A digital twin to train deep reinforcement learning agent for smart manufacturing plants: Environment,interfaces and intelligence

Filling the gaps between virtual and physical systems will open new doors in Smart Manufacturing. This work proposes a data-driven approach to utilize digital transformation methods to automate smart manufacturing systems. This is fundamentally enabled by using a digital twin to represent manufacturing cells, simulate system behaviors, predict process faults, and adaptively control manipulated variables. First, the manufacturing cell is accommodated to environments such as computer-aided applications, industrial Product Lifecycle Management solutions, and control platforms for automation systems. Second, a network of interfaces between the environments is designed and implemented to enable communication between the digital world and physical manufacturing plant, so that near-synchronous controls can be achieved. Third, capabilities of some members in the family of Deep Reinforcement Learning (DRL) are discussed with manufacturing features within the context of Smart Manufacturing. Trained results for Deep Q Learning algorithms are finally presented in this work as a case study to incorporate DRL-based artificial intelligence to the industrial control process. As a result, developed control methodology, named Digital Engine, is expected to acquire process knowledges, schedule manufacturing tasks, identify optimal actions, and demonstrate control robustness. The authors show that integrating a smart agent into the industrial platforms further expands the usage of the system-level digital twin, where intelligent control algorithms are trained and verified upfront before deployed to the physical world for implementation. Moreover, DRL approach to automated manufacturing control problems under facile optimization environments will be a novel combination between data science and manufacturing industries.     

An integrated framework for blockchain-enabled supply chain trust management towards smart manufacturing

With the development of a new generation of information technology, smart manufacturing has put forward higher requirements for supply chain. It is necessary to ensure the synchronization of the supply chain operation and maintain the reliability of the supply chain management, therefore the trust evaluation for the supply chain becomes extremely important. Traditional supply chain management has problems such as information flow is easy to be tampered with, logistics is difficult to trace, and capital flow is not true, which leads to increased opportunity costs due to the lack of trust among transaction entities in the supply chain. The emergence of blockchain technology provides an opportunity to improve the supply chain ecosystem. In this paper, an integrated framework for blockchain-enabled supply chain trust management towards smart manufacturing is proposed to explain how to enhance trust management with the help of blockchain from the perspectives of information flow, logistics, and capital flow. An optimized trust management model is designed for better entities evaluation in supply chain. A coal mine equipment manufacturing industry scenario is presented to illustrate the effectiveness of the proposed framework.     

Smart experience-oriented customer requirement analysis for smart product service system: A novel hesitant fuzzy linguistic cloud DEMATEL method

Smart product service system (PSS) has become an essential strategy to transform towards digital servitization for manufacturing companies. By leveraging smart capabilities, smart PSS aims to create superior user experience in a smart context. To develop a successful smart PSS, customer requirement management from smart experience perspective is necessary. However, it is a challenging task to identify and evaluate diverse, implicit and interrelated smart experience-oriented customer requirement (SEO-CR) in smart PSS context. Hence, this paper proposes an effective methodology to elicit and analyze SEO-CRs. At first, a generic, two-dimensional SEO-CR system is presented as a basis to derive the tailored SEO-CRs for various smart PSS applications. Second, a novel HFLC-DEMATEL (hesitant fuzzy linguistic cloud-based Decision-making and trial evaluation laboratory) method is proposed to accurately evaluate the priority and complicated interaction of SEO-CRs, considering the hesitancy, fuzziness and randomness under uncertain decision environment. Some new operations (e.g., cloud total-relation matrix and weight determination method) and a cloud influence relation map are developed to fully take advantage of cloud model in DEMATEL implementation. Finally, a real case of smart vehicle service system (SVSS) is presented. The 18 SEO-CRs of the SVSS are derived based on the generalized SEO-CRs. By using HFLC-DEMATEL, some important SEO-CRs in context of SVSS are identified, such as autonomous and convenience. The finding of results can help designers make proper decisions in design and development of SVSS with a superior smart experience. The effectiveness and reliability of the proposed method are validated by conducting some comparative analyses.     

A hybrid intelligence approach for sustainable service innovation of smart and connected product: A case study

Nowadays, smart and connected product (SCP) is gradually replacing the traditional functional products, which has attracted widespread attention from the industry and academia. Service innovation, as a crucial part of SCP iterative improvement, is a multi-criteria decision-making (DM) process facilitated by intelligent automation (IA) and cognitive technologies. However, product user’s intelligence (e.g. physiological feeling) that can intuitively reflect and evaluate the product service satisfaction is rarely considered in the process of service innovation. Hence, it is difficult to measure the product users’ preferences with precise numerical terms to make a strategic decision. In order to fill this gap, a hybrid intelligence approach is proposed to perform the service innovation for SCP. The product-user data (e.g. subjective data and physiological data) and product-sensor data are collected and used for the process of service innovation. A smart group spinning bicycle system is presented as an elaborated case study to illustrate the proposed architecture and approach. The service innovation of real-time and dynamic monitoring, user participation improvement and smart feedback manners are achieved. In addition, an ergonomic experiment is conducted to validate the effectiveness of the proposed approach in implementing the service innovation for SCP.     

