FDD
Frequency Domain Decomposition — SVD of the cross-spectral matrix for clean, fast identification from ambient response. The reference method the family is built on.
Brincker · Zhang · Andersen
The engine library
Nine identification engines.
One canonical mode set.
The complete frequency-domain family and the full stochastic subspace family, cross-checked against one another and arbitrated through a literature-cited commit gate into one canonical set of identified modes.
0FD
Frequency-domain engines, FDD through the AFDD wizard.
0TD
Time-domain engines: three SSI variants and NExT-ERA.
0SPECTRUM
Interrogable spectral tools on the bench beneath the engines.
0GATE
One commit gate; one authoritative set of identified modes.
01 — Frequency domain
The FDD family, complete.
From the classic singular-value decomposition of the cross-spectral matrix to spatial filtering for close modes, parametric curve fitting for damping, and a peak wizard that scores candidates by modal coherence — every branch of the frequency-domain literature, in one instrument.
EFDD
Enhanced FDD — SDOF bell isolation around each peak and correlation-function decay for refined frequency and damping estimates.
Brincker · VenturaFSDD
Frequency-Spatial Domain Decomposition — spatial filtering that sharpens close, weakly separated modes where plain FDD blurs them together.
Zhang et al.CFDD
Curve-fit FDD — parametric SDOF fitting in the frequency domain, contributing an independent damping estimate to the cross-check battery.
Jacobsen · AndersenAFDD Peak Wizard
Automated peak selection scored by modal coherence — guided identification that proposes, explains, and stays reviewable at every step.
Modal-coherence criterion
02 — Time domain
The subspace family,
The subspace family,
with uncertainty attached.
Covariance-driven and data-driven stochastic subspace identification with stabilization analysis and physical-pole discrimination — and a UPCX variant that puts analytic standard deviations on every parameter. NExT-ERA stands apart as an independent time-domain cross-check.
SSI-COV
Covariance-driven stochastic subspace identification — multi-order stabilization analysis with physical-pole discrimination, the workhorse of ambient identification.
Van Overschee · De MoorSSI-DATA
Data-driven SSI via orthogonal projection of the raw data matrix — robust identification when covariance estimates run short on record length.
Peeters · De RoeckSSI-UPCX
Subspace identification with analytic covariance — first-order sensitivity propagation from the identified state-space model puts a standard deviation on every frequency, damping ratio, and mode shape.
Method family — Döhler · MevelNExT-ERA
Natural Excitation Technique with Eigensystem Realization — a genuinely independent time-domain estimate, used to cross-examine every SSI result rather than echo it.
James · Juang · PappaEvery engine reads from the same measured bench — Welch spectra, coherence, zoom transforms, and a universal-length Hybrid FFT. See the analysis bench →
03 — Research instrument
And one research engine,
And one research engine,
badged as exactly that.
TOMA (PSDTM-SVD)
Research-badged and evidence-only: identification from the singular values of the
power-spectral-density transmissibility matrix — a mathematical route independent of
both FDD and SSI. Its candidate table corroborates review; it never commits modes.
Real-data gate
Badged after clearing a three-structure replication gate on recorded ambient data
(July 2026): archived protocol, hashed inputs, byte-identical candidate tables across
independent runs.
Nine engines produce candidates. One gate decides. Candidates converge through a literature-cited commit gate into a single canonical mode set — and when a mode is removed, every downstream diagnostic recalculates. No stale results, no orphaned conclusions.
— One source of truth, enforced by architecture
04 — Validation battery
Every candidate interrogated
Every candidate interrogated
from every angle.
Between the engines and the commit gate stands the battery: quality, correlation, complexity, and contamination diagnostics that every candidate must face before it can become a committed mode.
MACModal Assurance Criterion matrices across the full identified set.
CrossMACCross-method shape correlation — FDD against SSI against NExT-ERA.
MPC · MPDModal phase collinearity and mean phase deviation for complexity screening.
StabilizationMulti-order stabilization diagrams with physical-pole classification.
Harmonic screeningSpectral-kurtosis detection protects damping from rotating-machinery contamination.
Aliasing riskPer-mode aliasing exposure assessed against the acquisition chain.
Mode qualityComposite quality indices so review effort lands where doubt is highest.
σ bandsAnalytic uncertainty from SSI-UPCX carried through MAC and downstream checks.
05 — Damping cross-validation
Four independent judges.
Four independent judges.
No self-confirmation.
Every committed damping ratio is cross-examined by up to four automatic estimators from two bias families: half-power bandwidth, CFDD curve fit, Hilbert envelope decay, and continuous wavelet transform. The estimator that produced the committed value is deliberately excluded from its own cross-check — so when the judges agree, the agreement is real. Disagreement is displayed, not averaged away. How the consensus works →
Private demonstration
Run the full library
on your record.
Bring your own ambient data and watch the library cross-examine itself — candidates, diagnostics, arbitration, and the committed set, with the evidence trail open.
Enterprise evaluations · Technical dossiers under NDA · Volume licensing