基于有源前端变流器的多电机系统有限状态模型预测控制

发布时间：2018-01-14 22:35

本文关键词：基于有源前端变流器的多电机系统有限状态模型预测控制　出处：《天津大学》2014年博士论文　论文类型：学位论文

【摘要】：采用有源前端（Active Front End，AFE）变流器和共直流母线技术的多电机系统在最近十几年得到的飞速的发展，并将成为未来市场上的主流产品。在多电机系统中，网侧与电机侧的变流器具有不同的控制目标和控制算法，这将增加系统设计的复杂度。而目前，有限状态模型预测控制（Finite Control Set ModelPredictive Control，FCS-MPC）在AFE变流器以及电机驱动器中均得到了广泛的研究，此算法利用对象数学模型预测未来状态，通过价值函数实现多变量优化控制，在未来多电机控制领域有着广阔的应用前景。本文以基于AFE变流器的多电机系统为对象，，针对FCS-MPC算法进行分析与研究，并提出相应的控制策略。针对多电机系统中FCS-MPC算法控制周期短而计算量大的问题，提出了一种算法的简化方法。采用等效变换的方法，将多次预测过程转变为单次虚拟参考电压矢量预测，进而实现了预测过程的简化；同时，在价值函数寻优过程中引入矢量分区法，通过对候选矢量的筛选，排除不合适的电压矢量，减少寻优过程的计算次数。该方法在保持原有控制性能的基础上，可大幅减少算法的执行时间，降低多电机系统对控制器计算性能的要求。在AFE变流器中，算法延迟和信号滤波延迟对FCS-MPC电流控制性能有着负面影响，针对单周期算法延迟和非整数周期算法延迟，设计了对应的延迟补偿算法，提出了基于价值函数的非整数周期延迟补偿策略；针对信号滤波所造成的时间延迟，引入状态观测器理论，结合AFE变流器以及信号滤波器的数学模型，提出了一种滤波延迟观测器，实现信号滤波延迟的观测与补偿。该方法能够在不改变硬件结构的基础上，提高多电机系统中FCS-MPC算法在AFE变流器上的控制效果。针对FCS-MPC控制策略对多电机系统数学模型的依赖度较高的问题，结合扩张状态观测器理论和自适应滤波理论，提出了一种能够对多个变量进行实时观测的混合并行观测策略。将混合并行观测器与FCS-MPC相结合，并对电流预测模型进行重新构造。在网侧AFE变流器中，实现了电阻、电感参数不确定情况下的AFE变流器无电网电压传感器控制；在多电机侧变流器中，实现了反电动势、定子电阻、定子电感的同步观测下的多电机协同控制。所提的FCS-MPC改进算法大幅降低了对多电机系统模型的依赖度，提高了FCS-MPC控制策略在参数不确定情况下的控制性能。
[Abstract]:The multi-motor system with active Front end converter and common DC bus technology has been developed rapidly in the last ten years. In the multi-motor system, the converters on the grid side and the motor side have different control objectives and control algorithms, which will increase the complexity of the system design. Finite Control Set ModelPredictive Control. FCS-MPC) has been widely studied in AFE converter and motor driver. This algorithm uses object mathematical model to predict future state and realizes multivariable optimal control by value function. This paper takes the multi-motor system based on AFE converter as the object, and analyzes and studies the FCS-MPC algorithm. The corresponding control strategy is put forward. In order to solve the problem of short control period and large computational cost of FCS-MPC algorithm in multi-motor system, a simplified algorithm is proposed. The method of equivalent transformation is adopted. The multiple prediction process is transformed into a single virtual reference voltage vector prediction, and the prediction process is simplified. At the same time, the vector partition method is introduced in the process of value function optimization, and the inappropriate voltage vector is excluded by the screening of candidate vectors. On the basis of maintaining the original control performance, this method can greatly reduce the execution time of the algorithm and reduce the requirement of the controller calculation performance for the multi-motor system. In AFE converter, algorithm delay and signal filter delay have a negative impact on FCS-MPC current control performance, for single-period algorithm delay and non-integer period algorithm delay. The corresponding delay compensation algorithm is designed and a non-integer period delay compensation strategy based on value function is proposed. According to the time delay caused by signal filtering, a filter delay observer is proposed by introducing the theory of state observer, combining with the mathematical model of AFE converter and signal filter. This method can improve the control effect of FCS-MPC algorithm on AFE converter without changing the hardware structure. Aiming at the problem that FCS-MPC control strategy is highly dependent on the mathematical model of multi-motor system, the extended state observer theory and adaptive filtering theory are combined. In this paper, a hybrid parallel observation strategy is proposed, which can observe multiple variables in real time. The hybrid parallel observer is combined with FCS-MPC. The current prediction model is reconstructed. In the grid-side AFE converter, the AFE converter with uncertain resistor and inductance parameters is realized without grid voltage sensor control. In the multi-motor side converter, the back EMF and stator resistance are realized. The proposed improved FCS-MPC algorithm greatly reduces the dependence on the multi-motor system model under synchronous observation of stator inductance. The control performance of FCS-MPC control strategy under uncertain parameters is improved.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM46

【参考文献】