Operational time-series data modeling via LSTM network integrating principal component analysis based on human experience

Today’s information technologies involve increasingly intelligent systems, which come at the cost of increasingly complex equipment. Modern monitoring systems collect multi-measuring-point and long-term data which make equipment health prediction a “big data” problem. It is difficult to extract information from such condition monitoring data to accurately estimate or predict health statuses. Deep learning is a powerful tool for big data processing that is widely utilized in image and speech recognition applications, and can also provide effective predictions in industrial processes. This paper proposes the Long Short-term Memory Integrating Principal Component Analysis based on Human Experience (HEPCA-LSTM), which uses operational time-series data for equipment health prognostics. Principal component analysis based on human experience is first conducted to extract condition parameters from the condition monitoring system. The long short-term memory (LSTM) framework is then constructed to predict the target status. Finally, a dynamic update of the prediction model with incoming data is performed at a certain interval to prevent any model misalignment caused by the drifting of relevant variables. The proposed model is validated on a practical case and found to outperform other prediction methods. It utilizes a powerful deep learning analysis method, the LSTM, to fully process big condition monitoring series data; it effectively extracts the features involved with human experience and takes dynamic updates into consideration.     

数据驱动下的半导体生产线动态调度方法

本文研究了一种数据驱动下的半导体生产线调度框架,该框架基于调度优化数据样本,应用机器学习算法,获得动态调度模型,通过该模型,对于半导体生产线,能够根据其当前的生产状态,实时地定出近似最优的调度策略.在此基础上,利用特征选择和分类算法,提出一种生成动态调度模型的方法,并且具体实现出一种混合式特征选择和分类算法的调度模型:先采用过滤式特征选择方法对生产属性进行初步筛选,然后再采用封装式特征选择和分类方法生成模型以提高模型生成的效率.最后,在某实际半导体生产线上,对在所提出的框架上采用6种不同算法实现的动态调度模型进行测试,并对算法性能数据和生产线性能据进行对比和分析.结果表明,数据驱动下的动态调度方法优于单一的调度规则,同时也能满足生产线调度实时性要求.在数据样本较多的情况下,建议采用本文所提出的方法.     

Predictive maintenance using cox proportional hazard deep learning

Predictive maintenance (PdM) has become prevalent in the industry in order to reduce maintenance cost and to achieve sustainable operational management. The core of PdM is to predict the next failure so corresponding maintenance can be scheduled before it happens. The purpose of this study is to establish a Time-Between-Failure (TBF) prediction model through a data-driven approach. For PdM, data sparsity is regarded as a critical issue which can jeopardize algorithm performance for the modelling based on maintenance data. Meanwhile, data censoring has imposed another challenge for handling maintenance data because the censored data is only partially labelled. Furthermore, data sparsity may affect algorithm performance of existing approaches when addressing the data censoring issue. In this study, a new approach called Cox proportional hazard deep learning (CoxPHDL) is proposed to tackle the aforementioned issues of data sparsity and data censoring that are common in the analysis of operational maintenance data. The idea is to offer an integrated solution by taking advantage of deep learning and reliability analysis. To start with, an autoencoder is adopted to convert the nominal data into a robust representation. Secondly, a Cox proportional hazard model (Cox PHM) is researched to estimate the TBF of the censored data. A long-short-term memory (LSTM) network is then established to train the TBF prediction model based on the pre-processed maintenance data. Experimental studies using a sizable real-world fleet maintenance data set provided by a UK fleet company have demonstrated the merits of the proposed approach where the algorithm performance based on the proposed LSTM network has been improved respectively in terms of MCC and RMSE.     

