| 摘要 |
PROBLEM TO BE SOLVED: To effectively, securely, simply, and economically compensate cyclic disturbance in a servo system. SOLUTION: The servo system under the influence of cyclic disturbance is controlled using a neural network that includes two layers (a hidden layer and an output layer). The hidden layer neuron 610 is connected by the pair, one neuron output of which is connected to the input of the other neuron, and neurons of each pair oscillate with each higher harmonic at a basic frequencyω. The outputs of neurons of each pair are sine and cosine of corresponding higher harmonic respectively. The output layer outputs the weighted sum of output of the hidden layer, and is used to control the servo system. In the neural network, the output of the output layer is synchronized so as to match with cyclic disturbance to the servo system, so that it can be used for this compensation. If there are hidden layer neuron units sufficient enough, cyclic disturbance in any dimension can be accurately compensated. COPYRIGHT: (C)2004,JPO
|
