ABSTRACT
Centrifugal compressor is a typical air compressor, which is an important subcomponent of the air supply system in fuel cell system. Optimizing the designing structure of centrifugal compressor plays significant influence on the output performance of fuel cell systems. However, existing experimental and numerical methods suffer from much economic and time cost and are inadequate for designing optimized centrifugal compressor. Thus, we develop a novel artificial intelligence (AI) framework integrated the data-driven surrogate model and stochastic optimization algorithm to achieve multi-objective optimization of the centrifugal compressor impeller. With the database obtained from the constructed three-dimensional (3D) steady-state centrifugal compressor model, the data-driven surrogate model based on Support Vector Machine (SVM) is trained. Then, the surrogate model coupled with a non-dominated sorting Genetic Algorithm (NSGA-III) is used to obtain the optimal solution of structural parameters. Compared with the original compressor design based on the established 3D model, the optimized compressor is comprehensively verified. Within the working range of the centrifugal compressor, the pressure ratio and isentropic efficiency of the optimized compressor have been significantly improved. The proposed optimized method is effective for the performance improved in fuel cell centrifugal compressor.
Nomenclature
ANN | = | artificial neural network |
BPNN | = | back propagation neural network |
CFD | = | computational fluid dynamics |
FCV | = | fuel cell vehicles |
HEV | = | hybrid electric vehicles |
ICE | = | internal combustion engine |
ICEV | = | internal combustion engine vehicles |
NSGA | = | non-dominated sorting genetic algorithm |
PEMFC | = | proton exchange membrane fuel cell |
PHEV | = | plug-in hybrid electric vehicles |
RANS | = | Reynolds-averaged Navier-Stokes |
RBNN | = | radial basis neural network |
SST | = | shear stress transport |
SVM | = | support vector machine |
SVR | = | support vector regression |
3D | = | three-dimensional |
Symbol | = | |
b | = | impeller outlet width (mm) |
cp | = | specific heat capacity (J kg−1 K−1) |
D | = | diameter (mm) |
I | = | current density (A cm−2) |
m | = | mass flow rate (kg s−1) |
p | = | pressure (Pa) |
R | = | radius (mm) |
T | = | temperature (K) |
U | = | velocity (m s−1) |
Greek letters | = | |
β | = | impeller blade angle |
δ | = | blade thickness (mm) |
η | = | isentropic efficiency |
ξ | = | stoichiometry |
π | = | pressure ratio |
ρ | = | density (kg m−3) |
σ | = | tip clearance (mm) |
Subscripts and superscripts | = | |
1 | = | inlet of impeller |
2 | = | outlet of impeller |
3 | = | inlet of diffuser |
4 | = | outlet of diffuser |
h | = | hub |
i | = | generation number |
s | = | shroud |
t | = | total |
Acknowledgments
Present study is supported by the National Key Research and Development Program of China (2018YFB0105505) and the National Natural Science Foundation of China (Grant No. 51976138).