it is a difficult problem for fault diagnosis to solve the problem of unbalanced data sets. Under unbalanced data sets, faults with little historical data are always difficult to diagnose and lead to economic losses. In order to improve the prediction accuracy under unbalanced data sets, this paper proposes MeanRadius-SMOTE based on the traditional SMOTE oversampling algorithm, which effectively avoids the generation of useless samples and noise samples. This paper validates the effectiveness of the algorithm on three linear unbalanced data sets and four step unbalanced data sets. Experimental results show that MeanRadius-SMOTE outperforms SMOTE and LR-SMOTE in various evaluation indicators, as well as has better robustness against different imbalance rates. In addition, MeanRadius-SMOTE can take into account the prediction accuracy of the overall and minority class, which is of great significance for engineering applications.

Fierce global competition drives automated manufacturing systems (AMSs) to be increasingly complex, which poses significant challenges on performance analysis and production control. The multi-stage production via serial stations will lead to the propagation of failures in AMSs, which will affect system performance by triggering complex competitions among multiple failure modes. Although machine performance and product quality have been considered, very little has been done to investigate the effect of imperfect quality inspection on competing failures. Focusing on a time balance serial AMS, this paper presents a new competing failure model to investigate the complex interactions among machine failures, product quality, and inspection process, which enables the characterizations of time-delayed propagation of failure, accumulation of degradation, and dynamics of states in serial AMSs. In order to further analyze the impact of competing behaviors on system performance, we have also developed decision-diagram models and algorithms, that are evaluated and validated on serial AMSs with imperfect inspection, revealing the characteristic of multi-state interactions. Experimental results show the proposed methods have strong potentials for performance modeling and analysis of serial AMSs, and also demonstrate general applicability for manufacturing decision making.

The reliability optimization problem arises along with the increasing demands for products’ performance in practical engineering. Importance measures are capable of selecting critical components to gain the greatest improvement in system reliability with the constraint of maintenance cost. The characteristics of importance measure for four kinds of typical systems are discussed to illustrate the usage of importance measure in the practical reliability optimization. A reliability optimization model is eventually established and importance measure–based genetic algorithms are developed to solve the reliability optimization problem efficiently. Finally, two numerical experiments are implemented based on the smoke alarm systems. Experiment I is to illustrate the effectiveness of the importance measure–based genetic algorithms compared with standard genetic algorithm. Finally, the relationship between the order of component reliability improvement and the component parameters is analyzed in Experiment II.