Accurately estimating a machine’s Remaining Useful Life (RUL) is essential for improving maintenance efficiency and ensuring operational reliability. This study proposes an Artificial Neural Network (ANN) to enhance RUL prediction for bearing failures. The model processes both historical and real-time data to improve accuracy. The ANN takes root mean square, kurtosis, time, and adjusted Weibull failure rate as inputs. These features capture essential degradation patterns. To reduce noise, the model uses standardized operational duration percentages as output. This approach improves stability and prediction reliability.A Feedforward Neural Network optimized with the Levenberg-Marquardt algorithm is implemented. Training and validation assess its performance. The model architecture includes two hidden nodes, ensuring a balance between complexity and efficiency. Results show low training and validation errors, confirming the network’s accuracy. The proposed method provides a robust tool for predictive maintenance. By detecting early degradation, it helps optimize maintenance schedules and reduce unexpected failures. Compared to traditional approaches, this model offers better adaptability to non-linear degradation trends. Future work will focus on extending validation to different bearing types and operating conditions. This research contributes to the advancement of predictive maintenance strategies by integrating advanced machine learning techniques for more precise RUL estimation.