Action
The method switch, harmonic treatment, or model-order adjustment applied.
action ⋅ notch_harmonic
Deep Estimate
Deep Estimate:
automated mode identification.
Run an iterative OMA workflow and review its decision log. Deep Estimate is a paid add-on for Pro and is included in Ultimate; it is unavailable in Core.
The loop
An iterative
An iterative
identification workflow.
Deep Estimate evaluates signal diagnostics, selects an identification strategy, and compares the resulting candidates. It can adjust eligible methods, model orders, and frequency bands, then repeat the analysis within a bounded run.
STEP 1Assess signalsReview signal stability, acquisition faults, and spectral content. Descriptive contrast stays separate from SNR, which requires a supported noise model.
STEP 2DiagnoseDetect rotating-machinery harmonics, close modes, weak excitation, aliasing exposure.
STEP 3PlanChoose the remediation: switch SSI variant, adjust model order, re-window, notch.
STEP 4Re-identifyExecute the plan and re-run identification with the adapted strategy.
STEP 5Cross-validateCompare candidates across methods and apply acceptance checks.
The current workflow uses FDD and SSI methods.
What "closed-loop" means here
Review the effect
Review the effect
of each adjustment.
After an in-loop adjustment, the regression guard compares the new evidence with the previous result. Strictly worse outcomes are reverted and the reason is recorded.
THE REGRESSION GUARD
Guarded adjustments
Method and model-order changes, harmonic treatment, and channel adjustments run in an isolated working session. The guard assesses their effect on the evidence before the final mode set is committed.
- GuardedCompare evidence before and after an adjustment.
- IsolatedRun adjustments in an isolated working session.
- RecordedKeep reverted actions and reasons in the log.
CROSS-METHOD ARBITRATION
Cross-method comparison
The primary acceptance route compares frequency-domain and subspace results using cross-method MAC (Modal Assurance Criterion). Guarded single-family routes are available when the data supports only one family, and are labelled in the evidence trail.
- Per bandApply supported adjustments to selected frequency bands.
- Data branchesCompare supported harmonic and channel-processing alternatives.
- ArbitratedApply the acceptance checks to the resulting candidates.
The decision log
A reviewable decision log.
Inspect the action, its diagnostic basis, the resulting evidence, and any rollback. The examples below illustrate the log structure; they are not results from your data.
Diagnostic
The signal or candidate diagnostic that prompted the action.
finding ⋅ spectral kurtosisOutcome
Whether the action was retained or reverted, with the recorded reason.
reverted ⋅ reason loggedAcceptance checks
The checks performed and the measured values compared with their thresholds.
MAC 0.981 ≥ 0.90Provenance
Evidence-bundle hashes identify the associated computed evidence. They do not validate its scientific accuracy.
bundle ⋅ hashedIteration history
Review the sequence of adjustments and outcomes from the run.
timeline ⋅ legible
Damping, cross-examined
Damping cross-checks.
Compare damping with half-power bandwidth and CFDD curve-fit estimates. The method that supplied the committed value is excluded from its own cross-check. Disagreement is displayed and affects the quality tier.
Half-powerBandwidth of the spectral bell around the committed frequency.
CFDD fitParametric SDOF curve fit in the frequency domain.
Scope, stated plainly
Scope and limitations.
Automation supports engineering review. It does not guarantee that every accepted candidate is a physical mode.
Evidence tools
Current v1.0 committed results come from the FDD and SSI identification families through the commit gate.
Spectral settings
The loop adjusts identification methods, model orders, bands, and data branches. It does not silently retune spectral-estimation parameters or resegment the record.
Confidence routes
The primary confident route requires two independent method classes. Guarded single-family routes exist for data that supports only one family — and the evidence trail labels them as such.
Guard boundary
The guard compares evidence for in-loop adjustments in the isolated working session. It does not guarantee the best possible result.
Explore Deep Estimate
Choose your Modal Master edition.
Compare Core, Pro, and Ultimate to find the features you need. Contact us for edition guidance or technical support.
Technical evaluations · Edition guidance · Licensing enquiries