一种新的动态神经网络及其在机器人监督控制中的应用

A New Dynamical Neural Network and Application to Robotic Supervisory Control

针对提高机器人控制性能这一类问题 ,在复合输入动态递归网络的基础上 ,提出一种新的动态递归神经网络结构 ,称为状态延迟输入动态递归神经网络 (StateDelayInputDynamicalRecurrentNeuralNetworks)。该动态网络具有新的拓扑结构及学习规则 ,各权值矩阵的含义更为明确 ,权值的训练过程更为简洁。网络增加了输入输出层前一步的状态信息 ,收敛速度及稳态精度与其它常用网络结构相比均有明显提高。将该网络用于机器人的监督控制系统 ,利用神经网络建立起被控对象的逆模型 ,与传统PD控制器结合 ,进一步确保了控制系统的稳定性 ,有效地提高系统的精度和自适应能力。

A new neural network model named State Delay Input Dynamical Recurrent Neural Network(SDIDRNN)is presented.The model with new topological structure and learning algorithm has explicit significance for weight matrixes.Learning process of weights become more distinct and straightforward.Speed of learning and convergence are improved by inputting the prior state knowledge of input output layer comparing with several popular neural networks,which make it possible to improve the control performance of robotic system.The new network is applied to the system of supervisory dynamical control for robots.By establish inverse model of the controlled object with the new network and combining it with conventional PD controller,the stability or robustness of the system is guaranteed and accuracy together with adaptive abilities is improved effectively.Simulation results prove the efficiency and superiority of the new neural network.