In the contemporary industrial landscape, machinery stands as the cornerstone of various sectors. Over time, these machines undergo wear and tear due to extensive use, leading to the introduction of subtle faults into the machine readings. Recognizing the pivotal role of machinery in diverse industries, the timely detection of these faults becomes imperative. Early fault detection is crucial for preventing costly downtimes, ensuring operational efficiency, and enhancing overall safety. This paper addresses the need for an effective condition monitoring and fault detection system, focusing specifically on the application of the Long Short-Term Memory (LSTM) deep learning model for fault detection in bearings using accelerometer data. The preprocessing phase involves extracting time domain features, encompassing normal, differentiated, integrated, and carefully selected signals, to create an informative dataset tailored for the LSTM model. This model is then meticulously trained on the dataset to discern and accurately diagnose faults within the machinery. The research meticulously observes and reports that the LSTM model achieves an impressive 100% accuracy in fault detection, showcasing its robust capabilities in identifying subtle anomalies within the machine vibrations. In conclusion, the study underscores the critical importance of early fault detection in industrial machinery and highlights the efficacy of the LSTM model in this domain. The singular focus on the LSTM model demonstrates its proficiency in achieving accurate fault detection, contributing significantly to the predictive maintenance field. This research not only advances fault detection methodologies but also fosters a more reliable and sustainable industrial landscape, emphasizing the potential of deep learning techniques, particularly the LSTM model, in ensuring the optimal performance and longevity of machinery in diverse industrial settings.