Manufacturing machines, especially those without computational capabilities, often require technicians to calibrate the machines to ensure the production line operates optimally. During the calibration process, there are times when the machine experiences transient faults, whose manual monitoring becomes unpractical and cost ineffective. In some cases, the machine's vibrations have regular beats, but when faults occur, there will be impulsive vibrations. This work addresses these cases and proposes a novel approach to detect packaging faults through monitoring the vibration of the machine by detecting impulsive vibrations from a series of beat vibrations. The novelty of the proposed solution is that it does not require calibration or fine-tuning. It uses Neural Basis Expansion Analysis for Time Series (N-BEATS) to learn and predict the beat vibration tempo. It detects impulsive vibrations by comparing incoming vibrations with the predicted beat using Spearman's rank correlation coefficient. Experimental results are promising, showing that the approach can achieve 97 accuracy in detecting impulsive vibrations. © 2025 IEEE.