复杂工况下热油管道泄漏识别与定位方法研究

LEAK DETECTION AND POSITIONING FOR HOT OIL PIPELINE UNDER COMPLICATED CONDITIONS

针对复杂工况下热原油管道泄漏难以准确识别的难题,提出采用基于多元支持向量机的管道泄漏诊断方法,并建立了识别模型,可以在小样本情形下完成模型的训练工作,实现多种工况下对压力波动信号的分类识别,从而提高评判泄漏的有效性和准确性。针对热油管道负压波波速受油品及温度等因素影响较大所导致的定位误差,分析了管道沿程轴向温降以修正负压波波速,并采用牛顿-柯特斯积分方法对传统泄漏定位公式进行了改进。现场实验表明,基于多元支持向量机的检测方法能有效地识别管道运行异常状态,改进的漏点定位算法使得定位精度从原来的2.5%提高到1.0%。

For overcoming the difficulty of leak detection for hot oil pipeline under complicated conditions,a method based on MSVM(Multi-Support Vector Machine)is proposed and the diagnosis model is established.The model training can be completed in a few samples to distinguish different conditions of pipelines.Moreover,in hot oil pipeline negtive pressure wave,velocity is affected by oil and pipeline axial temperature drop.Positioning usually has obvious error.For solving this problem,axial temperature drop is analyzed and pressure velocity is revised.By means of Newton-Cotes integration method,positioning formula is improved.The field experiment shows that abnormal conditions including leakage of pipeline can be distinguished effectively by using the method based on MSVM and the improved positioning formula makes the positioning more accurate,from 2.5% up to 1.0%.