Sparse Optimistic Based on Lasso-LSQR and Minimum Entropy De-Convolution with FARIMA for the Remaining Useful Life...

Area of Science:

• Mechanical Engineering
• Data Science
• Signal Processing

Background:

• Predicting remaining useful life (RUL) is crucial for reducing maintenance costs and ensuring machinery operational safety.
• Long-term health monitoring systems require robust methods for analyzing complex operational data.

Purpose of the Study:

• To introduce a novel hybrid method for accurate RUL prediction in machinery health management.
• To improve the reliability and efficiency of long-term machinery health monitoring.

Main Methods:

• Utilized optimized Lasso and Least Square QR-factorization (Lasso-LSQR) for sparse reconstruction in compressed sensing (CS).
• Applied Minimum Entropy Deconvolution (MED) to identify fault characteristics and extract significant operational information.
• Employed Skip-over, sample entropy, and approximate entropy as health indicators, with Skip-over showing superior performance.
• Implemented a Fractal Autoregressive Integrated Moving Average (FARIMA) model with the R/S method for predicting the Skip-over indicator.

Main Results:

• The hybrid method demonstrated effective sparse optimization and fault characteristic identification.
• Skip-over was identified as a superior health indicator for tracking machinery degradation.
• The FARIMA model accurately predicted the Skip-over indicator, enabling precise RUL estimation.

Conclusions:

• The novel hybrid method achieves highly accurate RUL prediction, significantly safeguarding machinery operation.
• This approach offers a promising solution for long-term health monitoring and predictive maintenance in industrial settings.