H∞控制理论应用于大滞后对象的研究

【作者】 黄定华；

【导师】 孙炳达；

【作者基本信息】 广东工业大学 ， 控制理论与控制工程， 2001， 硕士

【摘要】 本文首先从控制理论的几个发展阶段入题，阐述了现阶段H_∞控制理论的提出对工程的实际意义及特点并对大滞后对象的控制意义及现状进行了说明。然后扼要概括了H_∞控制理论发展简史及其有关概念和定理，为本文在H_∞控制方面的研究作了理论铺垫。在深入分析了前人的各种大滞后控制方案基础上，对被控对象中的大滞后环节的近似方案进行了分析，给出了H_∞控制理论中的模型匹配问题，然后将模型匹配问题转换成Neharri问题，利用Neharri问题的解法给出了一阶大滞后的H_∞控制器的结构，并在仿真中验证了此控制器的正确性。之后又按照模型匹配问题的精确解法和H_∞标准问题的转化，给出了一种新的模型匹配控制器结构，仿真表明，此控制器可以对典型工业过程进行十分有效的控制，其响应曲线能够很好的逼近给定的模型，从仿真曲线可以看出，比起第三章中列举的两种鲁棒控制方案，我们的控制策略具有更好的性能指标。 其次，对于现在控制界中比较盛行的神经网络控制进行了深入的研究，首先给出了PID控制+滤波器控制器对于大时滞对象控制的有效性的结论，我们采用RBF神经网络预估器对所给对象进行在线辨识，并用在先调整参数的单神经元的PID控制器对于所给对象进行控制，仿真结果表明该方案具有稳态精度高，动态性能好和鲁棒性能强的特点。证明了新型控制器的实用性和正确性，同时，也为大滞后系统的控制提出了一种新的思路，具有一定的理论意义和实用价值。更多还原

【Abstract】 The thesis first gives an overall view of control theory development and elaborates the H Infinite Control Theory significance and it advantage to the control engineering, in which the status and signality of the control problem of large time-delay plant plays an important role. After the introducing of the H Infinite Control Theory history and it correlated concept and theorem, both with the analyzing of the previous work of time-delay system control strategy, we give an approximation of the pure delay unit in the system, and propose the problem of Model Matching in H Infinite Control Theory. Then transform the Model Matching problem to a NeharrI Problem, and present controller architecture with the solution for the Neharri problem, the simulation verified that the controller is effective and stable to be used in a large time-delay system. Afierwards~ on the synthesis of Standard H infinite control problem and the exact Modeling Matching problem, another new control strategy of Model Matching problem was proposed and illustrated with the example of typical industnal process, which is a one order time-delay plant, the result show it has better performance than some other design for the robust controller and the response can be very close to the given model.Secondly, the thesis also takes deep insig)lt into the Neuron Network technology, which is prevailed in the control-engineering field. A conclusion reduced from H infinite theory is presented and on the basis of it, the plant is predicted with the using of RBF neuron network, and an auto-adaptive PH) nerve-cell controller is used to accommodate time-varying plant. The result of experiments shows that this strategy has high quality in both the dynamic performance and stability precision, strong robustness as well. This new control strategy composed of robust control and neuron network presented us with a new idea in control of the large time-delay system, it has both the theoretical and the practical value.更多还原