Machine health prognostics using survival probability and support vector machine

Prognostic of machine health estimates the remaining useful life of machine components. It deals with prediction of machine health condition based on past measured data from condition monitoring (CM). It has benefits to reduce the production downtime, spare-parts inventory, maintenance cost, and safety hazards. Many papers have reported the valuable models and methods of prognostics systems. However, it was rarely found the papers deal with censored data, which is common in machine condition monitoring practice. This work concerns with developing intelligent machine prognostics system using survival analysis and support vector machine (SVM). SA utilizes censored and uncensored data collected from CM routine and then estimates the survival probability of failure time of machine components. SVM is trained by data input from CM histories data that corresponds to target vectors of estimated survival probability. After validation process, SVM is employed to predict failure time of individual unit of machine component. Simulation and experimental bearing degradation data are employed to validate the proposed method. The result shows that the proposed method is promising to be a probability-based machine prognostics system.     

A decision support system for production scheduling in an ion plating cell

Production scheduling is one of the major issues in production planning and control of individual production units which lies on the heart of the performance of manufacturing organizations. Traditionally, production planning decision, especially scheduling, was resolved through intuition, experience, and judgment. Machine loading is one of the process planning and scheduling problems that involves a set of part types and a set of tools needed for processing the parts on a set of machines. It provides solution on assigning parts and allocating tools to optimize some predefined measures of productivity. In this study, Ion Plating industry requires similar approaches on allocating customer's order, i.e. grouping production jobs into batches and arrangement of machine loading sequencing for (i) producing products with better quality products; and (ii) enabling to meet due date to satisfy customers. The aim of this research is to develop a Machine Loading Sequencing Genetic Algorithm (MLSGA) model to improve the production efficiency by integrating a bin packing genetic algorithm model in an Ion Plating Cell (IPC), such that the entire system performance can be improved significantly. The proposed production scheduling system will take into account the quality of product and service, inventory holding cost, and machine utilization in Ion Plating. Genetic Algorithm is being chosen since it is one of the best heuristics algorithms on solving optimization problems. In the case studies, industrial data of a precious metal finishing company has been used to simulate the proposed models, and the computational results have been compared with the industrial data. The results of developed models demonstrated that less resource could be required by applying the proposed models in solving production scheduling problem in the IPC.     

An approximate nondominated sorting genetic algorithm to integrate optimization of production scheduling and accurate maintenance based on reliability intervals

With the development of intelligent manufacturing, production scheduling and preventive maintenance are widely applied in industry to enhance production efficiency and machine reliability. Therefore, according to the different processing states and the physical degradation phenomena of the machine, this paper proposes an accurate maintenance (AM) model based on reliability intervals, which have different maintenance activities in diverse intervals and overcome the shortcoming of the single reliability threshold maintenance model used in the past. Combining the flexible job-shop scheduling problem (FJSP), an integrated multiobjective optimization model is established with production scheduling and accurate maintenance. To strengthen the ability of the evolutionary algorithm to solve the presented model/problem, we propose a novel genetic algorithm, named the approximate nondominated sorting genetic algorithm III (ANSGA-III), which is inspired by NSGA-III. To improve the performance of the Pareto dominance principle, the local search, the elite storage for the original algorithm, the approximate dominance principle, the variable neighborhood search, and the elite preservation strategy are proposed. Then, we employ a scheduling example to verify and evaluate the availability of the above three improved operations and the proposed algorithm. Next, we compare ANSGA-III against five recently proposed algorithms, representing the state-of-the-art on similar problems. Finally, we apply ANSGA-III to solve the integrated optimization model, and the results reveal that the machine can maintain higher availability and reliability when compared to other models in our experiments. Consequently, the superiority of the proposed model based on accurate maintenance of reliability intervals is demonstrated, and the optimal reliability threshold between the yellow and red areas is found to be 0.82.     

