| 摘要: |
| 针对单力臂机械手的控制，提出了一种基于RBF神经网络模型的控制方法；RBF神经网络即径向基函数，它本身是具有单隐层的三层前馈网络，从输入空间到输出空间的映射呈现非线性，但是从隐含层空间到输出空间的映射却呈现线性，又由于RBF网络它采用高斯基函数作为作用函数，它的输出与部分调参数有关，在输入空间的有限范围内不为零，是一种局部逼近的神经网络方法，所以采用RBF网络能够很快地进行学习并且避免出现局部极小问题，能够满足控制的实时性要求；通过建立相关模型的Simulink控制系统仿真以及M语言的离散数字化仿真发现，采用RBF神经网络模型方法进行控制，其控制的精确度和实时性很高，控制效果很好。 |
| 关键词: RBF神经网路 Simulink 控制系统 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| Control of One arm Robot Based on RBF Neural Network Model |
|
DOU Qin-qin
|
|
Artificial Intelligence Institute,Maanshan College，Anhui Maanshan 243100,China
|
| Abstract: |
| This paper proposed a control method for the single manipulator of control based on RBF neural network model.RBF neural network，the radial basis function，is a three layer feedforward network with single hidden layer.Its mapping is nonlinear from input space to the output space，but from the hidden layer to the output space，its mapping presents linear space，and because it used Gaussian basis function as role function of the RBF network，its output is associated with some adjustable parameters，and it is not zero in the limited range of input space，so it is a kind of local approximation of the neural network method，it can quickly learn and avoid local minimum problem in the RBF network,and it satisfies the requirement of the control in the real time.It is found that the RBF neural network model can improve the control accuracy，robustness and adaptability significantly through the establishment of Simulink control system simulation of the relevant models and the discrete digital simulation of M language. Its Control effect is good. |
| Key words: RBF neural network simulink control system |
