Automatic Identification Of The Pipeline Undetected Signal

The pipeline transportation system which is one of five big transportation professions holds the very high status in the national economy, whose characteristics are to protect environment,to save money,to transport quickly. The Daqing Oil Field is our country's petroleum important habitat, the petroleum transports from the pipeline to many areas of our country, supports the national economic construction. But as a result of influences such as pipeline natural life-span and human factor and so on, the accident of the pipeline leakage occurs frequently, has brought about the massive economic losses, the environmental pollution, the significant person casualties and so on. The operations of discovering pipeline leakage in time and fixing position accurately are extremely important to reduce the losses and to maintain the work of pipeline safely.Thinking the fact that the petroleum pipeline is restricted by many complex factors such as the fluid characteristic, the Physical feature of a place and the pipeline own characteristic and so on, simultaneously unifying the actual situation of the Daqing Oil Field pipeline transportation, and profiting from the various present international advanced method about leakage detected and localized , being aimed at the characteristic of petroleum pipeline leakage malfunction, this article proposes the technology of detecting leakage and fix position based on the combination of the negative pressure wave wavelet analysis and the BP neural network. Regarding this, this article mainly carries on such work:the collecting signal is located by using the time difference of the transient state negative pressure wave divulging localization method, This needs certain high-accuracy sensor, and will unify the system time of the first terminal data acquisition system. the collecting signal contains the many noise jamming, this article applies the wavelet to remove noise, adopts the MALLAT algorithm and porous algorithm to reconstruct, obtains a signal that its characteristic inflection point is quite obvious, makes use of the Lipschitz exponent of the wavelet transformation to detect the Extremely large value of the signal center mold and the leakage point of the signal. Possibly there are several extremely large value spots in a signal, Sends the maximum spot to the BP neural network to carry on the training by using the certain method; makes the leakage sample storehouse to recognize the leakage signal. This article uses the MATLAB language to deal with the laboratory signal, in order to prepare for the industrial application, may continue to use the C++ or VC language to debug.