The Model of Dam Displacement Based on Improved Ant Colony Algorithm-Neural Networks
Yufeng Jiang, Juan Wang · 2009
According to the problems of the nonlinearity and non-norm on dam displacement prediction, the dam displacement mode based on improved ant colony algorithm-neural networks was proposed. The binary ant colony algorithm has been brought into the optimization of weights in Neural Networks. So that the shortcomings of the ant algorithm using in the combinatorial optimization in continuous field have been overcome, while the embarrassment of BP algorithm being vulnerable into the local optimum have been avoided. Therefore, this improved ant colony algorithm-neural networks can have both rapid global convergence ability of binary ant colony algorithms and extensive mapping ability of neural networks. The dam displacement model based on the new ant colony algorithm-neural networks is built by mixed programming, and it has been used for project application. The analysis result shows that this mode is feasible in nonlinear fitting with a high accuracy, and so provides a new method for dam displacement prediction.