Comparison of HST and ANFIS.

For the determination of global stability after long term advanced monitoring, artificial intelligence have been used for the data analysis of water level and displacements of Songloulou earth dam at Cameroon. Measurements of safety and reliability indicators follow changes set by piezometric and pendulums measurements. The results obtained from the artificial intelligence on the base of many years recording data have confirmed the relevance and robustness of this model. The ANFIS model combining the concept of neural network and fuzzy logic was used to simulate the behaviour of piezometers and pendulums in the dam. This model has provided satisfactory results, given in the large amount of data to be processed. The water level evolution is modelled using the ANFIS function integrated in the MATLAB software and the result is compared to that obtained by the HST method. Afterwards, the state of stress on the structure and stability of the slope at shear have been assessed based on the hydro mechanical behaviour using the GEOSTUDIO Finite Element computation software. The input parameters are: the head of water recorded in the piezometers and geotechnical parameters of the dam. The modelling results in terms of displacement are accurately consistent with the displacement measurements. The horizontal displacement of pendulums obtained by GEOSTUDIO is 80 mm and those measured directly of the pendulums have 70 mm of average value. The safety factor for slope stability according to 530 m water level is 1.5.