Rule-based explanations based on ensemble machine learning for detecting sink mark defects in the injection moulding process

Manufacturing quality control (QC) in plastic injection moulding is of the upmost importance since almost one third of plastic products are manufactured via the injection moulding process. Moreover, smart manufacturing technologies are enabling the generation of huge amounts of data in production lines. This data can be used for predicting the quality of manufactured plastic products using machine learning methods, allowing companies to save costs and improve their production efficiency. However, high-performance machine learning models are usually too complicated to be understood by human intuition. Therefore, we have introduced a rule-based explanations (RBE) framework that combines several machine learning interpretation methods to help to understand the decision mechanisms of accurate and complex predictive models – specifically tree ensemble models. These generated rules can be used to visually and easily understand the main factors that affect the quality in the manufacturing process. To demonstrate the applicability of RBE, we present two experiments with real industrial data gathered from a plastic injection moulding machine in a Singapore model factory. The collected datasets contain condition data for several manufacturing processes as well as the QC results for sink mark defects in the production of small plastic products. The experiments revealed that it is possible to extract meaningful explanations in the form of simple decision rules that are enhanced with partial dependence plots and feature importance rankings for a better understanding of the underlying mechanisms and data relationships of accurate tree ensembles.     

A novel diagnostic and prognostic framework for incipient fault detection and remaining service life prediction with application to industrial rotating machines

Data-driven machine health monitoring systems (MHMS) have been widely investigated and applied in the field of machine diagnostics and prognostics with the aim of realizing predictive maintenance. It involves using data to identify early warnings that indicate potential system malfunctioning, predict when system failure might occur, and pre-emptively service equipment to avoid unscheduled downtime. One of the most critical aspects of data-driven MHMS is the provision of incipient fault diagnosis and prognosis regarding the system’s future working conditions. In this work, a novel diagnostic and prognostic framework is proposed to detect incipient faults and estimate remaining service life (RSL) of rotating machinery. In the proposed framework, a novel canonical variate analysis (CVA)-based monitoring index, which takes into account the distinctions between past and future canonical variables, is employed for carrying out incipient fault diagnosis. By incorporating the exponentially weighted moving average (EWMA) technique, a novel fault identification approach based on Pearson correlation analysis is presented and utilized to identify the influential variables that are most likely associated with the fault. Moreover, an enhanced metabolism grey forecasting model (MGFM) approach is developed for RSL prediction. Particle filter (PF) is employed to modify the traditional grey forecasting model for improving its prediction performance. The enhanced MGFM approach is designed to address two generic issues namely dealing with scarce data and quantifying the uncertainty of RSL in a probabilistic form, which are often encountered in the prognostics of safety-critical and complex assets. The proposed CVA-based index is validated on slowly evolving faults in a continuous stirred tank reactor (CSTR) system, and the effectiveness of the proposed integrated diagnostic and prognostic method for the monitoring of rotating machinery is demonstrated for slow involving faults in two case studies of an operational industrial centrifugal pump and one case study of an operational centrifugal compressor.     

Integrated scheduling and flexible maintenance in deteriorating multi-state single machine system using a reinforcement learning approach

This paper studies an integrated optimization problem of production scheduling and flexible preventive maintenance (PM) in a multi-state single machine system with deteriorating effects. A flexible PM strategy is proposed to proactively cope with machine failures while ensuring relatively regular PM intervals, which is composed of time-based PM (TBPM) and condition-based PM (CBPM). TBPM is conducted within every flexible time window and CBPM is implemented immediately after the most deteriorated yet still functional state. An illustrative case is presented using the enumeration approach to demonstrate the integration of production scheduling and machine maintenance. Then, Q-learning-based solution framework (QLSF) is further designed with proper state and action sets and reward functions to facilitate the determination of appropriate production scheduling rule under the constraint of the flexible maintenance. Numerical experiments show that the proposed QLSF outperforms the other four state-of-the-art scheduling rules in different scenarios. Moreover, the performance of the proposed flexible PM strategy is also examined and validated in comparison with three candidate maintenance strategies, i.e., run-to-failure corrective maintenance (CM), combination of TBPM and CM, and CBPM. The proposed flexible maintenance and solution approach can enrich the relevant academic knowledge base, and provide managerial insights and guidance in practical production systems.     

