System-level Optimization Design And Dynamic Performance Analysis Of Multi-stage Magnetic Gearbox For Wind Powe

Developing wind power is a major strategic demand of China to achieve “double carbon” goal.As one of the important devices of wind turbine,the high-quality wind power gearbox has become an urgent need for the wind power industry and has great developing prospects.However,the existing wind turbine gearbox has some inherent defects,such as high failure rate and high operation and maintenance cost,which restrict the development of high-power wind turbine.In recent years,magnetic gear(MG)has been widely concerned for its potential advantages such as high reliability and overload protection.Therefore,this paper proposes a multi-stage magnetic gearbox topology for wind turbine based on magnetic transmission and makes a thorough research on its system-level optimization design and dynamic performance.The main content of this paper is as follows:First,the magnetic field modulation behavior of the coaxial magnetic gear(CMG)which is the key component of the magnetic gearbox,is analyzed based on the magnetic field modulation theory.Then the expression of the operating torque of the CMG is deduced by the Maxwell tensor method and energy conservation relationship.Furthermore,a finite element model of magnetic gearbox with rated power of 1.5MW is established,then its electromagnetic performance is analyzed,and the gear ratio of MG at each stage is optimized based on the mass torque density.Second,compared with the conventional motor,the structure of MG is relatively complex and its optimization parameters are numerous.Only using the finite element analysis(FEA)method to obtain solution will consume a lot of computer resources,resulting a long optimization period.To shorten the development cycle,this paper proposes a single objective optimization strategy based on central composite design,quadratic regression,and genetic algorithm(CCD-QR-GA).The simulation results show that the optimization cycle by this method is short,but the parameters that can be optimized are relatively few.This method is suitable for the situations that require rapid design but do not require high performance.Third,to solve the above problem that the CCD-QR-GA optimization strategy can optimize relatively few parameters,an optimization strategy based on Latin hypercube sampling,BP neural network and multi-objective genetic algorithm(LHS-BPNN-NSGA-Ⅱ)is proposed.Then,the axial length is scanned based on FEA method to determine the final value of the axial length that meets the design requirements.The simulation results show that although this method takes a little longer time than CCD-QR-GA,it can optimize more parameters and the optimization results are more superior.And it is convenient for designers to choose different design schemes according to different design requirements.Thus,this method is more suitable for the optimization of high-performance machine system.Finally,the dynamics of the magnetic gearbox is studied.The torque of magnetic gearbox is transmitted between the rotors through the magnetic field.Therefore,when exerted by a random time-varying shock of wind direction and wind speed,its dynamic response becomes particularly important.Thus,the dynamic behaviors of the MG at lowspeed stage(LSMG)and the two-stage magnetic gearbox are studied respectively.For the LSMG,the first and second-order asymptotic solutions of load angle and magnetic torque are obtained by using the combined Newton harmonic balance(NHB)method.Compared with the Runge-Kutta(R-K)method,the first asymptotic solution can meet the engineering requirements.Then,the influence of physical parameters on system dynamic performance is analyzed.For the two-stage magnetic gearbox,the motion equation based on the Taylor expansion of the load angle is first solved,and then the accuracy of the proposed model is evaluated.In the end,the influence of each physical quantity on the dynamic performance of the magnetic gearbox under different operating conditions is analyzed.The analysis results show that the proposed two analysis methods do not need to solve complex differential equations,so the solution efficiency is high.Meanwhile,the feasibility of applying magnetic gearboxes to wind power systems is verified.