Chapter 1: Introduction to Industry 4.0 Technologies Free

The Fourth Industrial Revolution, also known as Industry 4.0, refers to the rapid digital transformation of industrial processes. Beyond the changes in industrial processes, the technological revolution is poised to drastically alter the way humans work, live and interact. Business models across various industries are also being disrupted by the revolution. Industry 4.0 is being driven by various emerging techonlogies including but not limited to the Internet of Things technology (IoT), artificial intelligence, cloud computing, augmented reality, robotics, rapid prototyping, and blockchain. The integration of some of these technologies facilitate the seamless linking of the physical and digital worlds, giving rise to cyber-physical systems. The above-mentioned technologies are briefly introduced in the following sections.

IoT describes the interconnection of ordinary objects over the internet with the ability to communicate with one another. IoT creates a wired or wireless communication link between physical objects with embedded sensors and actuators (Madakam et al., 2015). As shown in Fig. 1.1, in general, IoT can be divided into three layers (Wu et al., 2010; Yang, 2011; Khan et al., 2012; Cheng and Lim, 2021; and Rejeb et al., 2022).

The Perception Layer is the physical layer that is responsible for sensing, gathering, and processing information from the physical environment (Sethi and Sarangi, 2017; Martín-Lopo et al., 2020; Lombardi et al., 2021; and Rejeb et al., 2022). The layer encompasses devices equipped with sensors capable of sensing environmental variables or other nearby smart devices (Sethi and Sarangi, 2017). These devices can be everyday devices such as television sets and refrigerators (Lombardi et al., 2021). The perception layer is the lowest layer of the IoT architecture (Alavikia and Shabro, 2022).

This layer concerns the transmission of data gathered at the perception layer to the appropriate endpoints (Bansal and Kumar, 2020; and Rejeb et al., 2022). It encompasses all the communication protocols and technologies that make this connection possible (Abdmeziem et al., 2016). Connectivity with other servers and network devices is provided at the network layer (Sethi and Sarangi, 2017). This layer is the core of IoT (Chanal and Kakkasageri, 2020).

The application layer is the user-facing layer of an IoT system. It is at the top of the IoT architecture and facilitates the delivery of application-specific services to users (Rejeb et al., 2022). In an IoT-based condition monitoring system, for instance, the application layer constitutes the running of the application software at the monitoring centre (Alavikia and Shabro, 2022). Data analytics on data received from the perception layer also comes under the application layer (Alavikia and Shabro, 2022). There are a plethora of applications of IoT in the energy sector (Makhanya et al., 2019; and Onibonoje et al., 2019a, 2019b).

Machine learning is a subdivision of artificial intelligence that deals with the idea of software learning from datasets using statistical techniques. It involves the development of mathematical models that learn from data and autonomously improve their performance (Alpaydin, 2020; and Murphy, 2022). Through the learning process, computers are trained to perform tasks like intelligent beings (Ibrahim et al., 2020). Such tasks include making decisions and/or predictions (Hossain et al., 2019). Figure 1.2 shows four broad categories of machine learning (Belfin et al., 2020):

This type of machine learning involves training algorithms with labelled data, such that they learn to predict or classify data. The goal is to learn the relationship between input and output pairs of labelled sample data (Murphy, 2012). The regression and classification problems can be handled by supervised learning (Tien et al., 2022).

Unlike supervised learning, the training data set for unsupervised learning is neither classified nor labelled; the machine learning algorithm categorizes data based on differences or similarities (Hinton and Sejnowski, 1999). These learning algorithms can handle more complex tasks but can be unpredictable (Ibrahim et al., 2020). Unsupervised learning algorithms extract features and patterns from datasets to make sense of the data (Tien et al., 2022). They find applications when labelled data are unavailable or the target is unknown.

This learning approach combines a small set of labelled data and large unlabelled datasets (Ahfock and McLachlan, 2022). This comes in handy when labelled data, which can be expensive to acquire, are in limited supply. Semi-supervised learning algorithms are suitable for tackling classification and clustering problems (van Engelen and Hoos, 2020).

Under this learning paradigm, agents learn by a trial-and-error interaction with their environment. Proper behavior attracts rewards, while bad actions incur penalties. Hence, agents can adjust their actions in response to feedback (Lewis and Liu, 2013). Since agents can autonomously determine their behaviour, reinforcement learning closely resembles the learning process of humans (Han et al., 2019).

