کنفرانس بین المللی ماشین و محرکه های الکتریکی

صفحه اصلی / The 4th International Conference on Electrical Machines and Drives

Electromagnetic Sensor Behavior Prediction Using LSTM Neural Network for Enhanced Accuracy and Reliable Performance

نویسندگان :

Ahmad Saasni¹ Mehrdad Ghafari² Mostafa Eftekharizadeh³

1- دانشگاه سمنان 2- شرکت جوانه نرم افزار 3- شرکت جوانه نرم افزار

کلمات کلیدی :

deep learning،high-accuracy data،industrial applications،long-term short-term memory (lstm) neural network،electromagnetic sensor،noise reduction،predictive modeling

چکیده :

Accurate sensors are essential in industries for system monitoring and control. High-precision sensor data can enhance predictions and prevent costly errors, especially in noisy environments that require detailed analysis. In this paper, we simulate the Long Short-Term Memory (LSTM) neural network model to predict the real magnetic signal from noisy data. The LSTM algorithm effectively removes Gaussian and random noise from the received data, providing accurate predictions of the real magnetic signals. Compared to previous works that used simpler methods, such as classical filters or basic models, our proposed algorithm demonstrates significantly higher accuracy. By using LSTM to remove noise and improve predictions, errors have been minimized, resulting in enhanced prediction accuracy. The simulation results indicate that the model can simulate the real behavior of the system with 97.67% accuracy, with minimal difference between the predicted and actual signals. This level of accuracy allows our model to significantly outperform previous methods in predicting magnetic signals in noisy environments and proves highly effective for industrial applications requiring precise accuracy.