基于PCA和BP神经网络的内检测技术在油气集输管道中的应用

Application of Internal Detection Technology Based on PCA and BP Neural Network in Oil and Gas Gathering and Transportation Pipelines

为了解决超声导波内检测技术在油气集输管道进行缺陷检测时缺陷识别率低的问题,在对国内外现状进行充分调研的基础上,结合现有的PCA(主成分分析)和BP神经网络技术,建立缺陷信号识别模型,利用PCA和BP神经网络相结合的技术对管道信号进行识别。通过实例验证,在基于PCA和BP神经网络的超生导波管道信号识别中,对凹坑、孔洞、裂缝等缺陷分别进行了准确率计算,综合准确率可达95%。由此得出结论:对于油气集输管道缺陷问题,通过PCA和BP神经网络相结合技术的超声导波信号识别相比其他识别方法更为准确,该检测技术提高了超生波内检测对管道缺陷的识别能力,具有一定的推广价值。

In order to solve the problem of low defect recognition rate when using ultrasonic guided wave detection technology to detect defects in oil and gas gathering and transportation pipelines,based on the full investigation of the current situations at home and abroad,combining the existing PCA(principal component analysis)and BP neural Network technology,a defect signal identification model is established to use PCA and BP neural network technology to identify the pipeline signal.By the way of example verification,in the ultrasonic guided wave signal recognition based on PCA and BP neural network,accuracy rates of pits,holes,cracks and other defects are separately calculated,and the comprehensive accuracy can reach 95%.Then it is concluded that the ultrasonic guided wave signals of oil and gas gathering and transportation pipelines can be identified by the combination of PCA and BP neural network,and compared with other signal recognition methods it is more accurate and improves the recognition capacity of ultrasonic internal detection for pipeline defects,so the detection technology has certain promoting value.