Dependability of wireless sensor networks for industrial prognostics and health management

Maintenance is an important activity in industry. It is performed either to revive a machine/component or to prevent it from breaking down. Different strategies have evolved through time, bringing maintenance to its current state: condition-based and predictive maintenances. This evolution was due to the increasing demand of reliability in industry. The key process of condition-based and predictive maintenances is prognostics and health management, and it is a tool to predict the remaining useful life of engineering assets. Nowadays, plants are required to avoid shutdowns while offering safety and reliability. Nevertheless, planning a maintenance activity requires accurate information about the system/component health state. Such information is usually gathered by means of independent sensor nodes. In this study, we consider the case where the nodes are interconnected and form a wireless sensor network. As far as we know, no research work has considered such a case of study for prognostics. Regarding the importance of data accuracy, a good prognostics requires reliable sources of information. This is why, in this paper, we will first discuss the dependability of wireless sensor networks, and then present a state of the art in prognostic and health management activities.     

Product Completion Time Prediction Using A Hybrid Approach Combining Deep Learning and System Model

The prediction of product completion time is critical in real time production scheduling and control to achieve customer demand satisfaction. However, it is a challenging task due to the increasing complexity of production systems and greater diversity of products. The recent advancement in data-driven approach and machine learning algorithms have provided unprecedent opportunities to tackle such problems that otherwise very difficult to solve using conventional methods in the manufacturing industry. However, most existing studies on product completion time prediction adopt a purely data-driven approach while ignoring the prospect of integrating domain knowledge in their machine learning models. In this paper, we propose a hybrid approach to predicting product completion time by combining the strengths of both machine learning techniques and analytical system model. A mathematical model for multi-product serial production line is proposed to describe the real-time dynamics of the system. With this model, the strict lower bound of product completion time can be efficiently computed for given system status, where the lower bound represents the least possible product completion time when assuming no random downtime in the system. Instead of directly predicting the product completion time, a deep learning model is developed to only predict the distance between the lower bound and actual product completion time. Guided by properties of production system, we discover a recurrent sequence in the prediction problem by modeling each machine and product as recurrent units. The Long Short-Term Memory (LSTM) method, a prominent variant of recurrent neural network (RNN), is used to combine with the system model to predict the product completion time in a real-time fashion.     

Development of Industry 4.0 predictive maintenance architecture for broadcasting chain

Industry 4.0 Predictive Maintenance (PdM 4.0) architecture in the broadcasting chain is one of the taxonomy challenges for deploying Industry 4.0 frameworks. This paper proposes a novel PdM framework based on advanced Reference Architecture Model Industry 4.0 (RAMI 4.0) to reduce operation and maintenance costs. This framework includes real-time production monitoring, business processes, and integration based on Design Science Research (DSR) to generate an innovative Business Process Model and Notation (BPMN) meta-model. The addressed model visualizes sub-processes based on experts' and stakeholders' knowledge to reduce the cost of maintenance of audiovisual services including satellite TV, cable TV, and live audio and video broadcast services. Based on the recommendation and the concept of Industry 4.0, the proposed framework tolerates the predictable failures and further concerns in similar related industries. Some empirical experiments have been conducted by using the Islamic Republic of Iran Broadcasting’s (IRIB) high-power station (located near the capital city of Iran, Tehran) to evaluate the functionality and efficiency of the proposed predictive maintenance framework. Practical outcomes demonstrate that interval times between data collection should be increased in audio and video broadcasting predictive maintenance because of the limitation of the internal processing performance of equipment. The framework also indicates the role of the Frequency Modulation (FM) transmitters’ data clearance to reduce the instability and untrustworthy data during data mining. The proposed DSR method endorses using a customized RAMI 4.0 meta-model framework to adapt distributed broadcasting and communication with PdM 4.0, which increases the stability as well as decreasing maintenance costs of the broadcasting chain in comparison to state-of-the-art methodologies. Furthermore, it is shown that the proposed framework outperforms the best-evaluated methods in terms of acceptance.     

