struktura 0.4.0

Predict failure before it happens. Universal anomaly detection via DFA scaling analysis — bearings, heartbeats, spacecraft, drones, DNA. Zero dependencies.
Documentation

Struktura

Crates.io CI License docs.rs

Predict failure before it happens.

The only Rust-native DFA (Detrended Fluctuation Analysis) implementation. No hyperparameters, no training, MIT/Apache licensed. Currently proposed into NASA fprime, ArduPilot, PX4, and cFS.

cargo install struktura
struktura demo

  STRUKTURA DEMO
  Bearing fault detection from CWRU vibration data
  ================================================

  ##############..............  Normal bearing       alpha=0.738  [EXACT]

  ####..........................  Inner race FAULT   alpha=0.217  [STRONG]
                                  shift=-0.522  CRITICAL

  The bearing's vibration structure changed BEFORE
  any amplitude threshold would have fired.

Try it now

cargo install struktura
struktura demo                              # builtin bearing fault demo
struktura check your_data.csv               # analyze any signal
struktura check data.csv --baseline 0.39    # compare against baseline
struktura compare normal.csv faulted.csv    # side-by-side comparison

As a library

use struktura::{analyze, health_check, HealthVerdict};

let law = analyze(&vibration_data);
let verdict = health_check(&law, 0.389);
// verdict == HealthVerdict::Critical

Real-time monitoring:

use struktura::{BaselineTracker, HealthVerdict};

let mut tracker = BaselineTracker::new(256, 1000);

for sample in sensor_stream {
    if let Some(verdict) = tracker.push(sample) {
        if verdict == HealthVerdict::Critical {
            trigger_alert();
        }
    }
}

Bearing fault detection (CWRU data, builtin)

Condition DFA alpha Shift Verdict
Normal 0.738 -- Healthy
Inner race fault 0.217 -0.522 CRITICAL

Reproduced by struktura demo using real CWRU Bearing Data Center samples bundled in the crate.

Cross-domain proof

Same algorithm. Zero configuration. Every number from an actual run.

Domain Signal DFA alpha R2
Bearings CWRU 12kHz vibration 0.389 0.872
Spacecraft Queue depth telemetry 0.593 0.789
Genome Human chr1 GC% (8K windows) 0.909 0.991
Cardiac RR intervals (HRV) 0.695 0.985

Genome: 8/8 chromosomes at R2 > 0.99.

How it works

DFA (Peng et al., Physical Review E, 1994 -- 3000+ citations) measures long-range correlation:

  1. Compute the cumulative profile
  2. Divide into boxes, detrend each
  3. Measure residual fluctuation vs box size
  4. Slope in log-log space = alpha
  • alpha ~ 0.5: random noise
  • 0.5 < alpha < 1.0: healthy structure
  • alpha shifts from baseline: degradation

The crate reports R2 alongside every alpha. If R2 < 0.3, quality = Abstain. It never bluffs.

Alternatives

Crate DFA License Dependencies Flight-software proposals
struktura native MIT/Apache 0 4 (fprime, ArduPilot, PX4, cFS)
anomaly_detection no GPL-3.0 many 0
extended-isolation-forest no MIT many 0

Proposed into

References

  1. C.-K. Peng et al., "Mosaic organization of DNA nucleotide sequences," Physical Review E 49(2), 1994.
  2. C.-K. Peng et al., "Quantification of scaling exponents," Chaos 5(1), 1995.
  3. CWRU Bearing Data Center: https://engineering.case.edu/bearingdatacenter

Links

License

MIT OR Apache-2.0