Research On Intelligent Design Method Of Motor Based On Cloud Computing

Research and manufacturing of high performance electrical equipment can make China to tansfer from electrical equipment large-country to electrical equipment,in accordance with the background of Intelligent Manufacturing in China.Typical electrical equipment contain motor,transformer and so on,which is researched and designed in favor of their research,manufacture and putting into production.Numerical analysis of equivalent magnetic circuit and manual calculation of according to the orthogonal table is traditional motor analysis and comparison method,this type of method usually exist deficiency for instance long design cycle,tedious design process and low accuracy.In the actual project,the motor intelligent research and design is usually has trait such as large calculation amount and data intensive mode.Efficient and real-time computing resources can be provided by cloud platform with flexible,dynamic and scalable features.Computing tasks is assigned to multiple virtual machines,so the calculation design time is shorten by running in parallel.Design motor by intelligent algorithm is a programmed method,which has advantage such as short design cycle,simple design process and high accuracy.So in this dissertation,intelligent design method of motor based on cloud computing is proposesed.In the first chapter,first of all we configure completely same test environment,the necessary hardware performance is tested for physical host and virtual machines of cloud platform resource pool,then their performance differences are compared and analyzed.In the second chapter,the 2D FEM model of 4 pole 36 slots PMSM with built-in radial is established,simulation analysis of multi field coupling such as electric field,magnetic field and so on is carried out.Then,in the third chapter script file of FEM model is regarded as intelligent algorithm subroutine,realizing data between FEM model and intelligent algorithm main program real-time transfer and update;genetic algorithm,particle swarm algorithm and BP neural network are used to respectively research the same design problem of PMSM.Finally,the fourth chapter research on cloud scheduling for intelligent motor design based on the third chapter,cloud scheduling research contain complex single task and multiple simple tasks,intelligent motor design based on cloud platform is achieved and implemented.The performance test show that the performance of virtual machine compare with physical host has a slight loss,but the loss can be ignored compared with the cloud computing resource bring to computing performance promotion.Through genetic algorithm,particle swarm optimization to obtain optimal solution of PMSM design question which are consistent with the theoretical analysis results;these two algorithms omit tedious repetitive manual updating calculation of FEM model,the motor design time is shorten and accuracy is improved.Through the BP neural network learning,verification and training sample data achieve fitting function relationship,which can forecast motor intelligently.BP neural network prediction PMSM case verifies that dividing the single task into multiple subtasks is scheduled and assigned in multiple virtual machines dynamically,efficiency motor intelligent prediction is improve significantly by parallel computing these subtasks;In addition,three independent motor intelligent design tasks are allocated to three virtual machines in the cloud resource pool,The parallel calculation results show that the motor intelligent design time is significantly shortened.