Single-machine and flowshop scheduling with a general learning effect model

Learning effects in scheduling problems have received growing attention recently. Biskup [Biskup, D. (2008). A state-of-the-art review on scheduling with learning effect. European Journal of Operational Research, 188, 315–329] classified the learning effect scheduling models into two diverse approaches. The position-based learning model seems to be a realistic assumption for the case that the actual processing of the job is mainly machine driven, while the sum-of-processing-time-based learning model takes into account the experience the workers gain from producing the jobs. In this paper, we propose a learning model which considers both the machine and human learning effects simultaneously. We first show that the position-based learning and the sum-of-processing-time-based learning models in the literature are special cases of the proposed model. Moreover, we present the solution procedures for some single-machine and some flowshop problems.     

A hybrid framework combining data-driven and model-based methods for system remaining useful life prediction

Remaining useful life prediction is one of the key requirements in prognostics and health management. While a system or component exhibits degradation during its life cycle, there are various methods to predict its future performance and assess the time frame until it does no longer perform its desired functionality. The proposed data-driven and model-based hybrid/fusion prognostics framework interfaces a classical Bayesian model-based prognostics approach, namely particle filter, with two data-driven methods in purpose of improving the prediction accuracy. The first data-driven method establishes the measurement model (inferring the measurements from the internal system state) to account for situations where the internal system state is not accessible through direct measurements. The second data-driven method extrapolates the measurements beyond the range of actually available measurements to feed them back to the model-based method which further updates the particles and their weights during the long-term prediction phase. By leveraging the strengths of the data-driven and model-based methods, the proposed fusion prognostics framework can bridge the gap between data-driven prognostics and model-based prognostics when both abundant historical data and knowledge of the physical degradation process are available. The proposed framework was successfully applied on lithium-ion battery remaining useful life prediction and achieved a significantly better accuracy compared to the classical particle filter approach.     

Tool wear monitoring and prognostics challenges: a comparison of connectionist methods toward an adaptive ensemble model

In a high speed milling operation the cutting tool acts as a backbone of machining process, which requires timely replacement to avoid loss of costly workpiece or machine downtime. To this aim, prognostics is applied for predicting tool wear and estimating its life span to replace the cutting tool before failure. However, the life span of cutting tools varies between minutes or hours, therefore time is critical for tool condition monitoring. Moreover, complex nature of manufacturing process requires models that can accurately predict tool degradation and provide confidence for decisions. In this context, a data-driven connectionist approach is proposed for tool condition monitoring application. In brief, an ensemble of Summation Wavelet-Extreme Learning Machine models is proposed with incremental learning scheme. The proposed approach is validated on cutting force measurements data from Computer Numerical Control machine. Results clearly show the significance of our proposition.     

Single-machine scheduling problems simultaneously with deterioration and learning effects under deteriorating multi-maintenance activities consideration

Maintenance is important to manufacturing process as it helps improve the efficiency of production. Although different models of joint deterioration and learning effects have been studied extensively in various areas, it has rarely been studied in the context of scheduling with maintenance activities. This paper considers scheduling with jointly the deterioration and learning effects and multi-maintenance activities on a single-machine setting. We assume that the machine may have several maintenance activities to improve its production efficiency during the scheduling horizon, and the duration of each maintenance activity depends on the running time of the machine. The objectives are to determine the optimal maintenance frequencies, the optimal maintenance locations, and the optimal job schedule such that the makespan and the total completion time are minimized, respectively, when the upper bound of the maintenance frequencies on the machine is known in advance. We show that all the problems studied can be solved by polynomial time algorithms.     

