Frictional running-in status identification via the visualization and extracted features of friction-induced attractors

Visualization-based quantitative characterization of high-dimensional trajectories constitutes a fundamental methodology for friction-induced attractor implementation in tribology studies. This study establishes an optimized frictional running-in status identification framework through systematic comparison of three high-dimensional visualization techniques: phase trajectory projection, radar plots, and recurrence plots. The findings suggest that the attractor phase trajectory-based approach constitutes a methodology that is independent of specific wear mechanisms, demonstrating a notable enhancement in characterization accuracy when compared to traditional time-domain approaches. Moreover, phase trajectory projection and radar plots achieve computational efficiency but exhibit limited feature resolution, while recurrence analysis delivers superior dynamic evolution interpretation at higher computational cost. Therefore, a hybrid framework integrating radar plots for rapid state screening and quantitative recurrence analysis for feature extraction achieves a balance between accuracy and efficiency. This resolves the efficiency-accuracy trade-off in tribosystem monitoring.