基于神经网络的磁性目标磁矩反演方法

针对磁性目标定位中的磁矩反演问题，提出一种基于神经网络的磁矩反演技术。首先，基于最小二乘原理，建立了磁性目标磁矩反演模型；其次采用Hopfield网络进行了优化求解，并针对模型求解过程中鲁棒性差的弊端，对网络进化策略进行了自适应修正；最后设计了仿真实验对其有效性进行了检验，仿真结果表明利用修正后的网络求解磁矩反演问题结果令人满意，具有一定的实用性。

Magnetic Moments Estimation Method of a Magnetic Object Based on Neural Network

In localization of magnetic objects, a magnetic moments estimation method is proposed based on neural network. Firstly, a mathematical model is founded to estimate magnetic moments by minimizing the least-squares error at the observation points. Secondly, Hopfield neural network is used to solve the above optimization model and a self-adaptive correction algorithm is taken to improve the robustness of the model. Finally, a numerical experiment is designed to verify the effectiveness of the proposed method. The results show that the practicable method has many advantages such as high accuracy, good robustness, and easy implement.