The massive amount of data generated by sensors and meters in a smart grid will only be useful if efficient machine learning techniques are applied to extract insights needed for the proper operation of the grid. Within the energy sector, machine learning finds application in consumption forecasting, electricity cost prediction, renewable generation forecasting, future optimum schedules, network intrusion detection, system sizing, adaptive control, and fault detection (Mellit et al., 2009; Negnevitsky et al., 2009; Kusiak and Zhang, 2011; Wenyi et al., 2013; Frincu et al., 2014; Liu et al., 2015; Dickson, 2016; and Esmalifalak et al., 2017). It can be used on historical data collected by sensors and energy meters to predict future energy usage in buildings (Nwulu and Fahrioglu, 2011; Nwulu, 2017; Ntsaluba et al., 2021; and Tien et al., 2022). Besides its application in the smart grid, it can be applied to individual grid systems (Gbadamosi and Nwulu, 2019). For instance, in the solar energy industry, fault detection and performance prediction are two popular application areas of machine learning (Sohani et al., 2022). In addition to a reduction of degradation in PV technologies, machine learning techniques can improve their operating times by effective fault detection (Sohani et al., 2022).

Cloud computing facilitates the on-demand use of remote computer resources over the internet. It enables individuals and organizations to achieve computing tasks using resources they do not possess. Although cloud service providers usually provide this service at a cost, cloud computing remains a preferred option compared to acquiring, operating and maintaining all the IT infrastructure needed to perform all computing tasks. For instance, an organization that only occasionally needs high computing power needs not purchase expensive, sophisticated computers that can be accessed via a cloud computing platform. Remote storage is also made possible by cloud computing. The cloud is an asset to modern smart grids in the electricity sector. For instance, customers' consumption data can be aggregated to the cloud, from which demand-side management would be performed to determine optimal appliance schedules (Yaghmaee et al., 2017). The authors of Rajeev and Ashok (2011) have also shown that the cloud environment is relevant for energy management in microgrids. Due to the scalability of cloud storage and computing, it is well suited for information processing in smart grids (Fang et al., 2013). Cloud computing also offers smart grids better security capabilities than traditional IP-based security approaches (Bera et al., 2015) (Fig. 1.3).

According to the National Institute of Standards and Technology, U.S. Department of Commerce (Mell and Grance, 2011), the essential characteristics of a cloud model are:

• Pooled resources: A cloud provider combines computing resources that may not be physically close to serve various users by employing a multi-tenant model. Generally, users do not know the location of the resources being utilized and, to some extent, have no control over that.

• Wide network access: Using heterogeneous client platforms such as personal computers, tablets, and mobile phones, users can access the various computing capabilities offered by the providers over the network.

• Automatic self-service: Without human involvement on the cloud provider's side, users can automatically access computing resources to accomplish tasks such as network storage and server time as and when required.

• Metered service: Cloud service providers can control, monitor, and report resource usage. They typically perform metering on a charge-per-use or pay-per-use basis.

• Prompt elastic resources: Cloud-based resources can be quickly ramped up or scaled down in response to demand, so users never experience any form of restriction regarding resource usage.

Figure 1.4 shows a pictorial summary of these essential characteristics.

The following are the various deployment models for cloud computing (Mell and Grance, 2011):

• Private Cloud

  A private cloud is an infrastructure dedicated solely to a single organization, which may not be responsible for its operation and management. It may or may not be owned by the organization.

• Community Cloud

  A cloud infrastructure dedicated for use by a community of users or organizations that share similar policies, missions, security requirements, etc., is known as a community cloud. One or more of these organizations may own, operate, and manage the infrastructure, which may or may not be located within its premises. A third party could also own it.

• Public Cloud

  A public cloud is accessible to the general public but is located within the cloud provider's premises. It is owned by an institution such as a government, business, or academic organization.

• Hybrid Cloud

  An aggregation of two or more private, public, or community cloud infrastructures that remain distinctive entities but have a connection that facilitates software and data portability is known as a hybrid cloud.

The following are three core technologies that enable cloud computing (Marston et al., 2011):

• Virtualization

  In cloud computing, multiple users can simultaneously access the cloud—which may be a single computing resource—with the help of virtualization. Figure 1.5 shows a cloud's virtualization. Virtualization creates multiple virtual versions of a computing resource, each of which appears to users as the actual computing resource. This computing resource may be a network, storage, application stack, memory, or database (Xing and Zhan, 2012). With virtualization, multiple operating systems, each running on a virtual version of the computing resource share the actual computing resource.

