USE OF QUANTUM ALGORITHMS FOR CLASSIFICATION OF ROLLING BEARING DAMAGE
Quantum Machine Learning; Bearing Datasets; Prognostic and Health Management; Fault Diagnosis;
We discusses the use of machine learning (ML) and quantum machine learning (QML) algorithms in reliability engineering. The study focuses on comparing neural networks created with QML, using three different sets of bearing data. The datasets include vibration data from accelerometers and must be pre-processed for QML algorithms. The approach aims to detect faults in rotary equipment and considers healthy, inner ring fault, and outer ring fault states of the bearings. The study compares established circuit designs, as well as hybrid models that combine quantum circuits and neural networks. Results are obtained on classical computers using analytical calculations that simulate a quantum computer. The ZFeatureMap circuit is found to be the best way to enter data, with quantum convolutional circuits yielding better results than other parameterized circuits. The accuracy rates achieved in the training dataset are around 96%, indicating the stability of the parameterized quantum circuits used. However, similar accuracy rates must be ensured in the testing dataset to avoid overfitting or underfitting issues.