Autonomous nonlinear dynamics engine
NLSYS filters ordered signals, compiles verified nonlinear dynamics and produces bounded control recommendations—without a hand-designed filter or a supplied governing model.
Workflow illustration — not measured performance data.
One engine, three jobs
Each product is independently usable. The same ordered data can then move from cleaning to modelling and from modelling to control without changing the underlying evidence discipline.
Self-tuning filtration in instant, bounded-latency and offline modes, with gap recovery, prediction, health and System Passports.
NDC searches alternative dynamic structures, challenges them in held-out free-run and packages the selected model as an executable artifact.
Use full-model SDRE under a real-time cadence and NMPC where a slower optimisation loop is acceptable. Execution remains with the operator or supervisory system.
Published evidence
The tables below give recorded outcomes and their stated protocol boundaries. Review the full rows, applicable cases and limitations before comparing results across methods.
Mean offline RMSE-reduction gain across the current 32-case real-waveform acceptance suite.
Open all 32 cases → NDC6 / 6Published free-run cases below the declared stable classical baseline; recorded ratios range from 0.10× to 0.91×.
Open model results → CONTROL0.434Mean normalized score for full-model SDRE versus 0.762 for the frozen linear LQR comparator under the current protocol.
Open control protocol →Self-service path
Choose the relevant product, inspect its input requirements and try the available example. Use public evidence when it helps your evaluation; a benchmark visit is not required before using the workspace.
Read the full protocol, baseline and limitations.
Use the example already available in the current account interface.
Upload a file or connect an ordered stream.
Download the cleaned output, Passport or executable model.
Move to a subscription, shared workflow, API or Controller only when the job repeats.
Entry points
Screen a cycler CSV or TSV locally in the browser. The file is not uploaded.
Run local checkStart with the built-in trajectory example, then compile one verified model for one target.
Open NDCTry the sample file, then reuse the workflow for live or recurring sensor channels.
Open FiltrationEstimate a completed engineering CSV or TSV before creating an account or paying.
Estimate a fileArtifacts, not screenshots
NLSYS outputs are designed for hand-off: cleaned files, event evidence, System Passports, executable nonlinear models, validation reports, support boundaries, manifests and checksums.
Start with evidence
Choose a workflow directly, or inspect its public results first. The account shows current access and the order details before a paid launch.