matlab/matlab-agentic-toolkit / skills-catalog/control-systems/matlab-identify-linear-system
matlab-identify-linear-system
VerifiedIdentify a linear dynamic model from input-output or time-series data using MATLAB System Identification Toolbox. Use when estimating transfer function, state-space, ARX, ARMAX, BJ, OE polynomial or process models from measurement data.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 20.4 KB |
| manifest.yaml | config | 570 B |
| references/data-preparation.md | markdown | 12.2 KB |
| references/estimation.md | markdown | 8.3 KB |
| references/model-structures.md | markdown | 4.5 KB |
| references/order-determination.md | markdown | 3.3 KB |
| references/special-topics.md | markdown | 12.4 KB |
| references/validation-and-diagnostics.md | markdown | 7.1 KB |
references/order-determination.md
View on GitHub# Order Determination
## 5.1 Estimate Input Delay (with caution)
```matlab
nk = delayest(ze);
fprintf('Estimated delay: %d samples (%.3f s)\n', nk, nk*ze.Ts);
```
**Warnings:**
* Delay estimation is fundamentally unreliable for noisy data
* `delayest` does not work well for MIMO datasets
* If physical knowledge of delay exists, prefer it
* Consider non-parametric impulse response estimation using `impulseest` and noting the delay as the number of samples for which the response is statistically zero
* When using `impulseest` for delay assessment, use `RegularizationKernel="none"`. Regularization kernels (TC, SE, etc.) smooth the impulse response and can shift or obscure the onset, giving a biased delay estimate. Regularization is appropriate for system bandwidth/order assessment but not for delay detection.
* Always include a model with zero or single-sample delay in the search to hedge against incorrect delay estimates
* `idpack.auto.arxstruc` often provides a better combined delay + order determination
## 5.2 Determine Model Order
### Method A — idpack.auto tools (try first)
The most effective order-determination tools are in `toolbox/ident/ident/+idpack/+auto`. In particular:
* `idpack.auto.ssmSelectBasis` — for state-space order selection
* `idpack.auto.arxstruc` — for combined ARX order and delay determination
### Method B — ARX structure search (polynomial models)
```matlab
maxNA = min(floor(size(ze.y,1)/20), 15);
maxNB = maxNA;
NN = struc(1:maxNA, 1:maxNB, max(nk-1,1):nk+2);
V = arxstruc(ze, zv, NN);
nn = selstruc(V, 'aic'); % [na nb nk]
fprintf('ARX orders: na=%d, nb=%d, nk=%d\n', nn(1), nn(2), nn(3));
```
### Method C — Subspace order (state-space)
```matlab
maxOrder = min(floor(size(ze.y,1)/20), 20);
opt = n4sidOptions('Display', 'off', InteractiveOrderSelection=false);
sysInit = n4sid(ze, 1:maxOrder, opt);
n = order(sysInit);
fprintf('State-space order: n=%d\n', n);
```
### Method D — Iterative complexity (transfer functions)
```matlab
fits = zeros(8,1);
for np = 1:8
nz = max(0, np-1);
sys = tfest(ze, np, nz, nk*ze.Ts);
[~, f] = compare(zv, sys);
fits(np) = f;
fprintf('np=%d, nz=%d: fit=%.1f%%\n', np, nz, f);
end
[~, best_np] = max(fits);
```
### Method E — Process model ladder
```matlab
structs = {"P1D","P2D","P2UD","P2ZD"};
bestFit = -Inf;
for k = 1:numel(structs)
sys = procest(ze, structs{k});
[~, f] = compare(zv, sys);
fprintf('%s: fit=%.1f%%\n', structs{k}, f);
if f > bestFit, bestFit = f; bestStruct = structs{k}; end
end
```
### Method F — Frequency-domain path
```matlab
% Convert to frequency domain, then use AAA initialization
opt = ssestOptions('Display', 'off', 'InitializeMethod', 'AAA', InteractiveOrderSelection=false);
sysInit = ssest(fft(ze), 1:maxOrder, opt);
n = order(sysInit);
```
A closely related variant: use `etfe` or `spa` to compute empirical frequency response (idfrd), then use `ssest` with `InitializeMethod='lsrf'` or `'AAA'` on the idfrd for order determination.
## Order Rules of Thumb
* Max useful order: `n_max ~ min(N/20, 30)` where N = data length
* MIMO state-space: start with `n = max(ny, nu) * 2` up to `5`
* If ARX(10) and ssest(4) give similar fits, prefer ssest(4)
* Stop increasing order when improvement < 2% per additional parameter
----
Copyright 2026 The MathWorks, Inc.
----
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |