Interface morphology inversion-based performance degradation staged comprehensive evaluation for wet friction transmission component

Abstract

Wet clutches are used to transmit torque between transmission friction pairs in armored vehicles, heavy machinery, and other mechanical devices. Their safe and stable operation plays a crucial role in power transmission. Among all the influencing factors, the interface morphology of wet friction components directly influences both the transmission efficiency and the service life of the clutch, which makes the study of interface morphology particularly significant for these wet friction components. This paper designs a disc-to-disc experiment to obtain interface morphology data of wet friction components under various working conditions. These interface morphology data then serve the purpose of being sample input for the Newton-Raphson-Based Optimizer-Bidirectional Long Short-Term Memory-Attention (NRBO-BiLSTM-Attention) inversion model, allowing for the prediction of the interface morphology. Further, the study derives several key health evaluation indicators for the transmission interface of wet friction components, specifically steepness (S_ku), roughness (R_a), coefficient of friction (μ), and the coefficient of variation (CV). Ulteriorly, by applying a subjective-objective weighting method, the study determines the appropriate weights for each indicator and establishes a performance degradation staged comprehensive evaluation system to evaluate the health of the friction components. The health status of the friction component is ultimately divided into five stages: Healthy Stage, Good Stage, Moderate Stage, Alert Stage, and Fault Stage. The results indicate that the R² value of the NRBO-BiLSTM-Attention inversion model predictions is 0.9719, which demonstrates the model’s effectiveness and high accuracy. Furthermore, the performance degradation staged comprehensive evaluation system effectively evaluates the health of the transmission interface, achieving an R² of 0.9758 in relation to the actual degradation curve. This outcome provides a robust basis for enhancing the reliability of friction components.