Research Of The Three-dimensional Detection Method On Vibration Fault Of The Pipeline

In power and petrochemical companies, pipeline has been widely used to conveyfluid and the pipeline vibration is a common phenomenon. If the pipe vibration is toosevere, it may rupture, leading to leaks and explosion. If it can be found in early times,the staff can immediately take appropriate preventive measures to repair theequipment and eliminate the failure to protect the property and personal safety. Sofault diagnosis method for piping vibration study has a very important theoretical andpractical significanceBased on the characteristics of pipe vibration, the common faults of the pipevibration are simulated through experiment, at the same time, the three-axisacceleration sensor is used to collect pipe vibration signals in all kinds of operationconditions. So the comprehensive vibration for the pipe can be obtained and it canimprove the diagnostic accuracy.Wavelet packet method is developed on the basis of multi-resolution and it is amore sophisticated method of time-frequency analysis. It can decompose signal inboth high frequency and low frequency at the same time and select the appropriatefrequency-band for analysis. In this paper, the wavelet method is used for pipevibration signal de-noising to remove redundant information and wavelet packetmethod is used to decompose three-channel vibration signals, Therefore, vibrationsignal is decomposed into multiple frequency-bands. Full information method is usedto construct three-dimensional feature vector of energy to signify the fault conditionof the pipeline vibration. At the same time, the one-way signal is used to constructenergy eigenvector.For the problem of insufficient samples on fault diagnosis, the support vectormachine is applied to fault diagnosis of the pipeline. In the paper, radial basisfunction as kernel function, parts of the samples are used to train the support vectormachine and the rest of samples is used to test the training model. It proves that the three-dimensional method has higher accuracy than one-dimensional for pipelinevibration.