基于改进LSTM的船体监测数据异常处理方法

Exception Handling Method for Hull Monitoring Data Based on Improved LSTM

为了解决船体监测数据异常识别模型适应性差、异常修复过程低效且准确率不高等问题,提出一种基于改进长短期记忆网络(LSTM)对异常数据进行识别和修复,并采用BiLSTM-AEE对应变和加速度数据进行试验验证。结果表明,该方法在识别精度和修复效果方面都有明显优势,其中异常识别精度平均值达到91.8%,异常修复平均误差不超过4%,能有效对船体监测数据进行异常的识别与修复。相比其他异常数据处理方法,该方法能够根据监测数据变化对异常进行同步识别,修复过程更加高效。

In order to solve the problems of poor adaptability of the abnormal recognition model of hull monitoring data,inefficient and inaccurate error of the abnormal repair process,an improved long short-term memory(LSTM)method is proposed to identify and repair abnormal data.The results show that the method has obvious advantages in recognition accuracy and repair effect,among which the average accuracy of abnormal recognition reaches 91.80%,and the average error of abnormal repair does not exceed 4%,which can effectively identify and repair abnormalities in hull monitoring data.Compared with other abnormal data processing methods,the method can synchronously identify exceptions according to the changes of monitoring data,and the repair process is more efficient.