GCA-CG Based Groundwater Level Prediction with Uncertainty in Lower Reaches of Tarim River

It is well known that no uniform prediction approaches were obtained regarding ground water level, though the neural network and some other so-called artificial intelligence methods consistently provide the smallest uncertainty and different medians warranting further research on their abilities. In the present paper, the lower reaches of Tarim River is taken as the study area, a grey correlation analysis and cloud generator (GCA-CG) based groundwater level prediction model is proposed. The most important characteristic feature of the novel model is that the observation data with uncertainty is taken into consideration. First of all, based on the GCA theory, the most important influencing indicator of groundwater level is selected. And then, the CG of knowledge reasoning is applied to predict the groundwater level. Finally, a numerical experiment based on the historical observation data is performed to verify the presented ground water level prediction model, which shows us that the fitting precision is 91.09% before water transportation and 87.84% after the water transportation. From the theoretic foundation and experiment results, we can see that the model could be widely used in other systems with uncertainty.