الكشف عن الأنماط الحركية في الكف الصناعي باستخدام الشبكات العصبونية الالتفافية

Abstract

In this research we Introduce an artificial intelligence (AI) model based on Convolutional Neural Networks (CNNs) to classify electromyographic (EMG) signals associated with hand movements such as Closing, Scissor, Opening, and the "OK" sign. EMG signal analysis is a vital field in medical and engineering applications, where it is used in the development of smart prosthetics, rehabilitation devices, and gesture control systems.

The database is obtained from the Kaggle website. A MYO Myoelectric Sensor collects the electromyographic signal data, which then undergoes processing that includes data standardization to ensure model accuracy. Subsequently, a multi-layer Convolutional Neural Network is designed, where convolutional layers extract distinctive patterns from the signals, while dense layers analyze the extracted features to make accurate classification decisions.

The designed model is trained using performance optimization techniques such as Dropout to reduce overfitting. The results show that the model achieves a classification accuracy of 97% on the test data, which reflects its efficiency in recognizing EMG signal patterns. Furthermore, the performance curves demonstrate stability during training, indicating the model's ability to generalize without experiencing overfitting.