• Multi-tenancy

  A multi-tenancy architecture enables a single instance of a software or hardware component to be used simultaneously by multiple users. While users use the same resource, personal data are, in theory, not accessible by other users.

• Web Services

  According to the World Wide Web Consortium (W3C), interoperable device-to-device communication over a network can be facilitated by a software system known as a web service. With the help of web services, a client software such as a browser can access applications on a server via a network.

Cloud service providers usually provide cloud computing services at a cost. Three pricing models exist for the different cloud services (Youseff et al., 2008):

• Tiered pricing

  This pricing model has various tiers, each offering predetermined computing specifications and service-level agreements at a fixed cost per unit time. Amazon's cloud services are offered using this pricing scheme.

• Per-unit pricing

  This pricing scheme is usually applied to data transfer or memory usage. It allows users to adjust their system's main memory allocation based on their needs.

• Subscription-based pricing

  The software as a Service (SaaS) model usually employs this pricing model. In the model, users are not charged for their actual resource usage; however, they are expected to predict the periodic cost of cloud service usage.

The services offered by cloud providers can be classified into three main service models, which are as follows (Mell and Grance, 2011) (Fig. 1.6):

• Software as a Service (SaaS)

  A SaaS cloud service provider makes applications available to subscribers. The applications are installed and run on the cloud infrastructure but can be accessed by users from their personal computers, tablets and phones. Therefore, users do not need to install the application. They also do not manage the other components of the cloud infrastructure, such as operating systems, networks, storage and servers. With SaaS, individuals and businesses can utilize expensive software while avoiding the high cost of such. Google docs and GitHub are examples of SaaS.

• Platform as a Service (PaaS)

  PaaS cloud service providers offer environments on the cloud infrastructure where users can deploy user-owned applications. The management of the cloud operating systems, network, storage, and server remains the responsibility of the cloud provider, not the user. Applications to be deployed should be developed with tools supported by the cloud provider. PaaS offers software developers a cost-effective way to build applications without acquiring sophisticated infrastructure or setting up the required environment.

• Infrastructure as a Service (IaaS)

  IaaS cloud service providers offer fundamental computing resources such as networks, processing power and storage capacity. Although a subscriber can deploy and run the software and operating systems on the cloud infrastructure and control them, the management of the underlying cloud infrastructure remains the responsibility of the cloud provider. IaaS offers more accessibility and flexibility than the other service models.

Figure 1.7 describes a concept termed “virtuality continuum” by Milgram and Kishino (1994). On the extreme left of the continuum is the real environment where only real objects exist, while the other end is exclusively for virtual, computer-generated objects. In between are augmented reality and augmented virtuality. Augmented reality offers users an enhanced view of the real world by superimposing dynamic digital objects/information on their immediate environment. Unlike virtual reality, which creates and presents a virtual environment as if it were real, augmented reality incorporates virtual objects into a real environment (de Souza Cardoso et al., 2020). It bridges the real and digital worlds (Timchenko et al., 2020). Augmented reality is real-time interactive and registered in three dimensions (Azuma, 1997; and Reitmayr and Schmalstieg, 2003). This technology can be implemented using smart glasses or camera-equipped smartphones (João et al., 2020). Google Glass is an example of augmented reality in practice (He et al., 2017).

The technology finds application in many areas such as the military, education, entertainment, and medicine (Syarofi and Sibarani, 2020). Field technicians, for instance, can be guided step-by-step while maintaining, repairing, or assembling machine parts with the aid of augmented reality. Using smart glasses, they can view instructions while executing critical tasks (Gehrke et al., 2015). Augment reality can also be used to create a more interactive training environment that facilitates efficient staff upskilling. Useful information about equipment and other assets can be digitally overlaid on them with augmented reality (João et al., 2020). Augmented reality systems comprise four main categories of devices, namely (Carmigniani et al., 2011): Input devices, computers, tracking devices, and displays. Various input devices are available, such as touch screens of smartphones, wireless wristbands, and gloves. Trackers or sensors include accelerometers, global positioning systems, and wireless sensors. Laptops and desktop personal computers may be used to accomplish the necessary processing tasks of AR systems. Displays are the devices which deliver augmented reality imagery to the users. According to the authors of Azuma et al. (2001), an augmented reality environment may be viewed with the aid of three main categories of displays.

