Research Of Leakage Detection Technology In Natural Gas Pipeline Based Intelligent Algorithms

The pipelines leak detection technique is a kind of important tool at the aspect of ensuring safe running of pipeline. In recent years, many kinds of leak detection and location methods have been developed and used by scholars. However, due to the changing conditions, various methods have certain limitations in practical applications. Based on the flow equations of pipeline, the model based leak detection and location method can determine the leak location and size of leakage, and is identified as a promising method in leaks. In practical applications, the gas pipeline will be affected by many factors, sach as noise, temperature, friction coefficient, gas compression coefficient and so on, although the simplified model to reduce the computation, but will also decrease the accuracy of the model.In order to improve the accuracy of the models, a natural gas pipeline network model is established. It can compensate the effects of dynamic parameters on pipe model.In this paper, from study the simulation modelation method based on the mechanism of natural gas pipeline leakage, the teady state and transient simulation model of natural gas pipeline leakage are established. And solve them by using the four orders Runge Kutta method and the finite difference method. Rise using the time lag effect of compensation and the friction coefficient correction two aspects to enhances the influence of the leak location of pipeline parameters.According to the robust and global search characteristic of genetic algorithm, genetic algorithm to optimize neural network is used to solve the defect of convergence speed slow and easy to fall into local minimal value of BP algorithm. After a comprehensive analysis, genetic algorithm existing premature convergence problem is found. Therefore, the improvement based on standard genetic algorithm has been conduct. Based on the fitness value of the population division quality, take the respectively different ways for each different part. And then, supple and reallocat of each part to improve the ability of local search algorithm. Applicat complex two dimensional function tests to experiment. The results show that, the improved algorithm optimization result is better.In a comprehensive analysis based on the topological structure and search pattern of the particle swarm algorithm, the particle swarm algorithm is used to optimize neural network. A new improvement method is studied:In the early and late algorithm respectively conduct different disturbance. Make the population not only can search with high accuracy in the neighborhood of the global optimum solution, but also can from the neighborhood local extrema jump to the global optimal solution of the neighborhood. Applicat Standard test functions to experiment. The results show that the improved algorithm optimization result is better.This paper’s methods are applicated in natural gas pipeline to diagnosis leakage. The results show that, traditional model and mechanism of intelligent technology can make accurate judgments on the pipeline conditions. And the improved method is more accurate.