New Methods To Improve Control Performance Of The Current Regulator And The Modulation Scheme For Permanent Magnet Synchronous Machines

High efficiency and high power density permanent magnet synchronous machine(PMSM)is widely used in the traction and servo applications.The motor control fundamental inner loop(referred as fundamental inner loop)is the basic framework of PMSM control,which includes three parts: current regulator,PWM and current feedback.First,the current regulator determines the current response and harmonic suppression capability,the most used current regulator is proportional-integral(PI)control,which futures simple structure and high robustness,but the theoretical bandwidth and dynamic response is limited.Second,PWM implements the voltage command of the current regulator,and the most used space vector PWM(SVPWM)suffers from high AC side current ripple at high modulation index,and the ripple leads to the increase of the eddy loss in the iron core and the temperature rise of the rotor permanent magnet.Last,in practical applications,the non-linearity and time-varying effect of the current sensor and conditioning circuit cause current feedback sampling errors,and that error leads to the first and second order harmonics in the synchronous d-q axis current and torque.This thesis studies the above three basic problems in the fundamental inner loop separately.Firstly,compare to PI,deadbeat-predictive control futures higher performance,but it's quite sensitive to motor parameters and high frequency current noises.And the traditional observer based steady-state compensation method is quite complex and the convergence is slow,so that the deadbeat-predictive cannot replace PI control in practical applications.Therefore,new direct error compensation and high frequency current noises suppression algorithms are proposed in this thesis,the new algorithm is tested in interior permanent magnet machine(IPMSM)test rig at 90 k W output power,moreover the execution time of the proposed compensation term is only0.2 micro-second.Secondly,a novel online current ripple minimization modulation is proposed and experimental verified in this thesis.Under the same switching frequency,it is able to reduce the current ripple RMS around 46% at the maximum linear modulation index,and reduce the power losses of the semi-conductors at most used power factor range,the maximum reduction is around 15%.Finally,the analytical relationship between d-q axis current oscillations and current sampling errors is revealed in this thesis,and an accurate new current sampling error analysis model is proposed.Then the online compensation is achieved based on that model,so that the first and second order harmonics can be suppressed dramatically.And the convergence time of the proposed method is much faster than available methodsThe researches in second chapter is the current regulator in the fundamental inner loop.To tackle the sensitivity to the model parameters of the deadbeat predictive control,the analytical relationship between parameters estimation error and the d-q axis steady-state tracking error is derived for the surface mount PMSM(SMPMSM)first,then a new direct error voltage compensation based closed form deadbeat-predictive control is proposed.Furthermore,the noise suppression scheme is added to alleviate the noises in practical system.Then,the closed loop transfer function of the compensated system is established,and it's proven theoretically that the proposed method is able to compensate the estimation errors of the inductance,stator resistance and rotor flux linkage.Moreover,the two-sample deadbeat response can be maintained with the proposed method.At last,the proposed method is extended to the IPMSM,and tested in the high power test rig at output 90 k W power.The researches in the third chapter is the PWM in the fundamental inner loop.To reduce the current ripple at high modulation index without increasing switching frequency and losses,a novel online current ripple minimization modulation is proposed.After detailed analysis the current ripple and switching loss of two categories hybrid PWMs,the three main drawbacks of these methods are pointed out,which are the dependence of offline optimizations and look-up tables,the extra switching losses during voltage sequence swapping and the dependence of the CPLD and FPGA.So,a current ripple vector envelope online calculation model is established first,then the ripple of different modulations is compared online,and the optimal sequence is selected in real-time,thus the look-up table is avoided and it can be integrated with vector control seamlessly.Then,the extra switching losses during sequences swapping is avoided in the new online modulation strategy.Lastly,the simple implementation method of the new modulation as used for the DPS is given.In order to compare with available modulations,the SMPMSM test rig is chosen for the experimental verifications,the effectiveness of current ripple reduction and lowest switching losses within mostly used power factor range is proven for the proposed method.The researches in the fourth chapter is the current sampling feedback error of the fundamental inner loop.To tackle the d-q axis current oscillation due to the current sampling error,an accurate sampling error analysis model is proposed firstly in this thesis,and new online compensation strategies are proposed based on the model.Compared to conventional model,the accurate model takes the scaling differences and DC offsets due to sampling errors in three-phase current into consideration,rather than the ideal three-phase sinusoidal currents in the conventional model.Base on the accurate model,the influence of the current sampling errors to the current and speed loop is analyzed separately.Then an online compensation strategies based on three-phase imbalance detection and speed harmonics detection is proposed,and the convergence time is much faster than available methods.Last,experiments based on the SMPMSM and IPMSM test rig verified the correctness and effectiveness of the proposed compensation method.To conclude,this thesis is focused on the three basic problems in the PMSM fundamental inner control loop,which are,the sensitivity to parameters and noises of the deadbeat predictive control,the current ripple in PWM modulations,and the current sampling error.Novel and practical solutions to these problems are provided in this thesis,and the correctness and effectiveness of these proposed methods are validated via simulations and experiments,hence the performance of the fundamental inner loop is improved.Although the new methods proposed in this thesis are based on PMSM,they can also be applied to other three-phase motors and grid connected converters.