背靠背三电平PWM变换器矢量控制系统研究

发布时间：2018-09-11 16:02

【摘要】：本文针对交流异步电机驱动,对背靠背三电平脉冲宽度调制(Pulse Width Modulation, PWM)变换器矢量控制系统进行研究。对三种常用的快速SVPWM算法进行分析与比较,在此基础上提出了一种计算量小、易于数字实现的任意多电平SVPWM通用算法。并将所提的通用SVPWM算法在现场可编程门阵列(Field Programmable Gate Array,FPGA)上设计实现。针对单侧二极管中点箝位型三电平PWM变换器,提出一种新的混合SVPWM的电容中点电位平衡控制算法。根据每个扇区不同的小矢量作用,并结合电容电压偏差量及三相电流的极性来设置不同的小矢量作用时间调节因子,然后根据三相电流的变化情况使调制在常规的SVPWM和基于虚拟矢量的SVPWM之间进行切换,以实现对中点电位进行分扇区的精细控制。针对背靠背的二极管中点箝位型三电平PWM变换器电容中点电压平衡控制问题,提出一种基于双侧信号的中点电位平衡综合预测控制算法。该算法对双侧的三相电流和直流侧电容电压进行采样和预测,进而给出一个能够表征电容中点电压预期控制指标的品质函数。通过递推运算使品质函数取最小值,从而获得下一个调制周期最佳中点电压控制的冗余矢量时间分配因子。在已有的三电平PWM整流器DPC的基础上,提出了在一个开关周期内具有双非零电压矢量输出的三电平PWM整流器DPC算法。该算法通过对下一时刻的有功和无功进行预测,每次选择对有功和无功调节作用相反的两个电压矢量进行输出,并通过一个分配因子对两个电压矢量的作用时间进行分配,从而实现了对功率的精细调整。基于递推最小二乘法,对三电平变频器的电机参数静态辨识进行建模和仿真,给出了中点电位不平衡与平衡的情况下,电机参数静态辨识的仿真结果。提出一种交流异步电机电机参数动态在线辨识方法,通过在负载恒定时控制电机速度使其保持恒定来消除转子磁链耦合项对参数动态在线辨识的影响,从而能够利用递推最小二乘法对电机参数进行在线计算,以达到电机参数动态在线辨识的实时性和高精确度要求。其中电机恒定速度的保持又是基于对转子磁链和转矩电流的恒定控制来实现。提出了一种能够使系统保持固定的开关频率和采样频率且随机效果不依赖于零矢量的三电平随机SVPWM方法。通过随机调整矢量在每个三角载波的上升段和下降段的时间分配来实现随机SVPWM,并设计一种三重随机化M序列来增强其随机效果。该方法可使输出的线电压和相电流频谱在较宽的频带范围内均匀分布,在整数倍开关频率处的谐波幅值也大大降低。
[Abstract]:In this paper, the vector control system of back-to-back three-level pulse width modulation (PWM) converter driven by AC asynchronous motor is studied.
Based on the analysis and comparison of three commonly used fast SVPWM algorithms, a general algorithm for arbitrary multilevel SVPWM is proposed, which is easy to realize digitally with less computation. The proposed algorithm is designed and implemented on Field Programmable Gate Array (FPGA).
A novel neutral-point potential balance control algorithm for single-side diode neutral-point-clamped three-level PWM converter based on hybrid SVPWM is proposed. According to the different small vector action in each sector, different small vector action time adjustment factors are set according to the capacitor voltage deviation and the polarity of three-phase current. The variation of the current makes the modulation switch between the conventional SVPWM and the SVPWM based on the virtual vector to realize the fine sector control of the midpoint potential.
Aiming at the problem of neutral-point voltage balance control of capacitor in back-to-back diode neutral-point clamped three-level PWM converter, an integrated predictive control algorithm for neutral-point potential balance based on bilateral signals is proposed. The quality function of the voltage expectation control index is minimized by recursive operation, and the redundant vector time allocation factor of the best midpoint voltage control in the next modulation period is obtained.
Based on the existing DPC of three-level PWM rectifier, a DPC algorithm of three-level PWM rectifier with dual non-zero voltage vector output in one switching cycle is proposed. By predicting the active and reactive power at the next moment, the DPC algorithm selects two voltage vectors with opposite active and reactive power regulation at each time and passes through. A distribution factor is used to distribute the action time of the two voltage vectors, thus realizing fine power adjustment.
Based on the recursive least squares method, the static identification of motor parameters of three-level inverter is modeled and simulated. The simulation results of static identification of motor parameters under unbalanced and balanced neutral-point potentials are given. A dynamic on-line identification method of motor parameters of AC asynchronous motor is proposed. The motor speed is controlled by constant load. By keeping the rotor flux coupling constant, the influence of rotor flux coupling term on on-line parameter identification can be eliminated, and the motor parameters can be calculated on-line using the recursive least squares method to meet the real-time and high accuracy requirements of on-line parameter identification. The constant control of the torque current is achieved.
A three-level stochastic SVPWM method is proposed, in which the switching frequency and sampling frequency can be kept constant and the random effect does not depend on zero vectors. The stochastic SVPWM is realized by randomly adjusting the time allocation between the rising and falling phases of each triangular carrier. A triple randomized M sequence is designed to enhance the randomness of the system. This method can make the output line voltage and phase current spectrum distribute uniformly in a wide frequency band and greatly reduce the harmonic amplitude at integer multiple switching frequency.
【学位授予单位】：中南大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM46

【参考文献】