Performance evaluation of ambient services by combining robotic frameworks and a smart environment platform

Mobile robots and smart environments are two areas of research that can easily profit from each other. Smart environments, which are spaces unobtrusively equipped with sensors and actuators, providing ambient services to the people living within. Mobile robots inside those smart environments can use the existing infrastructure to increase their performance while decreasing the cost of local sensor systems. On the other side, evaluation of ambient services is often a laborious task. This work presents an approach that simplifies the evaluation by making use of two frameworks from robotics to perform tests in simulated smart environments. A method based on the language as action principle is used to extract realistic behavior of people living in real-world smart environments. Using this data, many different scenarios with varying configurations (different floor layouts, numbers and types of sensors, different number of people and pets) can easily be simulated and the performance of the ambient services evaluated.     

Digital twins-based smart manufacturing system design in Industry 4.0: A review

A smart manufacturing system (SMS) is a multi-field physical system with complex couplings among various components. Usually, designers in various fields can only design subsystems of an SMS based on the limited cognition of dynamics. Conducting SMS designs concurrently and developing a unified model to effectively imitate every interaction and behavior of manufacturing processes are challenging. As an emerging technology, digital twins can achieve semi-physical simulations to reduce the vast time and cost of physical commissioning/reconfiguration by the early detection of design errors/flaws of the SMS. However, the development of the digital twins concept in the SMS design remains vague. An innovative Function-Structure-Behavior-Control-Intelligence-Performance (FSBCIP) framework is proposed to review how digital twins technologies are integrated into and promote the SMS design based on a literature search in the Web of Science database. The definitions, frameworks, major design steps, new blueprint models, key enabling technologies, design cases, and research directions of digital twins-based SMS design are presented in this survey. It is expected that this survey will shed new light on urgent industrial concerns in developing new SMSs in the Industry 4.0 era.     

Digital twin-based smart assembly process design and application framework for complex products and its case study

With rapid advances in new generation information technologies, digital twin (DT), and cyber-physical system, smart assembly has become a core focus for intelligent manufacturing in the fourth industrial evolution. Deep integration between information and physical worlds is a key phase to develop smart assembly process design that bridge the gap between product assembly design and manufacturing. This paper presents a digital twin reference model for smart assembly process design, and proposes an application framework for DT-based smart assembly with three layers. Product assembly station components are detailed in the physical space layer; two main modules, communication connection and data processing, are introduced in the interaction layer; and we discuss working mechanisms of assembly process planning, simulation, predication, and control management in the virtual space layer in detail. A case study shows the proposed approach application for an experimental simplified satellite assembly case using the DT-based assembly application system (DT-AAS) to verify the proposed application framework and method effectiveness.     

Augmented Lagrangian coordination for energy-optimal allocation of smart manufacturing services

The rapid development of information and communication technologies has triggered the proposition and implementation of smart manufacturing paradigms. In this regard, efficient allocation of smart manufacturing services (SMSs) can provide a sustainable manner for promoting cleaner production. Currently, centralized optimization methods have been widely used to complete the optimal allocation of SMSs. However, personalized manufacturing tasks usually belong to diverse production domains. The centralized optimization methods could hardly include related production knowledge of all manufacturing tasks in an individual decision model. Consequently, it is difficult to provide satisfactory SMSs for meeting customer's requirements. In addition, energy consumption is rarely considered in the SMS allocation process which is unfavorable for performing sustainable manufacturing. To address these challenges, augmented Lagrangian coordination (ALC), a novel distributed optimization method is proposed to deal with the energy-optimal SMS allocation problem in this paper. The energy-optimal SMS allocation model is constructed and decomposed into several loose-coupled and distributed elements. Two variants of the ALC method are implemented to formulate the proposed problem and obtain final SMS allocation results. A case study is employed to verify the superiority of the proposed method in dealing with energy-optimal SMS allocation problems by comparing with the centralized optimization method at last.     

