In this paper, we present a new low-cost system for surface electromyogram (sEMG) acquisition. developed and designed for rehabilitation application purposes. The noninvasive device delivers four-channel EMG bio-signals describing the electrical activity for the right upper limb muscles. The recorded EMG signals obtained from several healthy subjects were exploited to build a database for movement detection and to evaluate the mechanical properties of the upper limb muscles. The proposed study focuses mainly on the influence of the use of the principal component analysis (PCA) method on the movement classification performance based on the sEMG extracted signals. Several tests were conducted, and the simulation results clearly showed the positive impact of PCA as a dimensionality reduction approach with respect to two performance metrics: the classification rate (CR) and the system’s response time. This advantage was confirmed via numerical tests using three different classifiers: K-nearest neighbor (KNN), probabilistic neural network (PNN), and learning vector quantization (LVQ), with and without PCA. The obtained classification rates highlighted the success of the proposed method since a clear improvement in the classification rates was achieved.
Electrocardiogram (ECG) signals have distinct features of the electrical activity of the heart which are unique among individuals and have recently emerged as a potential biometric tool for human identification. The paper attempts to address the problem of ECG identification based on non-fiducial approach using unsupervised classifier and a Deep Learning approaches. This work investigates the ability of local binary pattern to extract the significant pattern/feature that describes the heartbeat activity for each person’s ECG and the use of staked autoencoders and deep belief network to further enhance the extracted features and classify them based on their heartbeat activity. The proposed approach is validated using experimental datasets from two publicly available databases MIT-BIH Normal Sinus Rhythm and ECG-ID and the results demonstrate the effectiveness of this approach for ECG-based human authentication.
This paper presents an approach based on an intelligent fault-tolerant (FT) mechanism for enhancing a monopulse auto-tracking architecture through the integration of machine learning (ML) techniques. This approach is applied for the case of Oran ground station (GS) communicating in the S/X band frequency range to ensure system robustness during faulty events. The latter are due to the degradation or sudden failure of neuralgic radio frequency (RF) components, with an emphasis on the low-noise amplifier (LNA). To address these vulnerabilities, a probabilistic likelihood function is defined, and its statistical parameters are optimised using the grey wolf optimiser (GWO) in order to provide a means of automatic switchover to the ML tracking signal receiver (ML-TSR) control mode upon anomaly detection. Three ML algorithms, namely eXtreme Gradient Boosting (XGBoost), decision trees, and K-nearest neighbors (KNN), were implemented and evaluated using a database spanning eight years (2017–2024), including the operational parameters of the ALSAT-1B GS. The experimental results show that the KNN model achieves higher accuracy than the other approaches. A validation process was then conducted during an actual satellite pass encompassing a nominal phase followed by a GS LNA failure phase. During the nominal phase, a root mean square error (RMSE) of only 0.46° in azimuth and 0.64° in elevation was observed. Activation of our recovery mechanism during the second phase demonstrated that the robustness of the likelihood function, combined with the accuracy of the ML-TSR controller, ensures mission continuity without data loss, leading to four successful image downloads.