Performance degradation model and prediction for industrial robot based on FMECA and JSLSTM method

Abstract

Equipment performance degradation analysis and precision prediction is a critical aspect of prognostics and health management (PHM) that can significantly improve production efficiency and quality. This study proposes a method that combines the analytic hierarchy process with failure mode, effects and criticality analysis (FMECA) to analyze the failure causes of industrial robot. By using an interval fuzzy mathematical algorithm, 34 types of failure modes were quantitatively analyzed, and the primary degradation performance indicators were identified. In addition, a novel long short-term memory deep learning method optimized by an artificial jellyfish search optimizer (JSLSTM) is proposed. This model can adaptively adjust its parameters to address overfitting or underfitting by updating online during the prediction process. The proposed performance degradation prediction algorithm was evaluated using performance degradation data of industrial robot. Experimental results demonstrated that the JSLSTM neural networks achieved repeatable predictions with an error margin below 0.06 μm across 18 different datasets.