整流器论文中英文资料外文翻译文献Word文档格式.docx

1、Transforming （1） into the reference frame using （4） （5）where p is a differential operator and .Expressing （5） in a vector notation （6）where, （7）Taking a transformation of （2） by using （4） （8） （9）Expressing （6） and （8） in a discrete domain, by approximating the derivative term in （6） by a forward dif

2、ference 9, respectively, （10） （11） Where T is the sampling period. Fig. 2. Overall control block diagram.B System ControlThe PI controllers are used to regulate the dc output voltage and the ac input current. For decoupling current control, the cross-coupling terms are compensated in a feed forward-

3、typeand the source voltage is also compensated as a disturbance. For transient responses without overshoot, the anti-windup technique is employed 10. The overall control block diagram eliminating the source voltage and line current sensors is shown in Fig. 2. The estimation algorithm of source volta

4、ges and line currents is described in the following sections.2 PREDICTIVE CURRENT ESTIMATIONThe currents of and can not be calculated instantly since the calculation time of the DSP is required. To eliminate the delay effect, a state observer can be used. In addition, the state observer provides the

5、 filtering effects for the estimated variable.Expressing （5） in a state-space form, （12） （13）where, And y is the output. Transforming （12） and （13） into a discrete domain, respectively, （14） （15）Then, the observer equation adding an error correction term to is given by （16）Where K is the observer ga

6、in matrix and “ ” means the estimated quantity, and is the state variable estimated ahead one sampling period. Subtracting （15） from （16）, the error dynamic equation of the observer is expressed as （17）where . Here, it is assumed that the model parameters match well with the real ones. Fig. 3 shows

7、the block diagram of the closed-loop state observer.The state variable error depends only on the initial error and is independent of the input. For （17） to converge to the zero state, the roots of the characteristic equation of （17） should be located within the unit circle. Fig. 3. Closed-loop state

8、 observer. Fig. 4. Short pulse region. 4 EXPERIMENTS AND DISCUSSIONSA. System Hardware ConfigurationFig. 5 shows the system hardware configuration. The source voltage is a three-phase, 110 V.The input resistance and inductance are 0.06and 3.3 mH, respectively. The dc link capacitance is 2350F and th

9、e switching frequency of the PWM rectifier is 3.5 kHz.Fig. 5. System hardware configuration.Fig. 6. Dc link currents and corresponding phase currents （in sector V ）.The TMS320C31 DSP chip operating at 33.3 MHz is used as a main processor and two 12-b A/D converters are used. One of them is dedicated

10、 for detecting the dc link current and the other is used for measuring the dc output voltage and the source voltages and currents, where ac side quantities are just measured for performance comparison.One of two internal timers in the DSP is employed to decide the PWM control period and the other is

11、 used to determine the dc link current interrupt. Considering the rectifier blanking time of 3.5 s, A/D conversion time of 2.6 s, and the other signal delay time, the minimum pulse width is set to 10 s.A.Experimental Results Fig. 6 shows measured dc link currents and phase currents. In case of secto

12、r V of the space vector diagram, the dc link current corresponds to for the switching state of and for that of . Fig. 7（a） shows the raw dc link current before filtering. It has a lot of ringing components due to the resonance of the leakage inductance and the snubber capacitor. When the dc current

13、is sampled at the end point of the active voltage vectors as shown in the figure, the measuring error can be reduced. Fig. 7. Sampling of dc link currents. Fig. 8. Estimated source voltage and current at starting. To reduce this error further, the low pass filter should be employed, of which result

14、is shown in Fig. 7（b）. The cut-off frequency of the Butterworths second-order filter is 112 kHz and its delay time is about 2 sec. Since the ringing frequency is 258 kHz and the switching frequency is 3.5 kHz, the filtered signal without significant delay is acquired.Fig. 8 shows the estimated sourc

15、e voltage and current at starting. With the proposed initial estimation strategy, the starting operation is well performed. Fig. 9 shows the phaseangle, magnitude, and waveform of the estimated source voltage, which coincide well with measured ones.Fig. 10 shows the source voltage and current wavefo

16、rm at unity power factor. Figs. With the estimated quantities for the feedback control, the control performance is satisfactory. The dc voltage variation for load changes will be remarkably decreased if a feedforward control for theload current is added, which is possible without additional cur-rent

17、 sensor when the PWM rectifier is combined with the PWM inverter for ac motor drives.Fig. 9. Estimated source voltage in steady state. （a） phase angle （b）magnitude （c） waveform. Fig. 10. Source voltage and current waveforms. （a）estimated （b） measured.4 CONCLUSIONSThis paper proposed a novel control

18、scheme of the PWM rectifiers without employing any ac input voltage and current sensors and with using dc voltage and current sensors only. Reducing the number of the sensors used decreases the system cost as well as improves the system reliability. The phase angle and the magnitude of the source vo

19、ltage have been estimated by controlling the deviation between the rectifier current and its model current to be zero. For line current reconstruction, switching states and measured dc link currents were used. To eliminate the effect of the calculation time delay of the microprocessor, the predictiv

