F. Zhang - College of Water Resources and Civil Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China
Y. Heng - School of Energy and Power engineering, Xihua University, Chengdu, Sichuan Province, ۶۱۰۰۳۹, China
Q. Liu - College of Water Resources and Civil Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China
R. Tao - College of Water Resources and Civil Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China
W. Yang - College of Water Resources and Civil Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China
D. Zhu - College of Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China
R. Xiao - College of Water Resources and Civil Engineering, China Agricultural University, Beijing, ۱۰۰۰۸۳, China

Semi-open centrifugal pump has the disadvantage of low efficiency in chemical industry for its unique structure. For the purpose of improving its efficiency as much as possible on the premise of meeting the head demand, in this case, an optimization strategy combining Bezier curve, Genetic Algorithm (GA), Fuzzy Logic (FL), Artificial Neural Network (ANN) and solution space model based on stratified sampling is proposed to optimize the efficiency of semi-open centrifugal pump. The fourth-order Bezier curve is used as parametric method to control the blade profile, GA combined with FL is used as optimization strategy to seek the optimal individual. Besides, the optimization solution space based on stratified sampling method and ANN training are used to shorten computing time. Finally, the semi-open centrifugal pump corresponding to the optimal parameter combination of blade profile has been found, and on the premise of meeting the head requirements, the efficiency has been significantly improved. Entropy production rate is carried out to analyze the reason of efficiency improvement. The whole optimization strategy proposed in this study provides a good reference for parametric control and performance optimization of bladed pump.

Centrifugal pump, Semi-open impeller, Genetic Algorithm, Fuzzy logic, Optimization design