Introducing the Effective Features Using the Particle Swarm Optimization Algorithm to Increase Accuracy in Determining the Volume Percentages of Three-Phase Flows

What is presented in this research an intelligent system for detecting the volume percentage of three-phase fluids passing through oil pipes. The structure detection consists X-ray tube, a Pyrex galss pipe, and two sodium iodide detectors. A fluid water, gas, has been simulated inside pipe flow regimes, annular stratified. Different percentages from 10 to 80% are considered each phase. After producing emitting X-rays source containing fluid, intensity photons recorded by simulation introduced Monte Carlo N-Particle (MCNP) code. implementation all regimes different percentages, signals detectors were labeled. Three frequency characteristics five wavelet transform extracted received detector, which collected total 16 test. feature selection based on particle swarm optimization (PSO) algorithm was applied determine best combination features. result introduction seven features as percentages. input Multilayer Perceptron (MLP) neural network, whose had neurons (selected characteristics) output (volume gas water). highest error obtained determining equal 0.13 MSE, low compared with previous works. Using PSO select most optimal features, current research’s accuracy significantly increased.