Estimating four electric-motor temperatures
A motor-temperature estimator can complement instrumentation when its operating domain is understood. This example predicts yoke, tooth, winding and magnet temperatures from eight operating measurements.
1. Industrial challenge
Motor design and monitoring require attention to several thermal locations. This model maps one operating point to four temperatures, allowing the error at each location to be assessed separately.
Four locations
Estimate yoke, tooth, winding and permanent-magnet temperatures.
Operating point
Supply eight electrical and mechanical measurements.
Thermal assessment
Examine each temperature error before considering a monitoring application.
2. Data set
The supplied model uses a prepared table of 175 operating records, with 35 in its test report. It is not trained on the full high-frequency public dataset. Session identifiers in the accompanying CSV make a session-level evaluation possible, but the saved split is by record.
The downloadable project, saved report and supplied source CSV define the exact version used here. Repository: original dataset/source record.
| Subset | Records |
|---|---|
| Training | 105 |
| Validation / selection | 35 |
| Testing | 35 |
| Unused | 0 |
| Variable | Role | Type | Encoding | Unit |
|---|---|---|---|---|
| ambient_temperature | Input | Numeric | °C | |
| coolant_temperature | Input | Numeric | °C | |
| speed | Input | Numeric | rpm | |
| torque | Input | Numeric | N·m | |
| voltage_d | Input | Numeric | As supplied | |
| voltage_q | Input | Numeric | As supplied | |
| current_d | Input | Numeric | As supplied | |
| current_q | Input | Numeric | As supplied | |
| yoke_temperature | Target | Numeric | °C | |
| tooth_temperature | Target | Numeric | °C | |
| winding_temperature | Target | Numeric | °C | |
| magnet_temperature | Target | Numeric | °C |
Interactive chart: tooth_temperature distribution. Enable JavaScript to explore it.
Interactive chart: yoke_temperature Pearson correlations chart. Enable JavaScript to explore it.
Interactive chart: tooth_temperature Pearson correlations chart. Enable JavaScript to explore it.
Interactive chart: winding_temperature Pearson correlations chart. Enable JavaScript to explore it.
Interactive chart: magnet_temperature Pearson correlations chart. Enable JavaScript to explore it.
Interactive chart: winding_temperature vs. ambient_temperature scatter chart. Enable JavaScript to explore it.
3. Model
The final model has 8 encoded input features and 4 outputs. The following dimensions describe the final saved network.
| Layer | Input shape | Output shape | Activation |
|---|---|---|---|
| Scaling | 8 | 8 | — |
| Dense | 8 | 3 | Tanh |
| Dense | 3 | 4 | Identity |
| Unscaling | 4 | 4 | — |
| Clamping | 4 | 4 | — |
The final output is a continuous estimate. Scaling, unscaling and any bounds follow the saved implementation; the calculator does not silently impose physical constraints.
4. Training strategy
The saved training configuration uses NormalizedSquaredError with QuasiNewton. Training minimizes the recorded objective; the validation subset monitors generalization during fitting. The testing subset is used for the evaluation below.
Interactive chart: Quasi-Newton method error history. Enable JavaScript to explore it.
Quasi-Newton method results
| Measure | Value |
|---|---|
| Epochs number | 71 |
| Elapsed time | 00:00:00 |
| Stopping criterion | Minimum loss decrease |
| Training error | 0.045 |
| Validation error | 0.076 |
5. Model selection
No model selection experiment is recorded in this supplied project. The displayed architecture is the trained model used for testing; earlier article claims about a different selected architecture do not apply to this version.
6. Testing analysis
The final model is evaluated on 35 testing observations. MAE, RMSE and prediction R² below are computed from the exact testing pairs stored by Neural Designer. Errors retain each target’s original units.
| Target | Unit | MAE | RMSE | Prediction R² | Pearson r² |
|---|---|---|---|---|---|
| yoke_temperature | °C | 2.466 | 3.032 | 0.971 | 0.976 |
| tooth_temperature | °C | 3.162 | 3.857 | 0.969 | 0.972 |
| winding_temperature | °C | 3.959 | 5.084 | 0.969 | 0.97 |
| magnet_temperature | °C | 5.846 | 7.247 | 0.849 | 0.852 |
The native goodness-of-fit task reports squared Pearson correlation (r²). Prediction R² here is 1 − squared prediction error / squared deviation from the test mean. They measure different properties: a strongly correlated prediction can still have a bias or an incorrect scale.
Interactive chart: yoke_temperature goodness-of-fit chart. Enable JavaScript to explore it.
Interactive chart: tooth_temperature goodness-of-fit chart. Enable JavaScript to explore it.
Interactive chart: winding_temperature goodness-of-fit chart. Enable JavaScript to explore it.
Interactive chart: magnet_temperature goodness-of-fit chart. Enable JavaScript to explore it.
7. Model deployment
Validated inputs → saved preprocessing → neural network → score or estimate → domain review. The ZIP contains the original project, source CSV, schema, test metrics and standalone interactive chart exports. The project hash in the schema identifies this exact version.
Directional response at a fixed reference point
Inspect the saved reference point
| Measure | Variable | Value |
|---|---|---|
| 1 | ambient_temperature | 24.426 |
| 2 | coolant_temperature | 18.667 |
| 3 | speed | 4500 |
| 4 | torque | 20.092 |
| 5 | voltage_d | -42.684 |
| 6 | voltage_q | 124.37 |
| 7 | current_d | -91.699 |
| 8 | current_q | 24.509 |
Interactive chart: winding_temperature – torque directional output. Enable JavaScript to explore it.
Explore the exported model
This research demonstration runs locally in your browser. Values outside the training range are outside the validated domain and are rejected. A valid input range does not guarantee that a combination is physically or operationally plausible.
This is not a certified control or protection system.
8. Scope and limitations
Thirty-five test records provide limited evidence for a thermal model. Hold out complete sessions or machines and test transient regimes separately. This static estimator is a component for investigation, not a validated dynamic twin, controller or thermal protection system.
No external validation or independent calibration study is included. Preprocessing statistics and model choices should be refitted within a prospective or grouped validation design. Correlations and directional responses describe associations, not causes. Human review is required before an operational decision.
