Digital Twin Modelling for Automation, Maintenance and Monitoring in Industry 4.0 Smart Factory

Goals and Objectives

  1. Produce state of the art digital twin model to work with Siemens/Festo Industry 4.0 Cyber physical facility for thorough evaluation, debugging and optimization of applications
  2. Use proposed digital twin to suggest and counteract delay inducing elements in safety, preventive maintenance and regulatory systems in Industry 4.0
  3. Digitizing manufacturing processes in industry 4.0 for optimal efficiency, including detecting and solving physical issues faster, predicting outcomes to a much higher degree of accuracy, scheduling activities in the most efficient and cost-conscious way

Brief Description of the Project

The rapid advancements in manufacturing technologies and industry transformation in 4th Industrial Revolution requires more sophisticated tools to enable high productivity, lower running costs, product quality improvement, minimized maintenance and shutdown. In Industry 4.0, fully automated smart industrial infrastructure relies on low latency feedback networks, high efficiency distributed control systems, fool-proof emergency and safety systems, energy efficient and self-sustaining processes and supportive digital technologies.

The existing industrial systems are highly complex and require several processes to operate simultaneously to achieve the desired objectives. To ensure efficient operations within industrial processes, human intelligence, intervention and feedback is widely used. To enable truly self-reliant and autonomous industries, the developments are on the way. One major hurdle in achieving fully autonomous industries is lack of software-based counterparts to support vigorous testing.

This project targets implementation of digital counterpart (a Digital Twin model) of Industry 4.0 to replicate its functionalities, data, communications, feedback, emergency and safety aspects. The proposed digital twin for industry 4.0 will not only offer a digitized replication of functionalities but will also enable development towards self-correcting smart process control facility. The digital twin will also facilitate debugging, testing and reforming processes. It is expected that the developments in the project will provide solutions for some of the most critical aspects of the present-day industries. The developments in this project will be cross-validated and vigorously tested in state of the art Siemens/Festo cyber factory facility installed at Middlesex University (MDX), which acts as the physical twin in the project.

The key research question that will be addressed is how intelligently digital twin can predict the chain of events triggered as a consequence of certain variations in some processes, within the hundred plus manufacturing industries in and around Sricity/Andhra-Pradesh.

Scientific & Technical Details

Industry 4.0 aims to offer next generation of industrial automation which emphasises on interconnected and decentralized intelligent systems, capable of self-sustaining. However, the complexity of smart industrial processes is unfathomable, given the interjection of countless smart processes, which need to work seamlessly perfect to achieve the desired outcomes. For such interconnected systems, the impact of changes in one process is hard to predict.

The use of digital twin encompasses the functionality and interconnection of different processes within the industry and bears the potential to replicate interlinked complex processes in digital domain. This can provide a framework to investigate experimental setup in the simulations with more confidence. It also offers a platform to evaluate system limitations and impact of malfunction in one process on the others. The digital model of industry 4.0 will provide limitless opportunities to observe the impact of failure in one small block and how it will impact the entire setup. It will also enable the development of backup solutions to deal with the arising situation. Notably, the IoT and Analytics are required for real-time data collection, analysis and decision making which are crucial for the proper operation of the Cyber Physical System (CPS). The interaction of the IoT-based smart objects within the CPS will generate large amounts of data needs to be processed for extracting valuable and timely information.

This project targets implementation of digital counterpart of industry 4.0 to replicate its functionalities, data, communications, feedback, emergency and safety aspects. The project will develop digital twin to mirror the smart cyber factory facility at Middlesex University supplied by Festo/Siemens which comprises a comprehensive six-station table top unit (two production cells of three stations), as well as two bridging stations that enable an Automated Guided Vehicle (AGV) to deliver the logistics/transport between the cells. The validation for the Digital Twin will focus on the following aspects: i) energy monitoring; ii) tracking components and goods by means of tags which transmit a radio signal using Radio Frequency identification (RFID); iii) digital maintenance; iv) augmented reality of a real-world manufacturing process; v) direct communication among the objects using near field communication (i.e., objects equipped with a chip to exchange information directly); and vi) manufacturing execution system. The primary objectives are to develop a self-correcting smart process control facility where the digital twin can extend the debugging, testing and reforming processes before physical implementation. It is expected that the developments in the project will provide solutions for some of the most critical aspects of the present-day industries. The project will also aim to minimize the sensing, communications and processing delays for such applications. It will also target regulatory control applications within industry 4.0 to improve the overall efficiency of the plant/factory. Since the effective operation of regulatory control in industries require feedback response within a fixed time window for optimal process efficiency, therefore, the digital twin will serve as a digital alternate to predict any expected variations in regulatory control systems’ delays.
The project is divided into four work packages (WP):

  • WP1 – Data collection and analytics (IIIT, MDX, SPL): The IoT-enabled CPS rely on the underlying large-scale and dynamic network infrastructure which enables the communication and coordination among the connected objects. However, the use of these IoT-enabled CPS produces large amounts of fine-grained data that needs to be processed. Thus, one of the biggest challenges is analyzing this data and creating timely and relevant information out of it. Statistics of data (e.g., fan speeds, humidity, temperature, energy) can be used to help optimize machinery’s operation and reduce the risk of outage. In this WP, we will utilise the MindSphere platform provided by partner Siemens to analyze the data in real time and provide classified groups of data for the development of a Digital Twin model in WP2. Partner IIIT and SPL lead the data analytics tasks while MDX will coordinate the data collection process.
  • WP2 – Development of the digital twin model (MDX, IIIT, Festo, Siemens): In case of lack of data diversity for training, we will simply fall back on standard machine learning (ML) procedures (data diversity is traditionally a hard ML problem). For low data richness (ie., small/medium data rather than big data), we propose to employ the Kernelised support vector approaches which are very well attuned to low data richness conditions, behaving far better than deep learning approaches under these circumstances. Among the machine learning algorithms, auto-associative neural network (AANN) is a widely used and effective tool for physical issue detection. The key idea of AANN is to learn the interrelation between system responses and issues, and later to use the learning capacity to evaluate the new cyber system pattern, when issues appears.
  • WP3 – Physical based model and statistical data approach (IIIT, MDX, SPL): In this WP, for fine-tuning the Digital Twin, we will focus on predictive maintenance, and particularly detecting the physical issues (e.g., wrong reading of energy consumption, wrong tracking of objects, disconnected communication among components). Finite element model updating as well as optimization techniques will be used to detect the issues in the physical objects. In model updating, the differences between the numerical results and experimental responses are minimized as a function of updating parameters in order to localize and quantify issues. Recently, advanced optimization techniques, such as Genetic Algorithm and other metaheuristic algorithms, have been used for the purpose of physical issues detection. These algorithms will serve as tools for evaluation of interactions between two models (Digital and Physical). On the other hand, statistical model-based methodologies concentrate on pattern recognition algorithms to discriminate the experimental measurements of the responses under intact condition and the responses under abnormal condition.
  • WP4: Testing and trial (all partners): Validate digital twin via interactions in real-world conditions; analyse the outcomes and assess further the correlation between two models (physical and digital), and adjust the Digital Twin model accordingly.


UK Lead Partner

Middlesex University, London, UK (Huan X. Nguyen, UK lead Applicant)

India Lead Partner

Indian Institute of Information Technology (IIIT) Sricity (Prof. Hrishikesh Venkataraman, India Lead Applicant)

Other Partners

Sricity Private Limited (SPL)

Festo Didactic Ltd