A Deep Learning Model for Predictive Maintenance in Cyber-Physical Production Systems Using LSTM Autoencoders

Abstract

Condition monitoring of industrial equipment, combined with machine learning algorithms, may significantly improve maintenance activities on modern cyber-physical production systems. However, data of proper quality and of adequate quantity, modeling both good operational conditions as well as abnormal situations throughout the operational lifecycle, are required. Nevertheless, this is difficult to acquire in a non-destructive approach. In this context, this study investigates an approach to enable a transition from preventive maintenance activities, that are scheduled at predetermined time intervals, into predictive ones. In order to enable such approaches in a cyber-physical production system, a deep learning algorithm is used, allowing for maintenance activities to be planned according to the actual operational status of the machine and not in advance. An autoencoder-based methodology is employed for classifying real-world machine and sensor data, into a set of condition-related labels. Real-world data collected from manufacturing operations are used for training and testing a prototype implementation of Long Short-Term Memory autoencoders for estimating the remaining useful life of the monitored equipment. Finally, the proposed approach is evaluated in a use case related to a steel industry production process.

A hybrid of convolutional neural network and long short-term memory network approach to predictive maintenance

“…Average F-Score (%) Gradient boosting decision tree (GBDT) [5] 94.59 Recurrent neural networks (RNN) [13] 86.00 Support vector machine (SVM) [14] 85.00 Support vector machine (SVM) [15] 81.00 LSTM autoencoders [17] 94.20 LSTM 93.34 The proposed hybrid CNN-LSTM 97.48…”

“…The PdM model is built using LSTM and RNN deep neural networks and used to predict the RUL for light bulbs with a minimum error rate of 0.79%. Bampoula et al [17] utilized LSTM autoencoders deep learning method to build a model for planning maintenance in Cyber-Physical Production Systems. The autoencoder deplaning model is used to classify classifying real-world machine status based on the machine's sensor data.…”

Prognostics and Health Management in Nuclear Power Plants: An Updated Method-Centric Review With Special Focus on Data-Driven Methods

“…LSTM-based networks have gained greater attention in applications of RUL prediction. Some recent studies include Ramuhalli et al (2020) using NPP asset data from the feedwater and condensate system (FWCS) of a boiling water reactor (BWR); Zhao et al (2017b) using a convolutional bidirectional LSTM and raw sensory data from high-speed milling machine cutters for a real-life tool wear test; Zhang et al (2018), Wu et al (2018b), andElsheikh et al (2019) using different variants of LSTM on the C-MAPSS data set; Shi and Chehade (2021) using a novel dual-LSTM framework for both change point detection and RUL prediction on the same C-MAPSS data; and Bampoula et al (2021) using LSTM autoencoders to estimate RUL in a cyber-physical production system. Besides the above two commonly used ANNs, several other variants-such as wavelet neural network (Javed et al, 2014), CNN variants (Wang et al, 2019c;Zhu et al, 2019), generative adversarial network , and reinforcement learning (Kozjek et al, 2020)-can be found in the literature of prognostics.…”

SOPRENE: Assessment of the Spanish Armada’s Predictive Maintenance Tool for Naval Assets

“…In the state of the art different PdM techniques using LSTM after an autoencoder to predict malfunctioning components or assets have been implemented on data from temperature sensors, flowmeters, pressure and speed sensors in industrial machinery [44][45][46][47][48] and based on data vibration [49]. However, it is necessary to highlight, at this point, the efforts of the academy to advance in the knowledge with respect to the application of unsupervised techniques on naval machinery as propulsion devices [50][51][52].…”