| 摘要 |
The invention relates to a method for computer-assisted learning of one or more neural networks based on a time series of data comprising first values at subsequent time steps. Furthermore, the invention relates to a method for computer-assisted prediction of values of a time series based on one or more neural networks. According to the invention, a rate of change of an initial time series of data is calculated, thus resulting in derivative values. Those derivative values are subjected to Empirical Mode Decomposition which is a well-known technique. The modes extracted by this Empirical Mode Decomposition as well as some time-lagged values of the time series are used as inputs for the neural networks whereas the output of those networks are future values of the time series which are to be predicted. Those networks are trained and used for prediction of future values of the time series. The invention provides good prediction results because derivative values are used as inputs and also past values are considered as inputs. The invention may be used for predicting values of any time series and is particularly for predicting the dynamics of a technical system, e.g. for predicting vibrations occurring in a technical system. |
