【作者】 刘军锋；

【导师】 万淑芸；

【作者基本信息】 华中科技大学 ， 控制理论与控制工程， 2006， 博士

【摘要】 本文以感应电机基于定子磁场定向的无速度传感器控制为主要研究内容,针对其中的几个关键问题进行了深入的理论分析和研究,并且构建了DSP(TMS320LF2407A)+CPLD框架的硬件控制平台,在此平台上进行了感应电机按定子磁场定向的无速度传感器矢量控制的调速实验,从实验结果看系统具有较好的调速性能,较宽的调速范围。磁链估算是无速度传感器系统的核心问题。定子磁链估算相对于转子磁链估算需要的电机参数少,因此,定子磁场定向的无速度传感器系统参数鲁棒性强。本文详细阐述了定子磁场定向技术的出发点和原理,深入讨论了磁链和转矩的解耦方法,建立了感应电机按定子磁场定向的无速度传感器系统的控制结构,并理论分析了电感参数对不同参考系下无速度传感器系统的影响。定子电流和电压对定子磁链的估算有着直接影响。本文详细论述了非线性形态滤波器的原理,并将其引入电流信号的滤波,实验结果表明非线性形态滤波器在有效消除电流信号中正负脉冲干扰的同时很好地保留原有信号的特性,取得良好的滤波效果。本文深入分析了死区效应的产生机理、死区对电压基波分量和谐波分量的影响及其对电流波形的影响,提出了一种基于检测功率管实际导通时间的死区补偿和电机电压直接检测方法,利用检测的死区补偿时间进行死区补偿可以有效地提高波形质量,抑制谐波,同时检测到的电机相电压也具有较高的精度。在定子磁链观测方面,纯积分器很容易导致输出信号饱和与偏移,一阶低通滤波器可以在一定程度上抑制饱和与偏移,但不可避免地引入了幅值和相位误差。本文深入研究了带反馈补偿积分器的工作原理、特性及其相关参数的选取原则,指出阈值固定的双积分器可以有效抑制饱和与直流偏移,并且在低频饱和状态下可以完全补偿幅值和相位误差。将阈值固定的双积分器应用于实际系统取得良好的定子磁链观测效果。在仿真条件逼近实际工况的仿真环境下进行系统计算机仿真可以为研究工作提供具有很强的实际指导意义。本文在具有半物理仿真特性的SaberDesigner软件下,建立了感应电动机基于磁场定向的无速度传感器调速系统仿真平台,在此平台上对按转子磁场定向和定子磁场的无速度传感器系统进行了深入研究和对比,并且通过仿真实验研究了电机参数对上述两种磁场定向无速度传感器系统的影响。理论分析和仿真实验都表明无速度传感器系统严重依赖于电机参数的准确性。本文在借鉴其他学者研究成果的基础上,提出了一种改进的感应电机参数的离线检测方法。该方法结合SVPWM调制方式,并引入了电流调节器,加快了参数辨识过程,增强了系统的安全性,利用离散快速傅立叶变换(DFFT)计算电流电压的有效值和功率因数,增强了系统的抗干扰能力。上述方法充分利用现有的软、硬件条件,完全由系统自动完成,无需人工连线,非常适合于无速度传感器系统,实验结果具有良好的一致性和较高的精度。感应电机参数会随运行条件和环境的改变而变化,按定子磁场定向的无速度传感器系统中,定子磁链的估算和定子电阻密切相关。本文提出了一种定子电阻闭环在线估计方法。该方法原理简单、实现方便,仿真结果表明该方法可以很好地跟踪定子电阻的变化,改善系统的性能。采用动态转速估算法,在负载情况下,转子参数和电感参数会影响转差频率的估算。本文将PI自适应转速估算方法引入定子磁场定向的无速度传感器系统,避免了系统在负载情况下,估算转速受转子参数和电感参数影响的问题。从仿真结果看,采用这种估算策略的定子磁场定向无速度传感器系统具有较强的参数鲁棒性。更多还原

