Investigation of input variables influence in patterns learning of fluid flow behavior using fuzzy differential evolution

Abstract

We examine flow characteristics within the pipeline using computational fluid dynamics (CFD) and the fuzzy differential evolution (DEFIS) technique. The CFD findings are being used in the machine learning training mode. Evolution is used as a training method in artificial intelligence. Then, we utilize fuzzy inference systems in the decision mode. A distinct pattern of inputs and outputs is used to select the optimum category for training intelligent algorithm techniques. We analyze two different patterns of inputs and outcomes using two separate case studies. The first case study looks at how various inputs (i.e., nodes' coordinates, edge length ratio, element volume ratio, turbulence eddy frequency, gas volume fraction, and velocity components) change as pressure changes. In another analysis, we look at various inputs as a function of gas velocity. We discovered that a classification technique is a fantastic way to train intelligent algorithms. The best categorization technique reduces the computing time required for the DEFIS method's training phase. The classification improves the model's accuracy; furthermore, we examine a new pattern recognition technique in this research. In this instance, we compute various outputs as inputs and then compare pattern recognition techniques for artificial intelligence (AI) and CFD. This integration approach, which combines CFD findings with AI, can substantially decrease computing time throughout the CFD optimization process. Furthermore, AI structures may link input and output parameters to show a meaningful correlation between process parameters and productivity.