Industrial controls / AI supervisory tuning

One AI supervisor. Multiple control algorithms.

Map meaningful signals and authorized tuning parameters once. Then observe performance, review an AI target, apply a bounded request, and evaluate the response.

CUSTOM NAV GUIDANCECUSTOM PI / PIDAI SUPERVISED APPLY

Offline interface demo: sample response curves, AI targets and PLC readbacks are scripted. No real AI computation, PLC connection, or external data transfer takes place. Descriptions and tags below are editable example configurations.

AI Supervised Tuning

Focus on the result: observed response → recommendation → one authorized request → PLC readback → evaluation.

Operator-approved demonstration

ACTIVE LOOP / OFFLINE EXAMPLE

Supervisor operation

AI SUPERVISED • DEMO

1 Observe2 Analyze4 Apply5 Verify

NAV Track Capture

Example recording

SIMULATED / READY

Response trendILLUSTRATIVE DATA / 20 SECONDS

BaselineCurrent simulated responseReference

Ready to analyze sample

A stored illustrative response is ready for review.

AI recommendation

NOT ANALYZED

Parameter	PLC applied	AI target	Next request

Swipe sideways to view the full recommendation table.

Select Analyze sample to see the illustrated target and bounded next request. The PLC applied/readback value is a separate part of the workflow.

Communications / data freshness valid (simulated) PLC tuning permission enabled (simulated)

PLC owns qualification, limits, slew rates and fallback. No device is connected.

One-time control loop setup

Define the controller, editable descriptions, monitored signals, and authorized tuning parameters. Save it as a reusable configuration.

Template mapping

01 / CONTROL LOOP

General information

No PIDE instruction or standard tag naming is required.

Editable loop name

Controller / reference address

Control algorithm

Implementation notes / gain convention (editable)

02 / OBSERVATION SIGNALS

Monitored signals

Descriptions are provided by the example template and fully editable. The customer assigns each tag a function.

In a connected product, MELD can browse tag names and read data types. Meaningful descriptions may come from imported controller documentation or manual entry—not from guessing tag names.

03 / PARAMETER AUTHORITY

Tunable parameters

Each adjustable variable has its own description, request tag, applied/readback tag and permitted range.

For custom PID, Kp / Ki / Kd may be separate PLC REAL tags. For NAV guidance, lookahead and intercept scales are the tunables. The PLC must independently qualify all requests.

04 / CUSTOMER OBJECTIVE

Desired behavior

The objective and write authority become part of the saved loop profile.

Smooth / conservativeFavor damping and minimal overshoot BalancedTrade response speed against overshoot Faster responsePrioritize convergence within limits

Maximum requested adjustment per step

The next request is limited by this percentage and the permitted parameter range. These are demo configuration values, not approved limits for any aircraft or industrial system.

05 / REUSABLE PROFILE

Save loop setup

Save locally and exchange configurations as JSON files.

Saving the configuration stores your edited descriptions, tags and tuning limits. It does not validate them against a real PLC.

Template values loaded. Save to keep your edits.

Import JSON

Browser storage is used where available. Use Export JSON to move a configuration to another computer. Never use these sample settings on real equipment without engineering validation.

WORKFLOW

Configure once. Tune repeatedly.

Step 1Name the loop and map its signals

Step 2Choose tunables and their limits

Step 3Save / review configuration

Step 4Use AI Supervised Tuning