基于MPSO的RBF耦合算法的桩基动测参数辨识

变异粒子群优化算法(MPSO)是一种基于群体智能的改进全局优化技术,其优势在丁减小陷入局部极值的机率,增加全局搜索能力.将变异粒子群算法与径向基函数(RBF)神经网络结构进行结合,建立了变异粒子群神经网络(MPSO-RBF)耦合算法,充分发挥了MPSO算法的全局寻优能力和RBF算法的局部搜索优势.数值计算结果表明,所建立的方法能够对桩基动测进行多参数的识别和非线性优化问题的求解,具有良好全局收敛能力,是一种行之有效的智能算法.

A RBF neural network coupling algorithm based on MPSO for parameter identification of piles in dynamic testing

Mutation particle swarm optimization (MPSO) is a kind of improved stochastic global optimization based on swarm intelligence. The advantages of MPSO are that the probability falling into the local extreme values can be reduced; and the global optimal searching capability is improved. A new algorithm which combined MPSO with radial basis function (RBF) is presented. It not only has the advantage of the global optimization of MPSO, but also has local accurate searching of RBF. Numerical example shows that the presented method can solve the problem which includes multi-parameters identification and nonlinear optimization problem. This approach has the characteristics of global convergence. The intelligent algorithm is simple and precise.