进化神经网络在倒立摆控制中的应用  

Application of Evolutionary Neural Network in Controlling Inverted Pendulum

作　　者：谢宗安[1] 张滔[1] 

机构地区：[1]贵州大学电气工程学院,贵州贵阳550003

出　　处：《计算机仿真》2006年第5期297-299,306-307,共5页Computer Simulation

摘　　要：倒立摆作为典型的非线性系统,伴随着多变量、快速运动和绝对不稳定的特征,难于建立精确的数学模型,这就使得对倒立摆的控制变得异常困难和复杂。智能控制理论则是解决此问题的一个有效途径,该文针对倒立摆控制的传统神经网络算法(即BP算法)的缺点,将遗传算法与神经网络结合起来,提出了倒立摆的进化神经网络控制方法。控制器在结构上采用神经网络,利用遗传算法优化神经网络的连接权值。实验研究表明,该控制器不仅具有良好的动态和稳态控制性能,而且对于干扰也具有很强的抑制能力。同时还具备结构简单,易于实现的优点。Inverted pendulum is a classic nonlinear system accompanied by characters of multivariable, rapid move and absolutely unstable, these make it hard to establish accurate arithmetic model. However, intelligence control theory is an effective approach for this ease. Taking account of the defects of traditional training algorithm （ namely BP algorithm） in pendulum control we put forward a method called evolutionary neural network to combine genetic algorithm with neural network. The controller＇s structure was based on neural network , instead of back propagation, genetic algorithm was used to optimize the weights of neural network. The experimental results illustrated that controller has not only good dynamic and static control performance but also strong capability to suppress interference. Furthermore, it has advantages of simple structure and is easy to be implemented.

关 键 词：遗传算法 神经网络 优化 倒立摆 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]