PVC 管線洩漏聲訊號檢測方法之設計與實驗驗證

Abstract

本論文提出了一種基於聲學方法的低成本無線感測器，旨在實現PVC管線洩漏點的精準檢測與定位。 有鑑於管線洩漏對水資源、能源運輸及環境保護造成的重大影響，本研究結合三軸加速規與無線傳輸技術，設計並搭建了一套模擬實際管線洩漏情境的實驗平台。透過滑動窗口、互相關、快速傅立葉轉換(FFT)、短時傅立葉轉換(STFT)以及離散小波轉換(DWT)等信號處理方法，結合色散曲線與到達時間差(TDOA)技術，本研究能夠有效分析洩漏引發的振動訊號，並估算洩漏點的空間位置。實驗採用六分PVC管系統，模擬不同洩漏位置，驗證了所提方法的有效性。 結果顯示，互相關方法在短距離管線中的定位準確度優於小波轉換，平均誤差率較低，尤其在時域分析中表現穩定。然而，小波轉換在處理非平穩訊號時具備理論優勢，但受限於頻率解析度及計算複雜度，實際應用效果不如預期。本研究進一步比較了各方法的計算複雜度與適用情境，分析其在不同管線條件下的表現差異。 針對未來發展，本研究提出可從管道長度、材質（如金屬管與非金屬管的差異）、洩漏孔徑以及環境因素等面向進行延伸實驗，以提升聲學檢測技術的穩健性與實用性，為管線洩漏監測提供更高效且經濟的解決方案。

This thesis proposes a low-cost wireless detection device based on acoustic methods for the precise detection and localization of leaks in PVC pipelines. Given the significant impact of pipeline leaks on water resources, energy transportation, and environmental protection, this study integrates triaxial accelerometers with wireless transmission technology to design and construct an experimental platform simulating real-world pipeline leak scenarios. By employing signal processing techniques such as sliding window, cross-correlation, Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), and Discrete Wavelet Transform (DWT), combined with dispersion curves and Time Difference of Arrival (TDOA) methods, this research effectively analyzes vibration signals induced by leaks and estimates the spatial location of leak points. Experiments were conducted using a 3/4-inch PVC pipeline system, simulating various leak positions to validate the proposed method’s effectiveness. Results demonstrate that the cross-correlation method outperforms wavelet transform in localization accuracy for short-distance pipelines, exhibiting lower average error rates, particularly in time-domain analysis where it shows stable performance. However, while wavelet transform holds theoretical advantages in processing non-stationary signals, its practical performance is limited by frequency resolution and computational complexity. This study further compares the computational complexity and applicability of each method, analyzing their performance differences under varying pipeline conditions. For future development, the study suggests extending experiments to explore factors such as pipeline length, material (e.g., differences between metallic and non-metallic pipes), leak aperture, and environmental factors to enhance the robustness and practicality of acoustic detection technology, thereby providing a more efficient and cost-effective solution for pipeline leak monitoring.