Optimization Design of Myring Streamlined Volume Sink Based on NSGA-Ⅱ Genetic Algorithm

Abstract

In order to find a new type of quantitative tank with better hydraulic performance and broaden the research ideas in the field of quantitative tank, the obstruction state of the quantitative tank in the channel has certain similarities with the resistance of the submersible when diving, so based on the structural characteristics of Myring streamlined submersible, the body shape design of the measuring tank was carried out, and the better linear type and its hydraulic characteristics with the least hindered volume tank were explored. The length of contraction section, the sharpness factor of contraction section, the length of diffusion section and the departure angle of diffusion section were the variables, totally 40 sets of experimental schemes were designed for streamlined parameters by using the optimal Latin hypercube design method, and 40 sets of simulation data values were obtained based on FLOW-3D software. These data included percentage head loss and upstream Froude number, which were used to train RBF neural networks, and the coefficients of determination were 0.98916 and 0.99978, respectively, indicating that the neural network had high accuracy;the neural network was used as the fitness value of NSGA-Ⅱ genetic algorithm, and then the Patero frontier solution evaluation and screening by TOPSIS method were used to obtain the comprehensive optimal individual;the contraction segment length was 45.9cm, the contraction section sharpness factor was 0.74, the diffusion segment length was 49.2cm, and the diffusion angle was 14.63°. The corresponding head loss and upstream Froude number simulation values were 13.00% and 0.327, respectively, and the error of the prediction results did not exceed 5%. A total of six contraction ratios were obtained by equal scale scaling of the optimal individuals, and model experiments were carried out under nine groups of flow rates, and it was found that the Ferude number of upstream under different working conditions was less than 0.5, which met the requirements of the flow measurement specification, and the hydraulic performance was better when the shrinkage ratio was in the range of 0.58~0.66. Based on the principles of critical flow measurement and dimensional analysis, the accuracy of the current measurement formula was high, and the average relative error was 2.09%. The research result proved the feasibility of applying streamline to the field of quantitative sink research and optimization through neural networks and genetic algorithms. At the same time, it was shown that the optimized Myring streamlined water tank had good performance and high flow measurement accuracy, which had a wide application prospect in irrigation channels.

Keywords:streamlined volume sink;optimization of body shape;numerical simulation;neural network;NSGA-Ⅱ genetic algorithm