| In recent years,the application of permanent magnet synchronous motor(PMSM)in industrial servo control system becomes more extensive,and the requirements for the control performance of the controller of the system are gradually improved.Model predictive control(MPC)has great development potential in the industrial servo control realm because of its advantages such as excellent dynamic performance,fewer tuning parameters and control structure close to the model of the controlled object.At present,the MPC cannot fully be adopted in the industrial servo control realm with multiple control objectives and limited hardware processor performance.Based on the PMSM rotor position control system combining with the direct speed model predictive control technology is selected as the research object,an extra voltage vector table and hybrid MPC structure are proposed to solve the problem of limitations of the control objective number of the weighting factor eliminating structures.The number of control objectives can be increased without any weighting factors and its tuning processes.The constant prediction horizon is improved as a positive discrete-time variable by the two kinds of proposed methods including constant changing step and variable changing step changing prediction horizon methods to solve the problem of larger operating speed and weaken flexibility of nonlinear MPC.The prediction horizon can be auto-adjusted according to the system operating state,and the flexibility and performance can be improved with lower calculation burdens for processor to make the MPC has stronger compatibility for industry servo control systems with different requirements.The optimizations of the above-mentioned two aspects make the control strategy more suits for industrial servo control system,and it has important theoretical significance and research value.The main research contents of this paper are listed as follow:(1)By analyzing the structure and model of PMSM,the basic three-loop vector control,direct torque control strategy and their parameter tuning methods are studied,and the MPC strategies are derived.The stator current model predictive controls are established in the ?? coordinate and dq coordinate,and the direct speed model predictive control is also established in the dq coordinate.The system operates in transient state continuously to compare the performances of above basic control strategies.(2)In order to increase the number of acceptable control objectives of improved structure,the limited reason and solving method of the number of control objectives are highlighted mainly.The numbers of voltage vectors and control objectives for three-phase voltage-source-type bridge inverter are increased to 14 and 6 respectively.Based on the extra voltage vector table,the weighting factors of a cost function with four control objectives can be eliminated by a proposed 4-layer sequential structure model predictive control in this paper,and the control performances with different sequential structure sorts are analyzed and compared.Moreover,a hybrid model predictive control strategy combining with sequential and parallel structures is proposed in this paper,and the number of control objectives is increased to 3 under the conditions of the adopting conventional voltage vector table and keeping better control performance of model predictive control.(3)In order to increase the flexibility dynamic performance of the control strategy,the fixed prediction horizon of the nonlinear model predictive control is changed as a positive discrete variable.Two control algorithms named as increment method and disturbance method are proposed to combine the prediction horizon with the operating states of the system.In the transient process,the prediction horizon is increased to improve the stability and rapidity of the system,while in the steady state process,the prediction horizon is decreased to reduce the prediction error and calculation burden.The redundant prediction horizon is eliminated to reduce calculation burden of the hardware processor effectively during operating process while the control performance is guaranteed.In addition,the prediction error,the prediction horizon and its upper bound of the effective operation range of the long-time model predictive control are discussed and analyzed.(4)On the basis of the variable prediction horizon,the changing step of the prediction horizon is also optimized as a positive discrete variable,which can be calculated together with the prediction horizon within each sampling period and applied to the model predictive control strategy to further improve the dynamic performance and flexibility of the control strategy.The first and second kinds of variable prediction horizon with changing step method,and the method of adjusting angular velocity error are proposed to realize the connection between the changing step of prediction horizon and the operating state of the system.Because the first kind method has largest sensitivities of weighting factors,the second kind method and the method of adjusting angular velocity error are mainly verified by simulation and experiment,and the effectiveness of the two control methods is illustrated by comparing the control performances.Besides,a linear active disturbance resistance based nonlinear model predictive control strategy is proposed to solve the problem of weaken resisting model parameter mismatches,and the disturbances are estimated and compensate by the linear extend state observer to improve dynamics.(5)An experimental platform of permanent magnet synchronous motor control system based on digital signal processor(DSP)and complex programmable logic device(CPLD)are designed,and the structure and main parameter are analyzed and introduced in detail.The platform is used to simulate industry servo control system and the proposed control structures and strategies are verified by the platform to demonstrate the effectiveness and practicability of the proposed control strategies in this paper. |
PDF Full Text Request