A computer-generated image or information is projected onto a real object with projected displays. The user views the imagery as a whole without the need to wear or hold any display device. Figure 1.8 shows a physical object upon which a projector overlays various colors. The augmented image can then be viewed as a real, colored object.

These display gadgets are worn around the head with the augmented imagery close to the user's eyes. The imagery is composed of a real-world environment overlaid with computer-generated content. Figure 1.9 shows a head-worn display device. With this technology, there exist two ways of viewing the real-world part of the augmented imagery. It may be viewed directly via a half-transparent mirror.

On the other hand, a real-world video captured by cameras may be viewed on an opaque mirror (Rolland and Fuchs, 2000). The former is an optical see-through head-mounted display, and the latter is a video see-through head-mounted display. Figure 1.10 shows a pictorial description of both optical display methods (Azuma et al., 2001).

These are portable devices that the user holds with the hand in the direction of interest to view augmented reality imagery. Camera-enabled smartphones are commonly used hand-held displays, with the video see-through display technique being the naturally adopted option. A limitation of the hand-held display technology is the restriction imposed on the user who has to hold the device with at least one hand. A field technician, for example, would find it difficult to work efficiently on a piece of equipment using a hand-held display. The portability of hand-held display devices also evidently restrains their battery capacity; however, with the advances in battery technology, this limitation will become less pronounced. Figure 1.11 shows a hand-held device being focused on a shelf to obtain computer-generated product information. The see-through video technology is being used in this case. The camera is focused on the shelf, and the user clicks on any product of interest to view more information about the product.

Robots are programmed machines capable of performing tasks in the physical world (Patel et al., 2018). They outperform humans at repetitive tasks with high precision. Today, highly sophisticated robots have been developed to perform more complicated tasks requiring some intelligence. Contemporary programmable industrial robots that are programmed to carry out certain repetitive tasks are increasingly being adopted in various applications and industries (Artuc et al., 2018). Figure 1.12 shows an upward trend in the installation of industrial robots. Figure 1.13 presents the estimated worldwide year-end supply of industrial robots by industries. While Fig. 1.14 shows the turnover from professional service robots, Fig. 1.15 depicts the number of units of sales of personal/domestic service robots. Robots have successfully replaced many human operators in many manufacturing processes, saving costs, increasing speed and improving performance. In the solar energy industry, they are particularly well suited to assemble solar panels and handle delicate materials like solar cells and silicon wafers (Iqbal et al., 2019). In place of human operators, they can also be employed to carry out inspections, repairs and maintenance of equipment. They are preferred to humans for deployment in high-risk sites. Robots can be used to refurbish and fabricate equipment in the hydro energy industry (Iqbal et al., 2019). Electricity distribution companies can also benefit from the advances in robotics. In Byambasuren et al. (2016), the use of a robot to detect illegal usage of electricity was investigated. The adoption of robots in the maintenance of electric grid infrastructure can improve energy efficiency (Nagarajan et al., 2019).

There exist various possible subdivisions of the entire system of a robot. One subdivision categorizes a robot’s system into mechanical, information, and Electrical/Electronics subsystems, as shown in Fig. 1.16. The mechanical subsystem consists mainly of the structures or mechanisms responsible for a robot's movement, such as actuators, transmission, joints, and links. Power supplies, electrical/electronic cables, and most other components that use or transmit electrical power constitute the Electrical/Electronic subsystem of a robot's general structure. The information subsystem includes software required for the control and other actions of the robot. As shown in Fig. 1.16, these subsystems are not distinctly separable as some components can be seen to extend beyond a single subsystem (Fig. 1.17).

Rapid prototyping describes the quick and automated fabrication of complex product models from three-dimensional computer-aided design using a layer-by-layer deposition approach (Udroiu et al., 2011). In the product development process, product ideas can be developed for testing, validation, and even user feedback before the final market-ready product is produced. With rapid prototyping and potential user engagement, a new product's commercial viability can be assessed economically (Bryant et al., 2020). Moreover, this avoids the possible losses associated with a defective final product. An idea that can be quickly fabricated and tested also encourages creativity and innovation. Bryant et al. (2020) presented a study which involved the rapid prototyping of a new battery technology and its packaging. Also, in the energy storage industry, the rapid prototyping of electrolyzer flow field helps to test different plate configurations. This, in turn, paves the way for developing high-performance electrolyzers (Hudkins et al., 2016). Multiple iterations of a small wind turbine blade design were fabricated, as reported in (Poole and Phillips, 2015), using rapid prototyping. Udroiu et al. (2011) have also reported the rapid prototyping of a small experimental Pelton turbine for hydroplants.

