On the Use of an Improved Artificial Fish Swarm Algorithm-Backpropagation Neural Network for Predicting Dam Deformation Behavior

Dam behavior is difficult to predict due to its complexity. At the same time, dam deformation behavior is vital to dam systems. Developing a precise prediction model of dam deformation from prototype data is still challenging but determinant in the structural safety assessment. In this paper, an artificial neural network (ANN), trained by the improved artificial fish swarm algorithm (IAFSA) and backpropagation algorithm (BP), is proposed for predicting the dam deformation. Initially, crossover operator is embedded into AFSA, which aims to enhance the performance. In light of the influence mechanism of many factors on dam deformation behavior, the hybrid (IAFSA and BP) model uses statistical input to obtain the optimal connection weights and threshold values of the neural network. The hybrid model integrates IAFSA’s strong global searching ability and BP’s strong local search ability. To avoid overfitting the training set data, a validation set is adopted to check the generalization capability. Subsequently, the obtained optimal parameters are applied to predict the dam deformation behavior. The hybrid model’s preciseness is verified against the radial displacements of a pendulum in a concrete arch dam and simulations of four models: statistical model, BP-ANN optimized by genetic algorithm (GA), particle swarm optimization (PSO), and AFSA. Results demonstrate that the proposed model outperforms other models and may provide alarms for safety control.