摘要
船舶动力设备在自身性能退化过程中的相当长一段时间内仍能完成规定功能,对具有重要特征参数或性能指标的船舶动力设备而言,若使用定基线进行健康状态评估会导致评估值连续较低甚至误报警问题。为了解决这一问题,以目标设备按性能退化时间序列采集的特征参数为研究对象,首先建立退化基线计算方法,利用滑动概率神经网络和性能可靠度与基线值间的转换函数获得目标设备的动态退化基线;然后建立ARMA预测模型获得预测参数,并与退化基线计算方法结合对退化基线发生动态变化的时间节点进行预测;最后利用海水泵对建立的方法可行性进行验证。结果表明,本文建立的退化基线计算方法能够获得动态基线,退化基线预测方法能够对动态基线的变化时间节点进行准确预测。
Marine power equipment can still complete the required functions for a considerable period of time during its performance degradation process.For the equipment with important characteristic parameters or performance indicators,using a fixed baseline for health status assessment can lead to lower assessment values and false alarms.In order to solve this problem,the object of study is the characteristic parameters collected from performance degradation time series of the target equipment.Firstly,establishing degradation baseline calculation method which uses the sliding probability neural network and the conversion function between performance reliability and baseline value to obtain the dynamic degradation baseline.Secondly,establishing ARMA prediction model to obtain prediction parameters,and combining with the degradation baseline calculation method to predict the change time of the baseline value.Finally,using the seawater pump to verify the feasibility of the established methods,the results show that the degradation baseline calculation method can obtain the dynamic baseline,and the degradation baseline prediction method can accurately predict the change time of the dynamic baseline value.
作者
蔡玉良
孙晓磊
张晋彪
方宇
CAI Yu-liang;SUN Xiao-lei;ZHANG Jin-biao;FANG Yu(China Classification Society,Beijing 100007,China;Weichai Power Co.,Ltd.,Weifang 261001,China;Qinhuangdao of China Classification Society,Qinhuangdao 066002,China)
出处
《舰船科学技术》
北大核心
2020年第4期141-147,153,共8页
Ship Science and Technology
关键词
船舶动力设备
退化基线
健康状态评估
滑动概率神经网络
时间序列
预测
marine power equipment
degradation baseline
health status assessment
sliding probability neural network
time series
prediction