Cooling System Design Optimization of an Enclosed PM Traction Motor for Subway Propulsion Systems

This paper presents the design optimization of a self-circulated ventilation system for an enclosed permanent magnet(PM)traction motor utilized in the propulsion systems for subway trains.In order to analyze accurately the machine's inherent cooling capacity when the train is running,the ambient airflow and the related heat transfer coefficient(HTC)are numerically investigated considering synchronously the bogie installation structure.The machine is preliminary cooled with air ducts set on the motor shell,and the fluidic-thermal field distributions with only the shell air duct cooling are numerically calculated.During simulations,the HTC obtained in the former steps is applied to the external surface of the machine to model the inherent cooling characteristic caused by the train movement.To reduce the temperature rise and thus guarantee the motor's working reliability,an internal self-circulated air cooling system is proposed according to the machine temperature distribution.The air enclosed in the end-caps is driven by the blades mounted on both sides of the rotor core and forms two air circuits to bring the excessive power losses generated in the heating components to cool regions.The fluid flow and temperature rise distributions of the cooling system's structural parameters are further improved by the Taguchi method in order to confirm the efficacy of the internal air cooling system.

Fault Diagnosis of Photovoltaic Array Based on Deep Belief Network Optimized by Genetic Algorithm

When using deep belief networks(DBN)to establish a fault diagnosis model,the objective function easily falls into a local optimum during the learning and training process due to random initialization of the DBN network bias and weights,thereby affecting the computational efficiency.To address the problem,a fault diagnosis method based on a deep belief network optimized by genetic algorithm(GA-DBN)is proposed.The method uses the restricted Boltzmann machine reconstruction error to structure the fitness function,and uses the genetic algorithm to optimize the network bias and weight,thus improving the network accuracy and convergence speed.In the experiment,the performance of the model is analyzed from the aspects of reconstruction error,classification accuracy,and time-consuming size.The results are compared with those of back propagation optimized by the genetic algorithm,support vector machines,and DBN.It shows that the proposed method improves the generalization ability of traditional DBN,and has higher recognition accuracy of photovoltaic array faults.