Analysis of long-term pore water pressure data for the sedimentary rock, 2; Analysis using the genetic algorithm method and the neural network method

Seno, Shoji*; Toida, Masaru*; Watanabe, Kunio*; Sohail, A. R.*; Kunimaru, Takanori

Horonobe Underground Research Center of Japan Atomic Energy Agency has been conducting long-term observation of the pore water pressure fluctuation as part of the research activities in the Horonobe Underground Research Project. The pore water pressures are now monitored at 70 points from 100 to 1,000m in depth of 9 boreholes. In this article, some generic algorithm (GA) method and neural network (BPANN, GAANN) methods were applied to the observed pore water pressure data. The cross-correlations for a single borehole data and for different two boreholes data, or for a pore water pressure data with any other observed data (e.g. earth tide, atmospheric pressure, groundwater level, river water level and flow rates) were investigated and used for the prediction analysis.