Metamodeling of Smart Environments: from design to implementation

A smart environment is a physical environment enriched with sensing, actuation, communication and computation capabilities aiming at acquiring and exploiting knowledge about the environment so as to adapt itself to its inhabitants’ preferences and requirements. In this domain, there is the need of tools supporting the design and analysis of applications. In this paper, the Smart Environment Metamodel (SEM) framework is proposed. The framework allows to model applications by exploiting concepts specific to the smart environment domain. SEM approaches the modeling from two different points of view, namely the functional and data perspectives. The application of the framework is supported by a set of general guidelines to drive the analysis, the design and the implementation of smart environments. The effectiveness of the framework is shown by applying it to the modeling of a real smart office scenario that has been developed, deployed and analyzed.     

Control for smart systems: Challenges and trends in smart cities

There have been tremendous developments in theories and technologies in control for smart systems. In this paper we review applications to various systems that are crucial for the future of smart cities, for example enterprise and manufacturing systems, transportation systems, energy systems, and data centres. Beyond discussing the existing technological trends and the methodological approaches developed so far for managing and controlling such systems, we also provide visions on the future challenges for these systems in these various aspects.     

A digital smart product service system and a case study of the mining industry: MSPSS

Mining equipment products and services no longer meet the needs of future development in the mining industry due to high safety and operational risk. The deep integration of the product-service system (PSS) and digitization is required in the mining industry to promote industry transformation and safe and efficient production without changing the traditional operation mode. This paper proposes a smart product-service system for the mining industry (MSPSS) consisting of a smart product subsystem, stakeholders, smart service subsystem, and smart decision-making subsystem. The analytic hierarchy process (AHP) and virtual reality (VR) are used for decision-making, product selection, operation, and maintenance. The smart product subsystem outputs reliable digital products using three stages: digital design, virtual simulation and planning, and virtual debugging. The smart service subsystem is driven by data and digital technology and provides fault diagnosis and online maintenance services for complex mining products. A case study indicates that all stakeholders can participate seamlessly in the design process. The smart product subsystem uses iterative optimization (more than 100 iterations) to obtain the design results interactively. The smart service subsystem provides digitalized services throughout the entire process. Thus, a stable, reliable, and comprehensive product and service solution is provided for complex mining conditions. The output is used to guide the design, debugging, and operation of physical equipment. The MSPSS has higher design quality and efficiency, a shorter design time, and lower design cost (key performance indicator (KPI)) than the traditional design method.     

Development of smart sensors system for machine fault diagnosis

Machine fault diagnosis is a traditional maintenance problem. In the past, the maintenance using tradition sensors is money-cost, which limits wide application in industry. To develop a cost-effective maintenance technique, this paper presents a novel research using smart sensor systems for machine fault diagnosis. In this paper, a smart sensors system is developed which acquires three types of signals involving vibration, current, and flux from induction motors. And then, support vector machine, linear discriminant analysis, k-nearest neighbors, and random forests algorithm are employed as classifiers for fault diagnosis. The parameters of these classifiers are optimized by using cross-validation method. The experimental results show that smart sensor system has the similar performance for applying in intelligent machine fault diagnosis with reduced product cost. Developed smart sensors have feasibility to apply for intelligent fault diagnosis.     

Modeling patterns of activities using activity curves

Pervasive computing offers an unprecedented opportunity to unobtrusively monitor behavior and use the large amount of collected data to perform analysis of activity-based behavioral patterns. In this paper, we introduce the notion of an activity curve, which represents an abstraction of an individual’s normal daily routine based on automatically-recognized activities. We propose methods to detect changes in behavioral routines by comparing activity curves and use these changes to analyze the possibility of changes in cognitive or physical health. We demonstrate our model and evaluate our change detection approach using a longitudinal smart home sensor dataset collected from 18 smart homes with older adult residents. Finally, we demonstrate how big data-based pervasive analytics such as activity curve-based change detection can be used to perform functional health assessment. Our evaluation indicates that correlations do exist between behavior and health changes and that these changes can be automatically detected using smart homes, machine learning, and big data-based pervasive analytics.     

A novel directional object detection method for piled objects using a hybrid region-based convolutional neural network

Digital transformation is an information technology (IT) process that integrates digital information with operating processes. Its introduction to the workplace can promote the development of progressively efficient manufacturing processes, accelerating competition in terms of speed and production capacity. Equipment combined with computer vision has begun to replace manpower in certain industries including manufacturing. However, current object detection methods are unable to identify the actual rotation angle of a specific grasped target while objects are piled. Hence this study proposes a framework based on deep learning that integrates two object detection models. Faster R-CNN (region based convolutional neural network) is utilized to search for the direction reference point of the target, and Mask R-CNN is adopted to obtain the segmentation that not only forms the basis of an area filter but also generates a rotated bounding box by minAreaRect function. After integrating the output from two models, the location and actual rotated angle of target can be obtained. The purpose of this research is to provide the robot arm with the position and angle information of the object located on the top for grasping. An empirical dataset of piled footwear insoles was employed to test the proposed method during the assembly process. Results show that the accuracy of the detection reached 96.26%. The implementation of proposed method in the manufacturing process not only can save man power who responsible for sorting out products but also reduce process time to enlarge production capacity. The proposed method can serve as a part of smart manufacturing system to enhance the enterprise’s competitiveness in the future.     

