By comparing the prediction results of GM(1,1) and BP neural network, the combination of GM and neural model is feasible in oil spectral analysis parameter prediction, which can overcome the deficiency of single model and get good effect.

Parameter prediction models (PPM) and parameter recovery models (PRM) were developed based on the three Weibull distribution parameters and stand attributes.

The system can realize such functions as parameter prediction binaural auralization and model visualization and therefore it can help the acoustic consultants to know a virtual sound field from not only vision, but also acoustic indexes and hearing which will lead to the best design plan.

To reasonably determine prediction parameters in a surface subsidence model for an underground non-bedded shape room environment,the internal relations between prediction parameters and underground room shapes were analyzed based on the principles of surface subsidence.

A method of estimating the thermal properties from the transient experiment based on the solution of the nonlinear inverse problem of parameter estimation by using the Levenberg-Marquardt iterative procedure was presented.