Skip to content

DisturbanceRecoveryEvents¤

Detect external upsets hitting a process signal and measure recovery.

Module: ts_shape.events.engineering.disturbance_recovery
Guide: Process Engineering

When to Use¤

Use to detect and characterize external process upsets (material changes, ambient disturbances, upstream failures) and measure how quickly the process recovers. Critical for root cause analysis and for quantifying process robustness against unplanned events.

Quick Example¤

from ts_shape.events.engineering.disturbance_recovery import DisturbanceRecoveryEvents

analyzer = DisturbanceRecoveryEvents(
    df=process_data,
    uuid="reactor_temperature_01"
)

# Detect disturbances exceeding 3 sigma from baseline
disturbances = analyzer.detect_disturbances(
    baseline_window="10m",
    threshold_sigma=3.0
)

# Measure recovery time for each disturbance
recovery = analyzer.recovery_time(
    baseline_window="10m",
    recovery_pct=0.95
)

# Count disturbances per 8-hour shift
frequency = analyzer.disturbance_frequency(window="8h")

# Compare process behavior before vs after each upset
comparison = analyzer.before_after_comparison(baseline_window="10m")

Key Methods¤

Method Purpose Returns
detect_disturbances(baseline_window='10m', threshold_sigma=3.0) Detect process upsets DataFrame of disturbance events
recovery_time(baseline_window='10m', recovery_pct=0.95) Recovery time per disturbance DataFrame with recovery metrics
disturbance_frequency(window='8h') Count disturbances per window DataFrame with counts
before_after_comparison(baseline_window='10m') Before vs after statistics DataFrame with comparison stats

Tips & Notes¤

Correlate disturbances with upstream events

After detecting disturbances, use before_after_comparison() to quantify the impact, then cross-reference timestamps with upstream signals or operator logs to identify root causes. Pair with disturbance_frequency() to spot recurring patterns.

Related modules

See Also¤