Intelligent machinery health prognostics under variable operation conditions with limited and variable-length data

High accuracy health prognostics are significant to machinery intelligent operation and maintenance. Current data-driven prognostics achieve great success that benefits from amply learning samples. In fact, data scarcity challenge widely exists in machinery prognostics and health management, especially for high-end equipment. This study aims to solve this dilemma and proposes a novel meta learning algorithm reconstructed by classic variable-length prediction mode and attention mechanism, namely meta attention recurrent neural network (MARNN). Specifically, we first develop the encoder-decoder with attention mechanism (EDA) cell to perform episodic learning for the subtask-level upgrade. Then multiple subtasks with EDA as prediction models are aggregated to accomplish meta-level upgrade, thus mining the general degradation knowledge from historical datasets. Finally, cross-domain prognostics tasks can be easily realized through fine-tuning tricks, and three rotating machinery run-to-failed experiments are conducted to prove the generalizations of MARNN, which can obtain desired results even when the on-site adaptation data is reduced to one-twentieth.     

Simulation and sensor data fusion for machine learning application

The performance of machine learning algorithms depends to a large extent on the amount and the quality of data available for training. Simulations are most often used as test-beds for assessing the performance of trained models on simulated environment before deployment in real-world. They can also be used for data annotation, i.e, assigning labels to observed data, providing thus background knowledge for domain experts. We want to integrate this knowledge into the machine learning process and, at the same time, use the simulation as an additional data source. Therefore, we present a framework that allows for the combination of real-world observations and simulation data at two levels, namely the data or the model level. At the data level, observations and simulation data are integrated to form an enriched data set for learning. At the model level, the models learned from observed and simulated data separately are combined using an ensemble technique. Based on the trade-off between model bias and variance, an automatic selection of the appropriate fusion level is proposed. Our framework is validated using two case studies of very different types. The first is an industry 4.0 use case consisting of monitoring a milling process in real-time. The second is an application in astroparticle physics for background suppression.     

An update method for digital twin multi-dimension models

Digital twin, as an effective means to realize the fusion between physical and virtual spaces, has attracted more and more attention in the past few years. Based on ultra-fidelity models, more accurate service, e.g. real-time monitoring and failure prediction, can be reached. Against the background, some scholars studied the related theories and methods on modeling to depict various features of physical objects. Some scholars studied how to use Internet of Things to realize the connections and interactions, thereby keeping the consistency between the virtual and physical spaces. During this process, a new question arises that how to update the models once digital twin models are inconsistent with the practical situations. To solve the problem, this paper proposed a general digital twin model update framework at first. Then, the update methods for multi-dimension models are further explored. The cutting tool is the core component of machine tools which are the key equipment in industry. The precise cutting tool models are essential for realizing the digitalization and servitization of machine tools. Therefore, this paper takes a cutting tool as the application object to discuss how to conduct physics model update based on the proposed framework and methods. Through model update, a more accurate and updated tool wear model could be obtained, which contributes to the prognostics and health management for machine tools.     

复杂系统维护策略最新研究进展: 从视情维护到预测性维护

对于复杂、可修复的工程系统, 设备维护是确保系统安全性、可靠性、可用性的重要手段之一. 系统维护策略已经历修复性维护、定时维护、视情维护等多种维护策略. 其中, 视情维护是目前最受关注的维护策略, 它通过收集和评估系统的实时状态信息进行维护决策, 具有全寿命周期内系统可靠性高、运营维护成本低等优点. 近年来, 随着物联网技术、信息技术和人工智能的快速发展, 一种更新颖的视情维护策略——预测性维护逐渐成为领域研究热点. 本文首先简要回顾了系统维护策略的发展历程; 然后, 重点介绍了视情维护的研究进展, 根据决策支持技术的不同, 将视情维护划分为基于随机退化模型的视情维护和基于数据驱动的预测性维护, 对每类技术的发展分支与研究现状进行了疏理、分析和总结; 最后, 探讨了当前复杂系统维护策略面临的挑战性问题和可能的未来研究方向.