20、e state observer was used. It was shown that the estimation algorithm is robust to the parameter variation. The whole algorithm has been implemented for a proto-type 1.5 kVAPWM rectifier system controlled by TMS320C31 DSP. The experimental results have verified that the proposed ac sensor eliminatio

21、n method is feasible.无交流电动势、电流传感器的三相PWM整流器控制1 三相PWM 整流器A 系统模型图一所示为三相PWM整流器的主电路，电压等式给出如下： （1） 图1 无交流传感器三相PWM整流器其中e，i和v分别是源电压，线电流和整流器的输入电压，R和L分别是输入电阻和输入电感。当已知线电压峰值E，角频率和初始相位角时，假定三相系统是平衡的，则源相位电压可以表达为 （2）其中一种基于估计相位角的变换矩阵，将三相变量变换成一个同步的，坐标系，这个矩阵是 （4）将（1）式变为坐标系使用式（4） （5）其中p是一个微分算子且将（5）式写成矢量形式 （6）用式（4）对（2）式

22、进行变换 （8） （9） 通过前向差分来接近微分的限幅，分别将（6）式和（8）式用离散域表示 （10） （11）其中，T是采样周期图2 总的控制模块图B 系统控制PI控制器是用来调节直流输出电压和交流输入电流的。对于解耦电流控制，交叉耦合项用前馈式补偿，同时，源电压作为扰动的补偿。对于没有过调的暂态响应，引入anti-windup技术。消除源电压和线电流传感器的总的控制模块图如图2所示。源电压和线电流的估计算法在以后的章节中介绍。2预测电流估计由于DSP存在计算时间，所以和不能立即计算。为了消除延迟的影响，可以使用状态监测器。另外，状态监测器可以对估计变量起到滤波作用。将式（5）用状态空间形式

23、表达为 Y是输出。分别将式（12）和式（13）分别变换成离散领域 则加入了误差调整的监测器等式为其中，k是监测器增益矩阵，“ ”是指估计量，是提前一个采样周期估计的状态变量。用式（15）和减去式（16），监测器的动态误差等式表述为 （17）这里，假设模型参数与真实系统吻合的很好。图7所示是闭环状态监测器的模块图。状态变量误差仅取决于初始误差，与输入无关。为了使式（17）趋于零状态，典型等式（17）的根应该限制在单位圆内。图3 闭环状态监测器 图4短脉冲区域3实验与讨论A系统硬件构造 图5 系统硬件结构 图6 直流电流和相应相电流 （扇区5 ）.图5所示是系统的硬件结构图。源电压是三相110V。

24、输入电阻和电感分别为0.06和3.3mH。直流侧电容为2350F，PWM整流器的开关切换频率为3.5KHZ.使用TMS320C31 DSP芯片设定在33.3MHZ作为主处理器，同时用到两个12位的A/D转换器：一个用来检测直流侧电流，另一个用来检测直流侧输出电压、源电压和电流。其中直流侧数量只是为了性能比较而测量的。DSP内部的两个时钟一个是用来决定PWM波的控制周期，另一个是用来决定直流侧电流中断。考虑到整流器空白时间3.5S，A/D转换时间2.6S和其他信号延迟时间，最小脉冲宽度设定为10S.C、实验结果图6所示是测得的直流侧电流和相电流。假设空间矢量图的扇区V，直流侧电流对应于。图7（a

25、）所示是滤波之前未经处理的直流侧电流。因漏电感和缓冲电容的共振，会产生噪声成分。如图中所示，当采样动态电压矢量末端的直流电流时，测量误差可以减小。 图7 直流侧电流采样 图8 开始时的估计源电压和电流 为了进一步减少误差，可以使用低通滤波器，结果如图7（b）所示。Butterworth的第二顺序滤波器的截止频率是112KHZ，开关切换频率为3.5KHZ，所以可以得到没有显著延迟的滤波信号。图8所示是开始时估计源电压和电流。使用提出的初始估计策略，开始操作效果很好。图9所示是估计源电压的相位角、数值和波形。它们和测量的结果十分吻合。图10所是在单位功率因数时源电压和电流波形。当PWM整流器与逆变

26、器相连时，在没有额外电流传感器的情况下对交流汽车驾驶来说是可行的。 图9稳态时的估计源电压. （a）相位角（b）数值 （c）波形 图10 源电压和电流波形（a）估计值 （b）测量值4结论这篇文章提出了一种PWM整流器新颖的控制方法。这种方法没有使用任何交流输入电压和电流传感器，而仅仅使用直流电压和电流传感器。减少传感器数量可以减少系统费用的同时就提高系统的稳定性。通过控制整流器的电流和它的模型电流的偏差为零，可以估计相位角和源电压的数值。对于线电流重建，使用开关状态和直流侧电流测量。为了消除因微处理器计算时间所带来的延迟影响，使用预测状态监测器。可以看出，估计算法对参数变化是健全的。整个算法已经通过TMS320C31 DSP作为控制器的1.5kVAPWM整流器原型执行。实验结果证明已经证明了提出的消除交流传感器方案的可行性。