Abstract:

In order to reduce the disturbance effect on model precision, the accumulated generating operation in grey modeling is introduced into the identification algorithm. Current sampled data play an important role in identification, thus more importance is attached to it by the weighted product, and the history data are weakened. A generalized grey modeling algorithm is proposed which inherits the least square method. This algorithm can use negative signals to modeling, which is different from the traditional grey modeling method. It includes the advantages of the traditional grey modeling method and can satisfy the demand of system identification.