Fusion plasma,
controlled by phase arrays.
An interactive simulator extending BMPA (Phase Array + AI Closed-Loop Optimization) to tokamak MHD instability-mode control. Compare mode amplitude, phase correction, and heat-load proxies under the same initial conditions and disturbances for baseline and BMPA control.
MHD Mode Amplitude · Time Response
12-Channel Phase Array
Combines circumferential coil phases to synthesize a correction vector for the selected MHD mode.
Phase Array × AI Closed Loop
Estimate complex MHD mode amplitude and phase from simulated magnetic diagnostics, such as Mirnov coils.
Monitor estimated phase, mode amplitude, disturbance response, and prediction residuals.
Evaluate candidate phases and gains to minimize predicted loss and update the correction angle.
Synthesize 12 phase-control channels and feed a bounded correction field back to the plasma model.
Simulation Event Log
From interactive demo to testable research proposal.
This research-oriented extension documents the reduced model, implemented algorithms, limits of the comparison, and requirements for further validation. It is an independent proposal relevant to KSTAR's MHD research interests; it is not an official study, endorsement, or validation by KSTAR or the Korea Institute of Fusion Energy (KFE).
An explicit model and control law.
What the browser actually computes, with the model boundary stated clearly.
Complex mode state
a=x+iy; |a| is normalized mode amplitude. γ=0.08+mode term+0.40D−0.28B. ω is selected by mode. The 2/1, 3/2 and 1/1 choices are reduced-model labels.
Prediction and online search
Predict a mode state and apply an approximately opposing phase. Five candidate offsets are evaluated online. This is a lightweight adaptive optimizer, not a trained neural network.
Idealized 12-channel array
Channel positions θⱼ=2πj/12 and phases φⱼ=φₜ−nθⱼ. Real coil transfer functions and electromagnetic coupling to a tokamak plasma are not modeled.
A simplified fixed-gain, fixed-offset feedback example. It does not represent a real KSTAR controller or a validated industrial PID/MPC implementation.
Predictive phase correction with an idealized coil array. The baseline and BMPA have different observer filters and output limits, so this is not a fair performance benchmark yet.
A reproducible validation pathway.
Proposed gates before any claim of real tokamak-control performance.
Use matched modes, initial states, disturbances, observation delays, sensor noise, actuator limits and control-energy budgets across PID, MPC and BMPA.
Freeze seeds, pulse amplitudes/timing, test windows and allowed tuning ranges before examining performance outcomes.
Compare peak amplitude, modal growth, recovery time after the pulse, phase estimation error, integrated control effort and failure rate.
Retest on held-out seeds and parameter sets. Move to offline data, then validated physics solvers, before considering any experimental integration.
| Gate | Comparator / input | Required artifacts | Status |
|---|---|---|---|
| V0 · Web demonstrator | Fixed feedback vs BMPA in a reduced oscillator | Settings JSON, amplitude CSV, event log | Implemented |
| V1 · Fair benchmark | PID / MPC / BMPA under equal sensor and actuator limits | Multi-seed statistics, control effort, failure rates | Planned |
| V2 · Physics integration | Validated MHD solver and coil-response model | Mode stability, delay/saturation sensitivity | Unverified |
| V3 · Research diagnostics | Offline analysis of authorized KSTAR/public diagnostics | Phase-estimation errors, reproducibility report | No data link |
Export settings alongside the signals.
The experiment runs locally in your browser. Download configuration metadata and a proposed validation protocol.
Implemented, hypothesized, or unverified?
A transparent evidence ledger for technical reviewers.
| Item | Evidence available | Status |
|---|---|---|
| Browser simulator | Implemented visualization, disturbance injection, CSV export and numerical oscillator | Implemented |
| Phase-array mode suppression | Hypothesis testable in this reduced model; no demonstrated tokamak performance | Hypothesis |
| Superiority over PID/MPC | Not compared under equal sensor/actuator constraints | Unverified |
| KSTAR diagnostic / device integration | No authorized data integration, institutional partnership or device experiment | Not connected |
| Confinement, heat flux, neutron damage | No physics-based calculations or experimental validation | Unverified |
A staged proposal for KSTAR-relevant research.
Begin with a small, reviewable offline study rather than a request for device access.
Discuss physical meaning, observable mode phases and available coil authority
Compare PID, MPC and BMPA under jointly agreed constraints
Evaluate predictions using authorized diagnostics, MHD and coil responses
Define any further study through the institution's own safety and review processes
Partner with us to
advance BMPA fusion control.
We welcome collaboration with research institutions, companies, engineering teams, and strategic partners — from validation of fusion-plasma phase-array control to co-development and commercialization.
Please do not include trade secrets or confidential technical materials in your initial message. Your submission is processed by Formspree to enable us to respond to your inquiry.