论文部分内容阅读
本文介绍了一种用递阶优化的计算结构算法解决非线性、线性离散时间系统的优化和参数辨识问题。该方法的优点在于把系统优化和辨识问题有机地结合起来,达到了总体最优化克服了以往先辨识、后优化两段分离的缺点,这是因为若分为两步走,势必先要在某时间段内对系统进行辨识,在下段时间内依据辨识的模型结构,参数进行优化,这对时变系统来讲很难达到总体最优。因为优化的精度依赖于模型精度、最大和模型精度相同,而在一段时间内用一个不变的模型代替时变模型,显然得不出最优;其次该方法不仅适用于解小系统问题也可解大的、复杂的系统优化问题;再者,在计算机实时控制系统中,被控制对象的数学模型是离散型的,因此,讨论离散系统优化和辨识问题有很大的意义。最后本文讨论了在线控制问题,收敛条件以及θ值的收敛上限和收敛下限。
In this paper, we use a hierarchical optimization computational algorithm to solve the optimization and parameter identification of nonlinear and linear discrete-time systems. The advantage of this method lies in the organic combination of system optimization and identification and the overall optimization to overcome the shortcomings of prior identification and post-optimization separation. This is because if it is divided into two steps, During the period of time, the system is identified, and in the next period of time, it is optimized according to the identified model structure and parameters, which is hard to achieve the overall optimum for the time-varying system. Because the accuracy of the optimization depends on the accuracy of the model and the maximum is the same as the accuracy of the model, it is obviously not optimal to replace the time-varying model with a constant model over a period of time. Secondly, the method is not only suitable for solving small system problems In addition, in the computer real-time control system, the mathematical model of the controlled object is discrete. Therefore, it is of great significance to discuss the optimization and identification of discrete systems. Finally, this paper discusses the online control problem, the convergence condition and the convergence upper limit and convergence lower limit of θ.