Volume 39, Number 2, April 2021
|Page(s)||407 - 413|
|Published online||09 June 2021|
A RUL prediction method of equipments based on MSDCNN-LSTM
School of Automation, Beijing Information Science & Technology University, Beijing 100192, China
In order to solve the problems of high data dimension and insufficient consideration of time series correlation information, a multi-scale deep convolutional neural network and long-short-term memory (MSDCNN-LSTM) hybrid model is proposed for remaining useful life (RUL) of equipments. First, the sensor data is processed through normalization and sliding time window to obtain input samples; then multi-scale deep convolutional neural network (MSDCNN) is used to capture detailed spatial features, at the same time, time-dependent features are extracted for effective prediction combining with long short-term memory (LSTM). Experiments on simulation dataset of commercial modular aero-propulsion system show that, compared with other state-of-the-art methods, the prediction method proposed in this paper has achieved better RUL prediction results, especially for the prediction of the life of equipment with complex failure modes and operating conditions. The effect is obvious. It can be seen that the prediction method proposed in this paper is feasible and effective.
针对设备剩余使用寿命（RUL）预测过程中数据维度高，时间序列相关性信息难以充分考虑的实际应用需求，提出一种多尺度深度卷积神经网络和长短时记忆网络融合（multi-scale deep convolutional neural network and long short-term memory，MSDCNN-LSTM）的设备剩余寿命预测方法。对传感器数据进行标准化和滑动时间窗口处理得到输入样本；采用基于多尺度深度卷积神经网络（MSDCNN）提取空间详细特征，采用长短时记忆网络（LSTM）提取时间相关性特征以进行有效的预测。基于商用模块化航空推进系统仿真数据集的实验表明，相较于其他最新方法，文中提出的方法取得了较好的预测结果，尤其是对于故障模式和运行条件复杂的设备寿命预测需求，该方法效果明显。
Key words: remaining useful life / multi-scale deep convolutional neural network / long short-term memory / sliding time window / hybrid prediction model / simulation experiment
关键字 : 剩余使用寿命 / 多尺度深度卷积神经网络 / 长短时记忆网络 / 时间窗口 / 融合预测模型 / 仿真实验
© 2021 Journal of Northwestern Polytechnical University. All rights reserved.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.