Research On Assessment Of Sand Liquefaction Potential Based On Compensative Fuzzy Neural Network

Assessment of sand liquefaction potential is the important link in the works ofantiknock and disaster-prevention, and is important contents in reconnaissance ofgeotechnical engineering. This paper has summarized some methods of assessment ofsand liquefaction potential at home and abroad, and indicated their merits anddemerits. The technique of fuzzy neural network has been applied to assess the sandliquefaction potential on the basis of the summarization. So as to a new path wasfound out in the field of assessment sand liquefaction potential. This paper has illustrated the essential theory of fuzzy logic system and neuralnetwork,and has also discussed some issues on compensative fuzzy neural network inthe assessment of sand liquefaction potential. Through detailed analyzing someinfluencing factors of sand liquefaction, seven parameters were selected as assessmentindexes. They are earthquake magnitude, peak ground surface acceleration, standardpenetration value, specific penetration resistance, relative compaction, averageparticle diameter, and water table. The compensative fuzzy neural network model hasbeen established that was applied to assessment of sand liquefaction potential. The netmodel had been trained by means of document of earthquake disaster in Tangshan andhad been assessed in virtue of sand liquefaction example from part district in SanshuiGuangdong province and building site of China international exhibition center bysimulating. The research expresses that compensative fuzzy neural network is a usefultool in assessment and predicting liquefaction potential of sand.