基于模糊神经网络的水下机器人推进系统故障诊断

Fault Diagnosis Based on Fuzzy Neural Network of Underwater Vehicle Propulsion System

为了解决水下机器人推进系统运行可靠性问题,提出一种基于模糊神经网络的机器人推进系统故障诊断方法,用以解决故障诊断过程中信息的不确定性问题,并提高推进系统的整体可靠性。该方法在常规神经网络基础上,引入模糊推理形成一种新型模糊神经网络结构,提出一种最小调整的模糊神经网络学习率,完成模糊神经网络训练算法的推导。通过对水下机器人实施定速直航与转向等试验完成神经网络的在线训练,利用已完成训练的神经网络对机器人进行运动建模。通过比对神经网络模型估计值与机器人传感器的实测值获取残差信息,并对残差进行故障信息提取以实现故障诊断。将上述方法应用于仿真试验中,结果表明,基于模糊神经网络水下机器人推进系统故障诊断方法具有较高的可行性和有效性。

To solve the operational reliability problem of the underwater vehicle propulsion system,the fault diagnosis of propulsion system of underwater vehicle based on fuzzy neural network is studied to deal with the non-definiteness of the process of fault diagnosis,so as to enhance system reliability.Based on the conventional neural network,a new fuzzy neural network structure is formed by introducing fuzzy inference,and a learning rate of fuzzy neural network with minimum adjustment is proposed to complete the derivation of the fuzzy neural network training algorithm.The online training of the neural network is completed through the experiments of constant speed direct navigation and steering of the underwater robot,and the motion modeling of the robot is carried out by using the trained neural network.By comparing the estimated value of the neural network model with the measured value of the robot sensor,the residual information is obtained,and the residual fault information is extracted to realize the fault diagnosis.The above method is applied to the simulation test,and the results show that the fault diagnosis method based on fuzzy neural network is feasible and effective.