蒸汽渦輪發電機故障診斷之研究

A Study on Faults Diagnosis of the Steam Turbine Gearbo

本研究目的是以振動診斷監測模式，針對海軍某艦艇之蒸汽渦輪發電機在故障發生初期進行診斷分析。由於大型旋轉機械系統組成複雜，故障類型非常繁多，在組件相互影響下，致使主要與次要故障原因難以區隔。因此必須利用各種不同的診斷方法，針對複雜多元的故障現象進行有效及可靠的診斷。本研究以自撰Matlab程式，運用不同的訊號分析處理方法，對艦用蒸汽渦輪發電機的故障現象進行診斷，其中包括功率倒譜（Power Cepstrum）、小波轉換（Wavelet Transform）及雙譜（Bispectrum）等。由本研究所得結果可知，上述各種訊號分析方法的組合運用，能更有效掌握旋轉機械之運轉狀況及診斷可能發生的故障原因。如此也可以協助維修人員提出正確之有效維修模式，以提昇艦艇上各運轉裝備的操作妥善率。 The main purpose of this study is to use condition monitoring and diagnosis to analyze the faults of a steam turbine generator of a naval ship. Due to the complexity of large rotational machinery systems, several types of fault may occur, and usually, it is difficult to diagnose the main faults exactly. In this study, signal processing methods, including power spectrum, power cepstrum, wavelet transform and bispectrum, are used to perform vibration diagnoses of rotational machinery systems. From the results, it is shown that a combination of the these signal analysis tools, can be used for a more reliable condition monitoring for rotary machinery, and thus, providing personnel a more efficient maintenance schedule, and increasing the operating hour of the machinery on board.