【Abstract】 This dissertation focused on the speed sensorless vector control of induction motor (IM) based on stator flux orientation (SFO). The profound theoretical analysis and study have been made against some critical questions. In this dissertation, the hardware platform with the structure of DSP(TMS320LF2407A)+CPLD has been constructed, and the speed sensorless AC variable speed drive experiments of IM based on stator flux orientation have been performed. The preferable variable speed performance and wider speed range of the system have been proved by experiments.The flux estimation is the most important question in speed sensorless control. Much less parameters in stator flux estimation are required than in rotor flux estimation. So the speed sensorless vector control of IM based on SFO is more robustness to parameters than it based on rotor flux orientation (RFO). In this dissertation, the start point and principle of the SFO have been represented in detail, the decoupling methods between flux and torque have been profound discussed, the control structure of the speed sensorless vector control based on SFO has been constructed, and the influences on the system performance by the inductances are also analyzed theoretically.The stator flux estimation is effected directly by stator current and voltage. In this dissertation, the theory of nonlinear morphological filter (NML) has been represented, and the NML is introduced into current filtering. It can be seen that the positive or negative plus disturbance is effectively eliminated in current signal and the primary characteristic of current signal is well preserved from experiment results. The production principle of dead-time, the effect of dead-time on fundamental voltage, harmonic voltage and current wave are discussed. A new method of dead-time compensation and phase-voltage measurement based on detecting the actual turn-on time of power tubes is presented in this dissertation. The harmonic is inhibited and the quality of wave is improved by using detected dead-time to compensation. At the same time, the measurement phase-voltage is high precision.In stator flux estimation, the saturation and DC drift will be easily produced by the pure integrator. The previous questions can be silenced by the one-order low pass filter (LPF), but the errors in magnitude and phase angle will be introduced inevitably. In this dissertation, the profound research on the theory, characteristic of the integrator with feedback compensation and the principle of parameter selection has been made. The limiting level fixed dual integrator can silence the saturation and DC drift questions, and can compensate completely the errors in magnitude and phase angle under the low frequency and saturation condition. The well stator flux observation result has been achieved by applying the limiting level fixed dual integrator to the stator flux observation.The computer simulation under the environment which approaches the actual work condition can supply the powerful guidance for research work. In this dissertation, the speed sensorless variable speed drive based on FO simulation platform of IM has been constructed on SaberDesigner which is a semi-physical simulation software. On the platform, the profound investigation and comparison between the speed sensorless system based on RFO and SFO has been performed, and the influences of IM parameter on the previous speed sensorless system has also been investigated.The parameter dependence of speed sensorless system can be indicated by means of theoretical analysis and simulation study. In this dissertation, based on the other scholars’research production, a modified off-line identification method of IM parameters has been presented. In this method, the SVPWM has been used. The identification procedure has been accelerated and the safety of system has been enhanced by applying the current regulator. The system anti-interference ability has been improved by utilizing the discrete fast fourier transform (DFFT) to calculate the effective value of voltage, current and the power factor. The existing conditions are utilized, the identification can be completed automatically, and no manual connection of electric wires action is required in the proposed method. This method is suitable to the speed sensorless scheme. The high uniformity and precision are presented from the experiment results.The parameters of IM will vary with the change of operating condition. The stator flux estimation nearly correlates with stator resistance in the speed sensorless system based on SFO. In this dissertation, a close loop on-line estimating method of stator resistance is presented. This method is easily realized and its theory is simple. The simulation results show that the stator resistance variation can be well traced and the system performance is improved by using this method. The slip frequency estimation will be effected by the rotor parameters and inductances when the dynamic speed estimation has been employed. The PI adaptive speed estimation method is introduced into the speed sensorless system based on SFO in this dissertation. So the influences of rotor parameters and inductances on speed estimation under the load condition have been avoided. It is showed that the speed sensorless system based on SFO used the previous speed estimation method is more robust to parameters from simulation results.更多还原