One binary. Infinite telemetry.

Validated across five domains · Same source · Same flags

The specification

Footprint — 8,480 bytes: One compact detector small enough to live on the microcontroller native to the equipment.
Enrollment — 1,700 samples †: Learns healthy operation in 1,700 samples—as little as two seconds at kilohertz rates—then locks. No labels. No tuning.
Heap allocations — 0: No dynamic allocation after initialization. Architected for constrained and safety-critical firmware environments.
Delivery — binary: Compiled for your target hardware. Deterministic and verifiable against published test vectors, with source available to certifying authorities under NDA.
False positives — 0 / 120 h: Zero false positives across 120 hours of healthy synthetic-vibration soak testing, with a 95% confidence ceiling below 0.025 per hour.
Behavior — bit-identical: Same signal. Same answer. Replayable and auditable.

† Enrollment requirements vary by equipment type, modes, and operational envelope.

 ADAPTIVE BASELINE  --  the band moves with the fault
  peak amp

      |                                    - - -      <-- band rose with it
      |
      |                              - - -  ####
      |                                     ####
      |                        - - -  ####  ####
      |- - - - - - - - - - - -        ####  ####
      |                         ####  ####  ####
      |             ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      +------------------------------------------
         w1    w2    w3    w4    w5    w6    w7

  REPORTS: NOMINAL  --  the fault became the new normal

 LOCKED BASELINE (FLYNN)  --  the band is fixed at enrollment
  peak amp

      |
      |
      |                                     ####      <-- BREACH
      |                                     ####
      |                               ####  ####
      |- - - - - - - - - - - - - - - -#### -#### -    <-- locked band
      |                         ####  ####  ####
      |             ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      | ####  ####  ####  ####  ####  ####  ####
      +------------------------------------------
         w1    w2    w3    w4    w5    w6    w7

  REPORTS: ANOMALY at w6  --  the same swell breaks the band

Why a locked baseline changes the result. The run-to-failure telemetry in Domain 02 below, fed to two architectures. A self-tuning threshold widens as the fault grows and never reports it. Flynn’s threshold is fixed at enrollment, so the same signal crosses it.

Domain 01 · Bearing vibration: the industry-standard bearing benchmark

The CWRU bearing dataset spans 27 fault pairs across inner-race, ball, and outer-race faults at three severity levels. The recommended operating point produces the deployed numbers below, with no oracle threshold and no per-fault tuning.

Protocol

Dataset: Case Western Reserve University bearing data
Fault pairs: 27 (inner race · ball · outer race × 3 severities)
Threshold: Self-calibrated · locked at deployment
Configuration: Default operating point

Result

Precision	1.0	of alarms raised, share correct
Recall	1.0	share of real faults caught
F1 score	1.0	harmonic mean

Domain 02 · Run-to-failure: caught both failing bearings, seventeen days early

An industry-standard run-to-failure dataset spans more than 30 days of continuous vibration from four bearings, two of which failed and two of which survived. Flynn ran on all eight accelerometer channels simultaneously — zero configuration changes between them.

Result

Channels	8	simultaneous · zero config
Failing bearings detected	2 / 2	both before documented failure
Lead time	~17 d	ahead of failure date
Adjacent-channel pickup	Ranked ‡	same fault, lower magnitude
Duration	30+ days	continuous vibration
False positives	0	no spurious faults reported

Protocol

Dataset: NASA IMS Bearing dataset †
Setup: 4 bearings · 8 accelerometer channels
Outcome: 2 failed · 2 survived
Sampling: Continuous, multi-day
Configuration: Same source, same flags across channels

† Lee, J., Qiu, H., Yu, G., Lin, J. and Rexnord Technical Services (2007). IMS, University of Cincinnati. “Bearing Data Set”, NASA Prognostics Data Repository, NASA Ames Research Center, Moffett Field, CA.

‡ On the healthy-bearing channels Flynn surfaced a low-magnitude signal correlating with the failing bearings. This was initially scored as a false positive. That scoring was incorrect: the signal is the same developing fault, conducted mechanically through the test rig and reported at proportionally lower magnitude than on the failing-bearing channels. Flynn detected the fault on adjacent channels and ranked its severity correctly.

Domain 03 · Ambient & diurnal: operationally compatible with shift-cycle review

336 hours of ambient time-series covering temperature, environmental, and process-control signals — slow-changing data with strong day-night cycles. Flynn maintains a false-positive rate that fits the cadence of normal operator review.

Protocol

Duration: 336 hours
Signal types: Temperature · environmental · process-control
Review cadence: Shift-cycle compatible

Result

FP / hour	0.074 – 0.080	across 336 hours
Alerts / day	~1.8 – 1.9	reviewable in normal rhythm
Tuning	None	same source as bearing

Domain 04 · Electrical grid stability: electrical-bus stability with no domain assumptions

10,000 instances across 12 features of electrical-grid stability data — a signal domain that shares nothing with rotating-machinery vibration. Same source, same flags, no per-domain configuration.

F1 score	0.532	no per-domain config
Instances	10,000
Features	12

Domain 05 · Soak testing: the number operators ask for first

The false-alarm rate is the first thing operators ask about. Across 120 simulated hours of synthetic vibration — 5 random seeds × 24 hours each — Flynn produced zero false positives, with a 95%-confidence ceiling of 0.025 per hour. On bearing normal-operation traces, zero across 1.2 million samples.

Protocol

Synthetic soak: 120 hours · 5 seeds × 24 hours
Bearing soak: 1.2 million normal-operation samples
Confidence: 95% CI bound on FP rate

Result

FP · synthetic	0	120 hours
FP · bearing	0	1.2 M samples
95% CI bound	< 0.025/h	false-positive rate

Zero false positives on bearing data. Operationally low on diurnal. The numbers operators actually care about.

Reproduce these numbers: benchmarks, source, reproduction

Flynn’s empirical claims are reproducible from committed benchmarks and committed source code. Artifacts and reproduction instructions are available to evaluation licensees.

Email: tripp@inlikeflynn.io
Available: Source code · benchmarks · scripts · datasets manifest
Under: Evaluation license · NDA required

Read the whitepaper