One modeling loop
Validate time series, estimate dynamics, test whether the inputs reveal enough, expose fit limits, and suggest a next observation.
ROOT PHYSICS / MODEL STUDIO
A model from
six measurements.
Identify how a system responds from a handful of measurements. See what the model explains, where evidence runs out, and which observation would help next.
Open the workbenchOBSERVE → IDENTIFY → ASKSCROLL TO INTERACT ↓
THE WORKFLOW
One workflow across different physical systems: supply inputs and observations, identify a response, and decide what evidence is still missing.
NO BACKEND / RUNS IN YOUR BROWSERBegin with
the evidence.
Choose a physical system. The same identification engine reads its commanded input and measured response. Domain meaning and approved test limits stay with the engineering team.
SHARED CORE
Time series in. Dynamic response out. Model fit, evidence checks, and next-observation logic stay the same.
A / OBSERVATIONS
Your measurements
Edit a value, add a reading, or import a CSV. Each row records time, commanded input, and measured response.
| NO. | TIME SEC | RUDDER DEG | YAW RATE DEG/SEC | Remove |
|---|
B / MODEL
WAITING FOR DATAWhat the data supports
Observed vs. fitted responseIN-SAMPLE FIT
Observed Fitted
RESPONSE TIME / T—seconds
RESPONSE GAIN / K—°/s per ° rudder
TRAINING ERROR—°/s RMS
The model will appear here after the measurements are analyzed.
Scope. This first-order response fit uses only the measurements shown. It needs independent validation before any operational decision. Physical tests require an approved procedure and a qualified engineer.
Repeat the method.
Not the project.
The vessel and thermal examples use one response model and one evidence workflow. A new physical family enters through a defined adapter; the identification and validation machinery stays reusable.
Physics stays explicit
Define inputs, outputs, units, constraints, and safe test procedures for a family of systems. Experts review these rules once per family.
Prove reuse
Hold out a new asset. Compare its required measurements and engineer hours against the current workflow. Reuse is earned, not assumed.
This workspace demonstrates reuse across two first-order response systems. It does not claim to model every physical system automatically.
From six readings
to the next question.
Watch the model update, inspect its limits, and see how the same core is applied to a second physical system.
Download product guideLess data.
More understanding.
The technical goal is to find a trustworthy model for a new physical system with less expert setup, fewer measurements, and a clear signal when that model stops working.
Bring us a physical problem