基于改进神经网络算法的舰船操纵性能预报系统

Research on prediction system of ship maneuverability based on improved neural network algorithm

针对传统舰船操纵性能预报系统预报精度低、稳定性差、效率慢等弊端,设计一种新的基于改进神经网络的舰船操纵性能预报系统。对标准神经网络算法容易陷入局部最优解的不足进行改进,用共轭梯度法取代原梯度下降算法,介绍了改进后的神经网络算法。通过矩阵方式对舰船操纵稳定性的构成进行描述,利用改进的神经网络算法设计舰船操纵稳定性预报系统。实验结果表明,所设计系统预报精度高、稳定性强、效率快,能有效完成舰船操纵稳定性的预报。

Aiming at the shortcomings of low prediction accuracy, poor stability and slow efficiency of traditional Ship Maneuverability Prediction System, a new Ship Maneuverability Prediction System Based on improved neural network is designed. The improved neural network algorithm is improved by using the conjugate gradient method instead of the original gradient descent algorithm to improve the standard neural network algorithm easily to fall into the local optimum solution.The composition of ship handling stability is described by matrix method, and the prediction system of ship handling stability is designed by using improved neural network algorithm. The experimental results show that the designed system has high prediction accuracy, strong stability and fast efficiency. It can effectively predict the ship handling stability.