By means of the simulation experiment, the conclusion is obtained that two hidden layers must be used to improve the precision of the load forecasting.

This paper proposes a method which can determine the suitable structure in the hidden layer of a neural network.

With discussion on problems of the hidden layer nodes number, the training sample space matrix in training processes, etc.

A new method to study provenance and subtle reservoir using continental elements and terrigenous minerals;

Influence of non-active concealed fault to seismic response of structure;

During the preparative and mining process of fully mechanized coal mining face, meeting small fold is a premonitor that there may be concealed fault around the face.

In order to eliminate the practical problems due to the predication of concealed faults during coal excavation, the paper presents a comprehensive way to forecast faults concealed in mining fields, based on fuzzy mathematics and mining geology.

Measurable forecast of buried fault in front of mining roadway;

The paper presents some application examples of detecting buried faults with near-surface seismic methods i.

For acquiring the properties and activity of the buried fault under the loose overburden in some large cities,the faults in depth from tens metres to two kilometers have been explorated by use of reflection-seismic prospecting method and combining the data in deep and shallow strata and the properties of the faults and the relations between deep faults and shallow faults are determined.