ABSTRACTA reciprocating hydrogen compressor status monitoring system for predictive maintenance is developed based on HHT (Hilbert−Huang Transform) with multiple functions, strong applicability, and high accuracy to address the problem of difficulty in identifying fault signals and failure to provide advance warning before faults occur in the reciprocating hydrogen compressor state monitoring system. Design framework of monitoring system is confirmed, and function modules are designed based on LabView platform. HHT is applied to monitor the status of reciprocating hydrogen compressor based on LabView platform. A reciprocating hydrogen compressor is selected as research object, status monitoring analysis is carried out. Five working states of reciprocating hydrogen compressor are collected, which conclude normal state, filler malfunction, cross‐head malfunction, air valve malfunction, and piston rod malfunction. HHT is carried out for five signals, and results show that HHT marginal spectrum of five signals has different characteristics. Based on comparison results, precision of HHT ranges from 0.757 to 0.784, recall of HHT ranges from 0.738 to 0.766, F1‐score of HHT ranges from to 0.788 to 0.804, HHT has better performance than other two methods. Proposed monitoring system designed in this study provides a comprehensive and efficient online monitoring and data analysis solution for reciprocating hydrogen compressors, which can achieve fault prediction of reciprocating hydrogen compressor, reduce failure rate, and effectively improve the reliability of the compressor oil injection system.
Read full abstract