Optimization of gas-liquid two-phase flow liquid hold-up prediction model with RBF neural network based on genetic algorithm
Aiming at the difficulty of network topology and slow convergence of traditional Radial Basis Function(RBF)neural network in predicting liquid holdup of gas-liquid two-phase flow,a prediction model based on Genetic Algorithm(GA)optimized RBF neural network was proposed to improve the prediction accuracy of liquid holdup of gas-liquid two-phase flow.The collected experimental data were processed by the system clustering algorithm and Gray Relational Analysis(GRA)to select the optimal model characteristics.The RBF neural network structure parameters were determined by GA.The training was carried out based on the laboratory experimental data and compared with the Back Propagation(BP)neural network,GA-BP neural network,and RBF neural network,which are commonly used for liquid holdup prediction.The accuracy and feasibility of the model were evaluated.The results showed that the mean square error of the model is 0.001 7,the root mean square error is 0.0416,the mean absolute error is 0.028 1,and the fitting degree is 0.948 3.Compared with other neural network models,the prediction model shows higher calculation accuracy and stronger generalization ability.
万方数据
维普
