基于小波分析和神经网络的供水管网管内泄漏声检测方法研究

摘要

传统固定声传感器方式的泄漏检测方法可用于供水管道中的泄漏定位。然而，这些方法需要部署大量的传感器，因此难以在埋地长管线上进行应用，也很难实现提早预警以及评估泄漏尺寸大小。有鉴于此，本团队研发了以声学传感器为核心的管内声学泄漏检测装置，该装置依托水流运动，记录所有音频信号并通过分析检测微小泄漏。该自漂流式装置内部搭载的高灵敏度声学传感器，能够检测带压供水管道中的微小泄漏。由于管内环境有许多因素干扰水声信号，例如水流噪音等。因此，对管内泄漏的检测及其泄漏量的分析仍具有很大的挑战性。本文提出了一种新的基于小波变换的管内水声信号泄漏检测方法。在这种方法中，首先监测相对较长时间间隔内的水声信号，通过短时傅立叶变换(STFT)获取泄漏信号特征，通过准确地选择母小波（调谐小波）获得在时频域中目标信号的定位。然后，提出了组合小波分析和神经网络的算法结构，利用小波变换分析泄漏信号并利用人工神经网络进行分类。该研究表明，时域方法难以表征带噪声泄漏信号的完整特征，选择合适的母小波来提取供水管道中噪声事件特征具有重要作用。本文提出的方法已在工程中获实际应用，并在设计的实验平台上进行了测试，对自漂流式的装置采集的声学信号进行分析。采集到的声学信号被用于识别干扰信号（瞬时管道振动，水流噪声，管道固有频率和背景噪声）和泄漏特征信号。仿真结果表明，基于神经网络，并通过选择合适的母小波可定位泄漏特征信号和背景噪声的位置，该方法提高了提取特征的分类性能。此外，还验证了通过结合管内声学装置及优化后的小波变换和神经网络方法，可以高效可靠地进行管内泄漏检测。

目录

声明

摘要

ACKNOWLEDGEMENT

PUBLICATIONS／RESEARCH ACHIEVEMENTS

Table of Contents

List of Figures

List of Tables

1．INTRODUCTION

1．1．Problem Statement

1．2．Scope and Emphasis

1．3．Outline of Dissertation

2．LITERATURE REVIEW

2．1．Leak Detection Methods

2．2．Hardware-based Out-of-pipe／fixed Methods

2．2．1．Acoustic Leak Detection Methods

2．2．2．Non-Acoustic Leak Detection Methods

2．2．3．Application Analysis of Hardware-based Out-of-pipe／Fixed Methods

2．3．Hardware-based In-line／free-swimming detection Methods

2．3．1．Smartball

2．3．2．Sahara

2．3．3．PipeDriver

2．3．4．PipeGuard

2．3．5．Application Analysis of Hardware-based In-line／Free-swimming Methods

2．4．Software-based methods

2．4．1．Numerical Methods

2．4．2．Non-numerical model methods

2．5．Summary

3．IN-LINE ACOUSTIC LEAKAGE DETECTION DEVICE AND TEST SYSTEM

3．1．Introduction

3．2．In-line Acoustic Device Technical Approach

3．3．Design and Development

3．3．1．Embedded System Design

3．3．2．Mechanical Design

3．4．Test System

3．4．1．Controlled Experiments Designing Methodology

3．5．Summary

4．UNDERWATER MEASURED SIGNAL PROCESSING AND LEAK DETECTION METHODOLOGY IN WAVELET TRANSFORM

4．1．Introduction

4．2．Underwater Acoustic Signal Processing Methodology in Wavelet Transform

4．2．1．Wavelet Transform

4．2．2．Significance of Selection of Optimum Mother Wavelet in Underwater Acoustic Signals

4．3．General Method of Underwater Leakage Identification

4．4．Experimental Methodology

4．5．Leak Detection

4．5．1．Acoustic Signal Signature Identification

4．5．2．Selection of Optimal Mother Wavelet

4．5．3．Clustering and Acoustic Signal Identification

4．6．Summary

5．IN-LINE ACOUSTIC DEVICE LEAKAGE DETECTION BASED ON WAVELET AND NEURAL NETWORK

5．1．Introduction

5．2．In-line Acoustic Signal Processing and Significance of Wavelet transform in Water-filled Pipes

5．2．1．Wavelet Transfolrm

5．2．2．The importance of Selection of Mother Wavelet in Water Distribution Pipes

5．3．In-line Leakage Signal Features Classification Based on Neural Network

5．3．1．Neural Network

5．3．2．Back-Propagation Neural Network Based Classification

5．3．3．Optimization of BPNN model for Leakage Signal Classification

5．4．Experimental Methodology

5．5．Leak Detection and Classification

5．5．1．Frequency Analysis and Selection of Mother Wavelet of in-line inspected Acoustic Signal

5．5．2．Leak Signal Classification Based on Optimized BPNN

5．6．Summary

6．CONCLUSIONS AND FUTURE WORKS

6．1．Conclusions

6．2．Future Works and Suggestions

REFERENCES