Leakage Detection Of Pipe Network Based On Wavelet Analysis And Estimation Of Distribution Algorithm And Differential Evolution

With the rapid development of transportation engineering of the gas pipe in ourcountry, the construction of gas pipeline is more and more widely. And most of theexisting natural gas pipelines are in aging. Due to corrosion because of natural orman-made destruction, pipeline leakage phenomenon is more and more frequent.But the pipeline leak will cause environmental pollution and economic losses andeven cause personal injury and affect national security. In order to reduce the losscaused by the pipeline leakage and reduce the number of pipeline accidents, needmonitoring the pipeline in time and precise positioning. Therefore, pipeline leakdetection and location receives great importance to the academia all over the world.Ensuring the timeliness and accuracy of the pipeline leak detection becomes a hotresearch problem.Due to the limited factors of the natural gas transmission pipeline such as flowcharacteristics, the condition of terrain and the nature of the pipeline, it is difficult toestablish accurate model.Considering the used time of pipelines as well as the existing various leakdetection and location method, and according to the characteristics of the gaspipeline leakage, we propose a method to detect leaks and positioning based onwavelet analysis and distribution estimation algorithm and differential evolutionalgorithm. Mainly in the following several aspects:Firstly, this paper introduces the research status and development trend at homeand abroad of pipeline leak detection. Through comparison and analysis of differentleak detection methods, we use wavelet transform algorithm to realize pipeline leakdetection. Using wavelet analysis method to judge the leakage of natural gas signal,and selecting suitable wavelet base and threshold value function and the layernumber of wavelet decomposition, the leak signal wavelet is decomposed andreconstructed, And using the method of modulus maxima to realize leak position,more groups of time lag will be obtained by computing, do the preparatory work forthe subsequent signal analysis.Secondly, this paper introduces the principle of distribution estimationalgorithm, the probability model and optimization algorithm, and the estimation ofdistribution based on the Archimedes copulas algorithm was improved, solved theslow distribution estimation algorithm in the optimization and complex faults ofprobability model.Thirdly, the basic principle of differential evolution algorithm and the specificoptimization steps are introduced. After analyzing the basic principle and performance advantages in the differential evolution algorithm, the disadvantages that late slowconvergence speed and easy to fall into local optimal solution were solved bychanging two aspects of crossover probability factor and variation factor of algorithm.And verifying the improved algorithm through routine testing function on itsperformance, and comparing the simulation results in convergence speed andprecision of the optimal solution of the performance, the improved algorithm havevery big improvement, so the effectiveness of the algorithm can be proved.Finally, this paper introduces the estimation of distribution algorithm based onwavelet and pipeline leak detection of differential evolution algorithm. Simulationanalysis was carried out on the three algorithms respectively, contrast analysis of thepositioning accuracy of the algorithm. And comparative analyzing the improveestimation of distribution algorithm and differential evolution algorithm inconvergence speed, optimal time and complexity.