Achieving Knowledge-as-a-Service in IIoT-driven smart manufacturing: A crowdsourcing-based continuous enrichment method for Industrial Knowledge Graph

The rapid development of the industrial internet of things (IIoT) enlightened more tools and techniques to optimize the smart manufacturing service (SMS) paradigm, emphasizing the services based on pervasive IIoT products. However, the serviceability in the smart manufacturing environment is still limited since most of the conventional services are conducted in a passively reactive mode. The knowledge as a service (KaaS) model is hence to be introduced to emphasize the cognitive capability by actively utilizing massive knowledge, where industrial knowledge graph (IKG) plays as the core to generate context-awareness and proactive services for the optimization of serviceability and productivity. Meanwhile, as building IKGs for specific industrial cases is still plagued by the small scale and the lack of continuous enrichment, ensuring the quality and availability of the IKG are still challenging. Aiming to fill this gap with a practical and systematic approach, this paper proposes a generic crowdsourcing approach for continuously evolving the IKG. Through the IKG continuous enrichment approach, IKG-enabled systems indicate the higher value creation ability to utilize knowledge as a kind of service, rather than just a kind of resource. To further illustrate the proposed approach, a case study of a printed circuit board (PCB) processing machine is given with discussions. As an explorative study, future perspectives are also discussed to attract more open and in-depth studies for more robust applications of KaaS and IKG in the IIoT-driven smart manufacturing environment.     

Active learning strategy for smart soft sensor development under a small number of labeled data samples

This contribution proposes a new active learning strategy for smart soft sensor development. The main objective of the smart soft sensor is to opportunely collect labeled data samples in such a way as to minimize the error of the regression process while minimizing the number of labeled samples used, and thus to reduce the costs related to labeling training samples. Instead of randomly labeling data samples, the smart soft sensor only labels those data samples which can provide the most significant information for construction of the soft sensor. In this paper, without loss of generality, the smart soft sensor is built based on the widely used principal component regression model. For performance evaluation, an industrial case study is provided. Compared to the random sample labeling strategy, both accuracy and stability have been improved by the active learning strategy based smart soft sensor.     

The value of handhelds in smart environments

The severe resource restrictions of computer-augmented everyday artifacts imply substantial problems for the design of applications in smart environments. Some of these problems can be overcome by exploiting the resources, I/O interfaces, and computing capabilities of nearby mobile devices in an ad-hoc fashion. We identify the means by which smart objects can make use of handheld devices such as PDAs and mobile phones, and derive the following major roles of handhelds in smart environments: (1) mobile infrastructure access point; (2) user interface; (3) remote sensor; (4) mobile storage medium; (5) remote resource provider; and (6) weak user identifier. We present concrete applications that illustrate these roles, and describe how handhelds can serve as mobile mediators between computer-augmented everyday artifacts, their users, and background infrastructure services. The presented applications include a remote interaction scenario, a smart medicine cabinet, and an inventory monitoring application.     

Taxonomy,technology and applications of smart objects

Deployment of embedded technologies is increasingly being examined in industrial supply chains as a means for improving efficiency through greater control over purchase orders, inventory and product related information. Central to this development has been the advent of technologies such as bar codes, Radio Frequency Identification (RFID) systems, and wireless sensors which when attached to a product, form part of the product’s embedded systems infrastructure. The increasing integration of these technologies dramatically contributes to the evolving notion of a “smart product”, a product which is capable of incorporating itself into both physical and information environments. The future of this revolution in objects equipped with smart embedded technologies is one in which objects can not only identify themselves, but can also sense and store their condition, communicate with other objects and distributed infrastructures, and take decisions related to managing their life cycle. The object can essentially “plug” itself into a compatible systems infrastructure owned by different partners in a supply chain. However, as in any development process that will involve more than one end user, the establishment of a common foundation and understanding is essential for interoperability, efficient communication among involved parties and for developing novel applications. In this paper, we contribute to creating that common ground by providing a characterization to aid the specification and construction of “smart objects” and their underlying technologies. Furthermore, our work provides an extensive set of examples and potential applications of different categories of smart objects.