Color Recurrence Plots from Uniform Delay Embeddings for Bearing Degradation Tracking and Prognostics.

Area of Science:

• Mechanical Engineering
• Data Science
• Signal Processing

Background:

• Prognostic health management (PHM) for rolling element bearings is crucial for industrial maintenance.
• Effective feature representations are needed for real-time degradation tracking.
• Recurrence-based vibration analysis offers potential but faces computational challenges.

Purpose of the Study:

• To investigate uniform time-delay embedding as a computationally efficient substitute for non-uniform embedding in vibration analysis.
• To develop a practical method for bearing condition monitoring and remaining useful life (RUL) prediction.
• To compare the effectiveness of color recurrence plots versus binary recurrence plots for RUL estimation.

Main Methods:

• Empirical investigation of uniform versus non-uniform time-delay embedding for phase-space reconstruction of bearing vibration signals.
• Construction of color recurrence plots from uniformly embedded phase spaces.
• Application of color recurrence plots for RUL prediction using the Intelligent Maintenance Systems (IMS) bearing dataset.

Main Results:

• Optimally chosen uniform delay vectors provide phase-space reconstructions comparable to non-uniform methods.
• Uniform embedding significantly simplifies parameter selection for vibration analysis.
• Color recurrence plots effectively track bearing degradation and enable reliable RUL prediction, outperforming binary recurrence plots.

Conclusions:

• Uniform time-delay embedding is a near-optimal and computationally tractable approach for recurrence-based vibration analysis.
• Color recurrence plots, derived from uniform embeddings, offer an efficient and deployable method for industrial predictive maintenance.
• This approach enhances condition monitoring by revealing global degradation trends obscured by local instabilities in traditional methods.