【作者】 李秋红；

【导师】 孙健国；

【作者基本信息】 南京航空航天大学 ， 航空宇航推进理论与工程， 2011， 博士

【摘要】 随着对航空发动机性能要求越来越高,航空发动机控制已向多变量、智能、鲁棒控制等先进控制方式发展。而鲁棒控制能克服模型的不确定性,在非设计点也具有令人满意的性能,在航空发动机控制中具有良好的应用前景。将鲁棒控制和智能控制算法相结合,优化控制器参数,可以进一步减少回路之间的耦合,提高系统性能。本文围绕航空发动机智能多变量鲁棒控制这一主题,研究了航空发动机建模,鲁棒控制器智能参数优化方法,智能推力估计和控制方法,最后通过控制结构分析设计了航空发动机四变量分块解耦控制系统,并进行了数字仿真和半物理仿真验证。论文首先建立了某型航空动机非线性部件级模型,为论文的研究工作提供了仿真平台。提出了基于智能优化的航空发动机状态变量模型建立方法,避免了传统拟合法公式推导的复杂性和对系统模态的要求,并将粒子群优化算法与最小二乘优化算法相结合,实现了航空发动机四输入系统状态变量模型的建立,提高了高阶系统建模的精度。在多变量鲁棒控制系统设计中,提出了基于参考模型的智能参数优化方法,用于航空发动机转速控制系统,避免了基于性能指标加权优化中,性能指标与其权重没有明确对应关系的缺点,使系统具有明确的性能要求。设计了基于小波神经网络的航空发动机PID转速控制系统和解耦ALQR转速控制系统,提高了系统性能。提出了基于控制结构设计的模型逆控制方法,用于航空发动机直接推力控制系统。研究了推力估计的神经网络、支持向量机算法、Kalman跟踪滤波方法,并提出了基于控制器的跟踪滤波推力估计方法。采用神经网络逆模型和线性逆模型,结合PI控制算法,进行了推力控制研究。仿真结果表明推力估计和控制效果良好。论文最后研究了航空发动机四输入鲁棒控制系统的控制结构设计方法,分析了四变量航空发动机控制系统中控制变量和被控制变量之间的单调性,提出了基于LQ/H∞的分块解耦控制算法,并通过数字仿真和半物理仿真验证了算法的有效性。更多还原

【Abstract】 Aeroengine are becoming increasingly complex, with more control variables, to meet future demands on performance. To utilize the potential of the engines, it is also necessary to use more advanced control cenceptes such as multivariable, intelligent, robust control. Robust control can get good performance both at design point and off-design point. It has good application perspective at aeroengine control domain. Using the intelligent algorithm to optimize the paremeters of robust controller, can further eliminate the coupling between the control loops and enhance the system performance. The multivariable robust control of aeroengine is the topic of this paper. It includes aeroengine modelling, robust controller parameters optimization, intelligent thrust estimation and control. The four-input controller for aeroengine is designed and tested based on the control structure design at last.The nonlinear component level model of aeroengine is researched in the paper. It provides the simulation platform. The state variable modeling method of aeroengine based on the intelligent optimization is proposed. It can get the linear model without complicated formula deduction and has no limitation on the system modes. The particle swarm optimization and the least square optimization are combined and used to build the aeroengine four-input state variable model. The precision of the model is improved by the combination optimization. In the design of multivariable robust control system, the intelligent parameters optimization method is proposed and used in aeroengine rotor speed control system. It can make the performance request clearly and avoid the shortcoming of performance indices weighted sum method that the weights have no clearly corresponding relationship with the system response. The aeroengine rotor speed control system is designed based on the wevelet network PID control and decoupling ALQR control. The inverse model control method is proposed based on the control structure design and used in aeroengine thrust control system. The neural network, support vector machine and Kalman filter are researched in thrust estimation. The tracking filter thrust estimate method based on controller is proposed. The neural network inverse model and the linear inverse model are adopted to control the thrust combined with the PI control algorithm. The simulation results show that the thrust estimate and control results are satisfied.At last, the control structure design of aeroengine four-input robust control system is researched. The monotony of the input variables to the output variables of the enigne is analysed. The block decoupling control method base on LQ/H∞is proposed and used to design the four-variable engine control system. The method is validated by digital and semi-physical testing.更多还原