基于小波包熵与AFSA-SVM的压力管道泄漏识别

为了快速、准确地诊断出输气压力管道不同的泄漏状态,提出了一种基于小波包熵与人工鱼群优化支持向量机(AFSA-SVM)相结合的压力管道泄漏模式识别方法。该方法首先对管道泄漏时产生的声发射信号进行小波包分解,并对分解的最后一层节点重构信号进行相关性分析,以获得敏感的节点信号。然后求取这些敏感节点信号的小波包熵值,作为管道不同泄漏信号的特征向量。最后将小波包熵值输入到SVM中,并运用AFSA方法对SVM分类器中惩罚因子C与核函数参数g进行全局优化,以提高其分类准确率。实验结果表明,该方法能准确地识别压力管道不同的泄漏状态,为天然气管道泄漏状态监测提供新方法。

Operation Condition Monitoring of Pressure Pipeline Based on Wavelet Packet Entropy and AFSA-SVM

To diagnose the different leakage states of the gas pipeline quickly and accurately,a wavelet packet entropy and pattern recognition method of the leakage of gas pipeline for the optimizing support vector machine(SVM)by using artificial fish swarm algorithm(AFSA)was proposed.Firstly,the acoustic emission signal generated by the leakage of pipeline was decomposed by wavelet packet and the correlation of the reconstructed signal of the last layer of the reconstructed signal was analyzed to obtain the sensitive node signals.Then the wavelet packet entropy of these sensitive node signals is obtained,which was used as the characteristic vector of different leakage signals.Finally,the wavelet packet entropy was input into the SVM and the AFSA method was used to optimize the penalty parameter C and the kernel function parameter g in the SVM classifier,so as to improve the classification accuracy.The experimental results showed that this method can accurately identify the different leakage states of the pressure pipeline and provide a new method for monitoring the leakage status of the natural gas pipeline.