Design and multi-objective optimization of hairpin motor cooling system based on orthogonal design method and composite algorithm

•Reducing variables appropriately can improve the efficiency of algorithm fitting.•Orthogonal design method effectively improves the efficiency of dataset fitting.•The PSO-BP composite algorithm has high fitting accuracy.•The PSO-BP-NSGA III composite algorithm has a good optimization effect. The ne...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Applied thermal engineering Ročník 259; s. 124971
Hlavní autoři: Chen, Song, Xu, Pingxiang, Liu, Yuxi
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Ltd 15.01.2025
Témata:
Backpropagation neural network
Hairpin motor
Multi-objective optimization algorithms
Oil cooling system
Orthogonal design method
Orthogonal design method
Hairpin motor
Backpropagation neural network
Oil cooling system
Multi-objective optimization algorithms
ISSN:1359-4311
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:•Reducing variables appropriately can improve the efficiency of algorithm fitting.•Orthogonal design method effectively improves the efficiency of dataset fitting.•The PSO-BP composite algorithm has high fitting accuracy.•The PSO-BP-NSGA III composite algorithm has a good optimization effect. The new thermal management structure of the motor is designed to ensure effective heat dissipation while maintaining low pressure loss, reduce the required power of the cooling pump. The dimensions of the new thermal management structure of the motor are optimized through the establishment of a multi-objective optimization platform. Computational Fluid Dynamics (CFD) simulations were conducted through single-factor analysis to compare and preliminarily determine the number of turns in the spiral channel. The key structural dimensions of the flow channel were adjusted using orthogonal design method for CFD simulations. The obtained complete orthogonal table was used as a dataset for comparing and analyzing the optimization of the Backpropagation (BP) neural network using the Genetic Algorithm (GA) and the Particle Swarm Optimization (PSO) algorithm. Results showed a higher fitting accuracy after optimization with the Particle Swarm Optimization algorithm. The Non-dominated Sorting Genetic Algorithm III (NAGA III) was employed for multi-objective optimization of the key flow channel dimensions. The optimal solution identified from the Pareto optimal solution set graph, which is the lowest point combining the highest stator temperature and maximum pressure in the flow channel, was verified through simulation to have a prediction accuracy of 99% for the comprehensive optimization algorithm. Furthermore, the simulation results of the cooling system obtained from multi-objective optimization were compared with the original cooling system, the comparison results show that the maximum pressure has decreased by 43.13%. Orthogonal design methods and composite algorithms are employed to replace traditional optimization methods for multi-objective optimization of motor cooling system in this study, offers a new perspective for the structural design of motor cooling systems.
ISSN:1359-4311
DOI:10.1016/j.applthermaleng.2024.124971