Artificial Intelligence for Smart Manufacturing

This chapter provides an overview of Artificial Intelligence-based methods, applications, and challenges for Smart Manufacturing in Industry 5.0. We elaborate on the essential issues related to the applications and the potential of Artificial Intelligence algorithms in Smart Manufacturing. We will introduce crucial topics that will be discussed in the following chapters of the book.

As the fourth industrial revolution has passed an early stage of development, many companies are developing intelligent systems and cutting-edge innovations of Industry 4.0 to improve productivity and quality. Meanwhile, the next phase of industrialization has been started to introduce, known as Industry 5.0. One of the most prominent features of Industry 5.0 is that it places the well-being of humans at the center of the manufacturing process. Advanced technologies are also created to keep up with the trend of the fifth industrial revolution. Artificial intelligence (AI) algorithms have proven to play a key role in Industry 4.0. Moving to Industry 5.0, with the human-centric orientation, AI was developed in combination with human intelligence (HI), leading to the new concept of Augmented Intelligence (AuI). AI and AuI algorithms are expected to bring significant benefits for enabling smart manufacturing in Industry 5.0. In this study, we provide a survey on AI-based methods, applications, and challenges for smart manufacturing in Industry 5.0. The discussions will help to clarify some important issues related to the applications and the potential of AI algorithms in smart manufacturing.

Smart manufacturing is widely accepted as the new emerging transformation of the manufacturing industry today. In addition, quality control, an important aspect that contributes to the successful process of smart manufacturing attracts attention from the community. However, there are certain challenges in implementing quality control methods in Industry 5.0. Thus, this chapter aims to provide a comprehensive background review of important notions and advanced techniques related to quality control for smart manufacturing such as Machine Learning, computer vision, the Internet of Things, and Artificial Intelligence. Then, several difficulties and opportunities in the implementation of these techniques for quality control in Industry 5.0 are discussed. Finally, a case study on monitoring wine production in the food industry is also considered to show the performance of Machine Learning-based techniques for quality control.

Machine learning methods have been widely used in different applications, including process control and monitoring. For handling statistical process control (SPC) problems, the existing machine learning approaches have some limitations. For instance, most of them are designed for cases in which in-control (IC) process observations at different time points are assumed to be independent and identically distributed. In practice, however, serial correlation almost always exists in the observed sequential data, and the longitudinal pattern of the process to monitor could be dynamic in the sense that its IC distribution would change over time (e.g., seasonality). It has been well demonstrated in the literature that control charts could be unreliable to use when their model assumptions are invalid. In this chapter, we modified some representative existing machine learning control charts using nonparametric longitudinal modeling and sequential data decorrelation algorithms. The modified machine learning control charts can well accommodate time-varying IC process distribution and serial data correlation. Numerical studies show that their performance are improved substantially for monitoring different dynamic processes.

Equipment monitoring and process fault prediction are increasingly concerned in the modern industry due to the growing complexity of the production process and the high risk derived from severe consequences on the paper mills in case of production failure. Whereas the paper manufacturing process is continuous that is difficult to be warned early of faults. To address such issues, this Chapter proposes a data-driven approach to predict fault in the papermaking process on the basis of correlation analysis and clustering algorithms. Historical operating data of key variables were acquired in normal operating conditions. The health benchmark dataset was constructed based on the Gaussian mixture model (GMM) and Mahalanobis distance (MD) to evaluate the operating status of the papermaking process. The verification results showed that the proposed model has a fault prediction accuracy of 76.8% and a recall rate of 72.5%, which allows anomalous data to be observed in advance, providing valuable time for subsequent fault diagnosis.

With the complexity of the production process, the mass quantification of production jobs, and the diversification of production scenarios, research on scheduling problems are bound to develop in a direction closer to the actual production problems. Considering the combination of workshop scheduling problems and process planning problems, the study of such problems is of great significance for improving the production efficiency of enterprises. Therefore, this chapter studies the intelligent scheduling problem of a flexible flow-shop and establishes a two-stage flexible flow-shop scheduling model. On this basis, the fast non-dominated sorting genetic algorithm II (NSGA-II) and the variable neighborhood search algorithm (VNS) are combined to optimize the established two-stage intelligent scheduling model. Finally, a papermaking production process is taken as an example to comprehensively evaluate the performance of the model and the hybrid intelligent algorithm. The experimental results show that the model and algorithm can effectively solve the presented problem.

