论文部分内容阅读
针对多机器人系统的应用场景,提出了Weibull分布和隐马尔科夫模型相结合的多机器人系统故障预测方法。首先,根据机器人的无故障运行时间估算出机器人可靠性的Weibull分布模型;然后对机器人运动数据采用小波包变换的方法进行特征提取,并训练好状态评估模型,将经过特征提取后的待诊断数据输入训练好的状态评估模型,实现性能评价功能;最后,使用隐马尔科夫模型中期望最大化算法(expectationmaximization,EM)结合Weibull分布进行故障预测模型的训练。通过仿真验证了该方法的可行性和有效性。
A multi-robot system fault prediction method based on Weibull distribution and hidden Markov model is proposed for multi-robot system. Firstly, the Weibull distribution model of robot reliability is estimated based on the robot’s trouble-free operation time. Then, the robot motion data is extracted by wavelet packet transform, and the state evaluation model is trained. After the feature extraction, the data to be diagnosed Input well-trained state evaluation model to realize performance evaluation function; finally, use expectation maximization (EM) combined with Weibull distribution in hidden Markov model to train fault prediction model. The feasibility and effectiveness of this method are verified by simulation.