Hazardous down lifting regions detection in a power plant site via artificial neural networks from microgravity data

For infra structures like towers, power plants, heavy petrochemical equipments and so on, one of the hazardous risks is down lifting of the ground surface. It is because of their high weight which is applied a high pressure to subsurface cavities and will trigger the down lifting process. To avoid this hazardous phenomenon, it’s necessary to detect the subsurface cavities, especially before the building stage of site will be started. As an example it ahs been observed down lifting of some cooling towers in Shahid Mofatteh Hamadan Power site, Hmadan,Iran.Subsurface cavities have a negative contrast density which cause to a reduction of earth’s gravity value. Also the gravimeter accuracy is necessary to be at least 1 μ gal to detect this reduction of gravity.In this way we have presented an intelligent method to detect these hazardous subsurface cavities. We designed a Multi Layer Perceptron(MLP) neural network that estimates the depth and size of the related cavities which plays an important role in triggering the dangerous downliftings . The most advantage of this method is no need to do or repeat complicated calculations for new collected microgravity data, after training the network. Flexibility of the network in present of noise and for different domains of depth is notable. We presented these abilities for noisy data and also real data.

Keywords:

Down lifting , Microgravity , Artifitial Neural Network