Various types of rapid prototyping technologies are described below (Dudek and Rapacz-Kmita, 2016):

• Stereolithography

  Stereolithography is an old additive manufacturing technique that uses laser light to create three-dimensional objects by solidifying a light-sensitive liquid resin layer-by-layer (Dudek and Rapacz-Kmita, 2016). As shown in Fig. 1.18, the liquid is held in a container that also holds a platform which supports the moulded object. The laser beam operates in the X–Y plane while the platform moves in the Z direction.

• Laminated object manufacturing

  In this technique, layers of materials are bonded together and cut as required using a blade or a laser cutter. Each layer is cut such that the portion of it that constitutes a part of the final object is kept intact while the rest is cross-sliced to facilitate easy removal from the final object. The heated roller rolls over each layer to melt its adhesive coating and bind it with an adjacent layer (Fig. 1.19).

• Three-dimensional printing

  This method makes use of materials in powder form. A layer of powder is spread over a platform, and a polymer binder is used to hold the particles of powder that make up the geometry of the cross-section of that layer of the final object. This process is repeated until the final product is obtained (Fig. 1.20).

• Fused deposition modelling

  In fused deposition modelling, an extrusion nozzle fixed to a printing head receives the supply of a thermoplastic filament. In a layer-by-layer approach, the plastic is melted and extruded onto a building platform according to the geometry of each layer of the final object, after which it solidifies and bonds with the preceding layer (Fig. 1.21).

• Selective laser sintering

  In this rapid prototyping method, a powder-form material is rolled onto a powder bed and sintered layer-by-layer according to the geometry of each layer. Sintering is achieved using heat from laser beams. The method facilitates manufacturing complex three-dimensional parts. It is frequently used in the automotive, aerospace, and medical fields (Fig. 1.22).

A blockchain is an immutable, distributed database of peer-to-peer transactions. It facilitates the transfer of value between peers on a network without a central controlling entity. Unlike centralized systems, which are susceptible to cyber attacks, in blockchains, a consensus on the system state can be reached even in the face of an attack (Münsing et al., 2017). It is well-suited for transactions within networks with untrusted parties. With blockchains, the security of transactions is ensured using cryptography and consensus mechanisms. It can also be used as a tamper-proof database for storing data. Blockchains are also being used to store and execute the terms of contracts between parties. The study presented in Münsing et al. (2017) explores the use of blockchains to facilitate the control and distributed optimization of energy resources as well as fair payments in a microgrid. In the electricity industry, the role of end consumers has been enhanced by the peer-to-peer transactions the technology enables (Diestelmeier, 2019). The low cost and high computing capacity offered by blockchain is one of its selling points in energy trading systems (Wang et al., 2019). The technology is being widely adopted to facilitate energy trading (Guo et al., 2020). Blockchain-based smart contracts have been explored for peer-to-peer energy trading, energy storage sharing, and thermal plant fuel trading in (Damisa and Nwulu, 2022; Damisa et al., 2022a, 2022b).

The book's first section briefly introduces some concepts and technologies associated with the ongoing digital transformation in manufacturing and production practices, also known as the fourth industrial revolution (Chaps. 1–3). It is followed by a section focusing on the key concepts around the revolution in the energy industry (Chaps. 4–5) and a third section, which details crucial applications that would facilitate the ongoing energy sector revolution (Chaps. 6–11). The book concludes with a discussion on the future of Energy 4.0 in terms of technologies, services, business models, and policy and regulatory frameworks (Chap. 12).

Some concepts and technologies connected to the fourth industrial revolution have been briefly introduced in this chapter. These technologies are also relevant to the energy industry's ongoing revolution—Energy 4.0. The rapid increase in data available to utilities from smart electricity meters paves the way for harnessing data science tools. Cloud computing will come in handy for complex computations required for the optimal performance of the modern electricity grid. With augmented reality, a field technician can be guided in the assembly or repair of a power grid component. Robots can replace underwater field technicians to avoid the possibility of life loss. A new wind turbine blade design sample can be quickly developed for testing and analysis using rapid prototyping technologies. Blockchain can facilitate secure peer-to-peer trading of distributed energy resources. The fusion of these technologies is driving the energy industry's revolution.