Design Of Fiber-optic Sensor For Pipeline Leakage And Research On Data Analysis Method

In recent years,the pipeline leakage generated by aging,corrosion and human destruction has caused environmental pollution and economic loss.In traditional detection method,pressure signal has the characteristics of large attenuation and low fidelity under the condition of long distance transmission and complex jamming environment.Therefore,the pipeline leak detection algorithm is inaccuracy due to the lowdetected signal-to-noise ratio and the submerging of real signal in noise.With the development of optical fiber technology,fiber grating provides an effective solution to the problem of detecting leak for the complex environment of long distance pipeline with the advantages of small size,high precision,intrinsic safety,anti-electromagnetic interference,corrosion resistance,low transmission loss,good fidelity,long transmission distance and so on.In view of the above discussion,this paper takes the pipeline leakage detection as the research object,introduces fiber optic sensing,and builds the pipeline leakage optical fiber detection system.A pipeline optical fiber pressure sensor with temperature compensation is designed.The sensor signal is processed by signal processing technology,and the leakage detection algorithm and alarm strategy are studied,using the sensing signal after processing as data source.The work done in this article is as follows:Firstly,a pipeline detection fiber pressure sensor with temperature compensation is designed and produced.The influence of temperature on the sensor is removed by the differential compensation method,the temperature drift of the sensor is solved,and the sensitivity of the sensor is increased.The COMSOL software is used to simulate the designed sensors under different parameters to determine the optimal selection of the sensor parameters.The sensitivity of optical fiber pressure sensor can reach 258.32 pm/MPa and the linearity can reach 0.998.The sensor is applied to the pipeline leakage optical fiber detection system for experimental test.Secondly,process the data obtain from the test of the optical fiber inspection system of the pipeline,including the detection and extraction of the peak value of the spectral signal,the filtering and interference removal of the signal,and the interpolation of the data.The optimal parameter filter is selected by comparing the signal to noise ratio(SNR)after filtering with different wavelet base parameters and decomposition layers.The simulation of signal missing,and different interpolation methods are used to compensate the data of different fluctuations.The data obtained after processing are the basis for data analysis.Thirdly,using the pipeline leakage detection algorithm to analyze the processed data information.The support vector machine(SVM)is used to determine the pipeline leakage detection.In order to eliminate the influence of the imbalance of data sample ratio,the penalty factor parameters are added to the target function.Aiming at the sensitive problem of support vector machine for parameter selection,genetic algorithm is used to find the appropriate penalty factor parameters so as to avoidthe limitations caused by artificial experience.Meanwhile we use the K-fold cross method to verify the result.Aiming at the problem of false alarm and small leak detection,different window strategies are designed to optimize it.Two test set samples are used to test and compare the results before and after optimization.