Vector prediction model to describe the signals of seismic sensors group

Signals recorded by seismic sensors, formed by the superposition of seismic waves propagating in the ground by multipath paths. Therefore, they obey a Gaussian distribution, and spectral-correlation properties allow the use of Markov models based on recurrent linear prediction mechanism. Models of this type are used in various fields, such as processing of speech signals, sonar systems, as well as a good description of the properties of seismic signals. The fact that sensitive sensors in the system are at a relatively short distance from each other, resulting in mutual dependence of local signals using which can further improve the quality of prediction, thereby reducing the residual background level and, therefore, increase the signal-to-noise ratio. So in the work as an alternative to the existing local processing is offered jointly (vector) processing of signals observed in the group of seismic sensors. Furthermore, a method of identifying a model parameter vector prediction is described.