Objective Centrifugal compressor units are critical power components of long-distance natural gas pipeline systems. Their faults not only disrupt the stability of gas transportation but also lead to high maintenance costs and pose safety risks. Therefore, timely and accurate detection of abnormal vibrations in these units is essential to prevent fault escalation and ensure safe, reliable operation.

Methods A comprehensive alarm method for centrifugal compressor units based on interval, rapid change and overlimit alarm strategies was proposed. The vibration characteristics of electric- and gas-driven units were thoroughly analyzed, and two key parameters—full-frequency vibration value and speed—were selected. The confidence interval for real-time full-frequency vibration value was predicted using the mean and standard deviation of historical full-frequency vibration values and the speed’s standard deviation. An alarm was triggered when the vibration values at two measurement points on the same unit simultaneously exceeded the confidence interval. To address sudden changes in the full-frequency vibration value at a single point caused by signal interference, loose wiring, or sensor failure, a rapid-change alarm strategy was introduced. To cover shutdown, start-stop, and normal operation states, an overlimit alarm strategy was developed to detect extreme full-frequency vibration values and signal interruptions. Alarm execution strategies for interval, rapid-change, and overlimit conditions were formulated according to the compressor’s operating state, leveraging the strengths of each strategy to ensure low false-alarm and missed-alarm rates.

Results By integrating the actual vibration patterns of centrifugal compressor units with the requirements of the interval, rapid change and overlimit alarm strategies, specific data for key parameters were established. The effectiveness of the proposed comprehensive alarm method was then validated using both normal and fault vibration cases. Results demonstrated that the method effectively eliminated the influence of electrical and mechanical jitter as well as speed adjustments. With straightforward, targeted parameter adjustments, it accurately identified abnormal phenomena—including sudden, abrupt, and gradual changes in full-frequency vibration—while maintaining a low false alarm rate.

Conclusion The proposed method enables on-site and remote technicians to promptly detect abnormal vibrations in centrifugal compressor units, preventing fault escalation and damage. It supports the development of unmanned or minimally manned stations, black-screen management, and the intelligent operation of long-distance natural gas stations.