Parameter Identification of Hydraulic Low Pressure Pipeline Transient Model Using Parallel Genetic Algorithms
Parameter identification of pressure transient models accompanying cavitation and gas bubbles inside low hydraulic pipelines using parallel genetic algorithms (PGA) in a cluster of computers is presented. In this paper, pressure transient models are given to describe the flow behavior in a pipeline, and the numerical models of cavitation and gas bubbles are built to calculate the cavitation volume and gas bubble volume during the transients. The unknown parameters of the pressure transient mathematical models include the initial gas bubble volume, gas resolving time and gas releasing time constants. Parameter identification of pressure transient mathematical models with genetic algorithms (GA) on a single computer usually takes seven or eight days, which is out of expectation. The execution time of GA becomes high due to time consuming of evaluating fitness function. In order to predict the pressure transients and shorten the execution time of GA, MATLAB Distributed Computing toolbox and MATLAB Genetic Algorithms toolbox are used to perform evaluation of fitness function at each generation by PGA, which is a parallel calculation on a cluster of computers. Based on the minimization of the least-square errors between experimental data and simulation results, PGA is applied to search for global optimal model parameters. Acceleration rate of computation time using PGA to using GA is acquired. Simulation results with identified parameters obtained by PGA are given. Comparison of simulation results with experimental data indicates that PGA is feasible to estimate unknown parameters in hydraulic low pressure pipeline transient model.
parallel genetic algorithms parameter identification pressure transients
Yang Chifu Li Songjing Jiang Dan
Department of Fluid Control and Automation, Harbin Institute of Technology, Harbin, Heilongjiang, 150001, China
国际会议
北戴河
英文
502-506
2007-06-06（万方平台首次上网日期，不代表论文的发表时间）