水下机器人推进器状态评估智能方法研究

Research on Intelligent Method for Underwater Robot Thruster Assessment

为了保证水下机器人推进器运行的稳定性与可靠性,提出了一种状态评估方法。利用小波包分析法提取推进器电机的定子电流能量值,经过BP神经网络分类后输出属于每种模式的概率,根据概率计算出推进器的健康度。对推进器进行故障仿真实验,结果表明:小波包分析法能有效提取推进器不同故障模式下故障特征;实现了在有限数量样本的情况下模式的分类,其准确率达到了95.6%,说明了所提取的故障特征的有效性;在正确分类的基础上进行状态判断的准确率达到了100%,利用健康度能得出推进器的实时健康状态。

In order to ensure the stability and reliability of the operation of the underwater robot(Romote Operated Vehicle,ROV)thruster,a state assessment method is proposed.The wavelet packet analysis(Wavelet Packet,WPT)is used to extract the stator current energy value of the propeller motor,and after classifying by BP(back propagation)neural network,the probability of each mode is output,and the health of the propeller is calculated according to the probability.The fault simulation experiments on the thruster show that the wavelet packet analysis method can effectively extract the fault characteristics of the thruster in different fault modes;the classification accuracy of the four modes reaches 95.6%,with a limited number of samples.Further illustrates the effectiveness of the extracted fault features;the accuracy of the state judgment based on the correct classification has reached 100%,and the real-time health status of the thruster can be obtained by using the health degree.