Sliding window constrained fault-tolerant filtering of compressor vibration data

Area of Science:

• Petrochemical instrumentation
• Signal processing
• Fault-tolerant systems

Background:

• Petrochemical instrumentation generates time-series data susceptible to noise and outliers.
• Existing filtering methods may struggle with complex noise patterns and fault tolerance.

Purpose of the Study:

• To develop a robust sliding window constrained fault-tolerant filtering method.
• To enhance the accuracy and reliability of sampled data in petrochemical processes.

Main Methods:

• A sliding window constraint function is applied to time-series data.
• Smoothed estimates are generated and used to derive a difference series.
• The difference series undergoes further smoothing for compensation.

Main Results:

• The proposed method demonstrates superior performance over sliding mean, median, and Savitzky-Golay filters.
• Effective filtering of random errors and reduction of outlier impact.
• Preservation of true data variations in the presence of noise.

Conclusions:

• The sliding window constrained fault-tolerant filtering method offers enhanced accuracy for petrochemical instrumentation data.
• The technique exhibits strong fault tolerance and noise reduction capabilities.
• It is suitable for processing noisy and outlier-contaminated time-series data.