壓縮機不穩定現象之偵測：結合模型與信號為基礎之技術

Abstract

本篇論文主要是在探討壓縮機系統不穩定現象之偵測與診斷議題，為了提升偵測效率及診斷可靠度，本論文結合了model-based與signal-based偵測與診斷方法之優點。在model-based部分，由於FIDF偵測方法具有快速及抗雜訊之特性，本論文選用FIDF設計法做為model-based的偵測方法，此外，為提升偵測的可靠度並提供壓縮機系統產生不穩定現象的類型，本論文採用FIDF所產生的殘量利用傅立葉轉換及自適應性模糊邏輯系統作為診斷的工具。模擬結果也驗証了所提出的整合方法之有效性。

This thesis studies the detection and diagnosis issues of compressor’s instabilities. In order to promote the detection efficiency and diagnosis reliability, this thesis combines the advantages of model-based and signal-based fault detection and diagnosis techniques. In this study, Fault Identification Filter (FIDF) design method was adopted as the model-based method because the FIDF design has the features of fast response and insensitivity to noises. In addition, the residual signal produced by FIDF design was employed to improve the detection reliability and to diagnose the type of instabilities by an adaptive fuzzy logic system. Simulation results demonstrate the effectiveness of the proposed scheme.