Before an anomaly or failure occurs, the machine will produce slightly abnormal vibration patterns. Install sensors on the machine to monitor its working condition. And use machine learning algorithms to learn its normal operating status directly from the machine. By analyzing the changes in vibration, it can be detected when an anomaly occurs. Similarly, this method is also suitable for other industrial equipment.