<p>No methods are available for predicting settlement of soft ground subjected to alternating radiant drain vacuum preloading (AR-VP). To predict settlement effectively, a novel approach was proposed based on a backpropagation (BP) neural network considering the clogging-layer thickness. A comparative analysis was carried out on the settlements predicted by the BP neural network, Asaoka, hyperbolic, and three-point methods on the basis of the results of a physical model test. The proposed method was found to be highly accurate and therefore, suitable for practical engineering applications. The settlement values predicted by the BP neural network closely matched the measured values, where the range of error was within -0.8 mm to 0.4 mm and the average relative error was 0.473%. The BP neural network considerably outperformed the other three prediction methods.</p>