Neural network combined with fuzzy logic to allow Pressure Sensitive Grouting (PSG)

Grouting of rock becomes more and more important in rock mechanics. If waste is stored in rock deposits sealing of cracks to prevent fluid flow is fundamental. The very important question is, how to control the grouting process in order to avoid dangerous crack extensions. The paper discusses the use of Transient Pressure Analysis (TPA) to control a grouting process using a fuzzy logic approach. The criteria which control this so called Pressure Sensitive Grouting (PSG) are the grouting pressure, the residual pressure, the pressure: grouting volume ratio and the time dependent behaviour. This system combines expert knowledge, experiences from physical and numerical investigations and adjusted rules from training data to control the grouting process. A neural network was used to find the rules fitting best due to known sets of input and output data for the fuzzy logic control algorithm. The advantages and disadvantages of this approach are discussed.