It is possible to obtain, in areas of active induced mine seismicity, within a short time interval, a sufficiently large data set of seismic and other physical parameters. These data can be used as input for multi-channel statistical extrapolation methods, which appear most suitable for the very reason that all the effects and parameters which enter in the process and affect the preparation and inception of a seismic event are not exactly known until now. Seismic and mining parameters obtained from January 1, 1994 to April 3, 1997 were used as input data for the 'application of linear extrapolation (Wiener filter) and non-linear methods (neural network). This proved that the seismic data completed by parameters of mining activities do not contain sufficient information on the development of seismic parameters within a future time interval and the prediction, from the statistical viewpoint, is on the average only by about 30% better than that obtained with the use of a primitive predictor, such as the last or average value.
展开▼
