Experimental Evaluation of Current Sensors’ Fault Detection and Classification Methods in PMSM Drives

Kamila Anna Jankowska

doi:10.2478/pead-2026-0005

.blurhash-client-img { display: none !important; }

Experimental Evaluation of Current Sensors’ Fault Detection and Classification Methods in PMSM Drives

Power Electronics and Drives

Volume 11 (2026): Issue 1 (January 2026)

By: Kamila Anna Jankowska

Open Access

|Mar 2026

Figures & Tables

Block diagram (a) and photos (b) of experimental set-up. PMSM, permanent magnet synchronous motor.

Control system diagram with measurement systems. PMSM, permanent magnet synchronous motor.

Sample transients with different types of faults – signal noise (a), gain error (b), and signal loss (c).

Block diagram of the phase detection and localisation system based on Cri markers.

Speed, current, and detector response waveforms in the standard version during periodic signal interruption in phases A (a) and B (b).

Speed, current, and detector response waveforms in the modified version during periodic signal interruption in phases A (a) and B (b).

Waveforms of speed, current, markers, marker differences, and detector response during signal loss in phases A (a) and B (b).

Waveforms of speed, current, markers, marker differences, and detector response during signal noise in phases A (a) and B (b).

Confusion matrices for the detector based on Cri markers for phases A (a) and B (b).

Confusion matrices for the classifier based on MLP for phases A (a) and B (b) for the test data. MLP, multilayer perceptron.

Speed, current, and classifier outputs transients during signal loss and signal noise for no-load conditions in phase A (a) and loaded motor conditions in phase B (b).

Speed, current, and classifier outputs transients during faults in phase B.

Confusion matrices for the classifier based on CNN for phases A (a) and B (b) for the test data.

Speed, current, and classifier outputs transients during signal loss and load condition in phase A (a) and non-load condition in phase B (b).

Speed, current, and classifier outputs transients during signal noise and gain error in regenerative mode conditions in phases A (a) and B (b).

Parameters of training and testing data for the classifier based on MLP_

Feature	Training data	Testing data
Number of samples	380,010	228,006
Speed	±0.1ω_ref, ±0.2ω_ref, ±0.3ω_ref	±0.075ω_ref, ±0.15ω_ref, ±0.225ω_ref
Motor load	0.1 T_N, 0.3 T_N	0.2 T_N
Regenerative mode	0.1 T_N, 0.3 T_N	0.2 T_N

Essential parameters of the tested motor_

P_N (kW)	P_p (−)	n_N (rpm)	T_N (Nm)	I_N (A)	J (kg . m²)	R_S (Ω)	L_S (mH)
0.894	4	6,200	1.4	1.9	0.000039	4.6615	7.9835

Structure of the CNN classifier network_

Input layer: Matrix 40 × 10

Feature detector
Convolutional layer 3 × 90	Batch normalisation layer	Activation function: ReLu	MaxPooling layer
Padding method: same	Batch normalisation layer	Activation function: ReLu	Stride: 20
Convolutional layer 3 × 120	Batch normalisation layer	Activation function: ReLu	MaxPooling layer
Padding method: same	Batch normalisation layer	Activation function: ReLu	Stride: 2
Convolutional layer 3 × 150	Batch normalisation layer	Activation function: ReLu	MaxPooling layer
Padding method: same	Batch normalisation layer	Activation function: ReLu	Stride: 2
Convolutional layer 3 × 180	Batch normalisation layer	Activation function: ReLu	MaxPooling layer
Padding method: same	Batch normalisation layer	Activation function: ReLu	Stride: 2
Classification
Fully connected layer (4)	Softmax layer	Classification layer
Output layer: 1 – no fault, 2 – signal loss, 3 – signal noise, 4 – gain error

Types of individual failures and equations that enable their simulation_

Type of the fault	Current value
Gain error	iAfault=1−γiAmeas i_A^{fault} = \left( {1 - \gamma } \right)i_A^{meas}
Signal noise	iAfault=iAmeas+nt i_A^{fault} = i_A^{meas} + n\left( t \right)
Signal loss	iAfault=0 i_A^{fault} = 0

Parameters of training and testing data for the classifier based on CNN_

Feature	Training data	Testing data
Number of samples	69,800	69,800
Number of training examples	698	698
Speed	±0.05ω_ref, ±0.1ω_ref, ±0.2ω_ref ±0.3ω_ref	±0.07ω_ref, ±0.15ω_ref, ±0.25ω_ref ±0.35ω_ref
Motor load	0.1 T_N, 0.2 T_N	0.15 T_N, 0.25 T_N
Regenerative mode	0.1 T_N	0.15 T_N

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/pead-2026-0005 | Journal eISSN: 2543-4292 | Journal ISSN: 2451-0262

Journal RSS Feed

Language: English

Page range: 79 - 94

Submitted on: Dec 3, 2025

Accepted on: Feb 19, 2026

Published on: Mar 16, 2026

Published by: Wroclaw University of Science and Technology

In partnership with: Paradigm Publishing Services

Publication frequency: 1 issue per year

Keywords:

current-sensor faults,

neural classifier,

fault-tolerant control,

CNN,

MLP

Related subjects:

Computer sciences,

Artificial intelligence,

Engineering,

Electrical engineering,

Electronics

© 2026 Kamila Anna Jankowska, published by Wroclaw University of Science and Technology
This work is licensed under the Creative Commons Attribution 4.0 License.

Volume 11 (2026): Issue 1 (January 2026)

Experimental Evaluation of Current Sensors’ Fault Detection and Classification Methods in PMSM Drives

Figures & Tables

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Parameters of training and testing data for the classifier based on MLP_

Essential parameters of the tested motor_

Structure of the CNN classifier network_

Types of individual failures and equations that enable their simulation_

Parameters of training and testing data for the classifier based on CNN_

Paradigm

My account