基于蚁群算法的径向基网络漏磁信号二维反演

2-D Inversion of MFL Signals of RBF Network based on Ant Colony Algorithm

利用蚁群算法的并行寻优特征和一种自适应调整挥发系数的方法,寻找径向基函数(RBF)网络基函数的中心点,能够大大提升网络的鲁棒性及训练速度。将优化算法下的RBF神经网络应用于漏磁信号的二维反演中,能够在训练样本较少的情况下快速地得到更为准确的缺陷模型,更好地满足了漏磁检测的实时性和在线性。实验结果表明,该方法能够达到预期的效果,是一可行的方法。

Applying the feature of parallel search optimum of the ant colony algorithm and a dynamic method to adjust the parameter of evaporation coefficient on searching the center point of each basis function of RBF network,which can improve the robustness and training speed efficiently.Using the RBF network based on this optimization algorithm in two-dimensional inversion of MFL signals,which can receive precise defect model fast,and satisfy the real-time capability and online capability of MFL inspection better.Experimental results show that this method can achieve the expectation.It is a feasible method.