【摘要】 针对流线型AUV舵故障,提出了基于Elman神经网络的故障诊断方法。基于蚁群算法优化改进型Elman神经网络,建立了AUV角速度运动模型,通过蚁群算法和梯度下降法对改进型Elman神经网络训练的对比分析,验证了蚁群算法优化的改进型Elman神经网络具有训练速度快,不易陷入最优解等特点。提出了基于角速度残差检测舵故障,再通过定角度航行和定速直航的主动诊断方式,判定舵故障类型的故障诊断方法,探讨了基于角速度残差和角度残差的变化趋势来诊断舵卡死和舵变形故障的故障决策方法。对流线型AUV的舵变形及舵卡死故障进行了水池模拟实验,实验结果验证了所提方法的有效性。更多还原

【Abstract】 Aiming at faults of rudders of streamline Autonomous Underwater Vehicles (AUV),a fault diagnosis method based on the Improved Elman Neural Network (IENN) is presented.Based on the IENN optimized by Ant Colony Algorithm(ACA),the angular speed model of AUV is established.Through the contrastive analysis of the IENN’s training program using ACA and decreasing gradient algorithm respectively,it is testified that the IENN optimized by ACA has the advantages of fast training, global convergence and etc.A fault diagnosis method is presented,which detects rudder’s faults firstly by angular speed residuals and then identifies fault type by an active fault diagnosis method.A rudder fault decision-making method based on the angular speed and the change trend of angular residuals is discussed.In pool environment,the faults of distorted rudders and locked rudders have been simulated.The experiment’s results show that the proposed fault diagnosis method is effective.更多还原