A unified moment approach for time-independent and time-dependent reliability analyses of robotic systems

Reliability of robotic systems throughout their entire service life is essential for ensuring the manufacturing quality of products in future factories. For this purpose, this study proposes a unified analysis of both static and dynamic reliability for robotic systems based upon a moment-based approach. Firstly, fractional moments are calculated to quantify the uncertainty in the output response of motion performance. A flexible mixture probability distribution model is then developed by weighting the log skewed-normal distribution model, enabling more precise modeling of asymmetric distributions. Two fractional moment-informed calibration strategies for the probability model are proposed by integrating the optimization and Bayesian methods, which meticulously encapsulate uncertainty in robotic performances. Subsequently, the calibrated probability model is applied to accurately reconstruct the probability distribution and compute the reliability of the static response in robotic systems. On the other hand, the time-dependent failure probability is calculated by employing the calibrated probability model at different time instances and the system reliability theory. Two numerical examples and four practical engineering cases of robotic systems are implemented to demonstrate the computational performance of the proposed method. The results show that the proposed method can accurately and efficiently evaluate the time-independent and time-dependent reliability of robotic systems.