Commercial garment recommendation systems have been generally used in the apparel industry. However, existing research on digital garment design has focused on the technical development of the virtual design process, with little knowledge of traditional designers. The fit of a garment has a significant role in whether a customer purchases that garment. In order to develop a well-fitting garment, designers and pattern makers should adjust the garment pattern several times until the customer is satisfied. Currently, there are three main drawbacks of traditional pattern-making: (1) it is very time-consuming and inefficient, (2) it relies too much on experienced designers, and (3) the relationship between the human body shape and the garment is not fully explored. In practice, the designer plays a key role in a successful design process. There is a need to integrate the designer’s knowledge and experience into current garment CAD systems to provide a feasible human-centered, low-cost design solution quickly for each personalized requirement. Also, data-based services such as recommendation systems, body shape classification, 3D body modeling, and garment fit assessment should be integrated into the apparel CAD system to improve the efficiency of the design process. Based on the above issues, a fit-oriented garment pattern intelligent recommendation system is possible for supporting the design of personalized garment products. The system works in combination with a newly developed design process, i.e. body shape identification—design solution recommendation—3D virtual presentation and evaluation—design parameter adjustment. This process can be repeated until the user is satisfied. The proposed recommendation system has been validated by some successful practical design cases.

Artificial Intelligence (AI) and especially Machine Learning (ML) are the driving energy behind industrial and technological transformation. With the transition from industry 4.0 to 5.0, smart manufacturing proves the efficiency in industry, where systems become increasingly complex, producing massive data, necessitating more demand for transparency, privacy, and performance. Federated learning has demonstrated its effectiveness in various applications, however, there are still exist certain challenges that should be addressed. Thus, in this chapter, a comprehensive perspective on federated learning-based anomaly detection is provided. The problems have posed concerns and should be taken into account when researching and deploying. Then, our perspectives about efficient and trustworthy federated learning-based explainable anomaly detection systems are demonstrated as an end-to-end unified framework. Finally, to provide a complete picture of future research direction, the quantum aspect is introduced in the subject of machine learning.

Prognostics and health management (PHM) is a crucial enabler to reduce maintenance costs and enhance the availability and reliability of manufacturing systems. In the context of Industry 4.0, these systems become more complex and can be monitored by different types of sensors. The quality and completeness of data are crucial factors for the success of any PHM task in this paradigm. Here, we investigate the possibility of exploiting additional data sources in manufacturing besides monitoring sensors, e.g. production line cameras or maintenance reports. We first present the terminologies of multimodal learning and the potential it holds for industrial PHM. We then further explore the development and notable works in this field applied to other domains, look at the relevant works in PHM, and finally present a case study to demonstrate how multimodal learning can be performed to improve PHM processes.

Industry 4.0 was first presented in 2011 and has revolutionized manufacturing in enormous applications by integrating artificial intelligence (AI), the Internet of Things (IoT), cloud computing, and other leading technologies. As technology continues to grow and expand, the concept of a new Industry 5.0 paradigm could be investigated. Industry 5.0 aims to transform the manufacturing sector into a more sustainable, human-centric, and resilient manufacturing industry. In this chapter, we demonstrate research for Cybersecurity in Smart Manufacturing in Industry 5.0 by leveraging AI and Explainable Artificial Intelligence (XAI) techniques. This chapter especially presents several essential perspectives for a potential approach of XAI to enable Smart manufacturing in the Industrial Revolution 5.0. There also is an illustrative example demonstrating the XAI approach for anomaly detection in the cyber network of an Industrial Control System in the Smart Manufacturing context.

The innovation of wearable Internet of Things devices has fuelled the transition from Industry 4.0 to Industry 5.0. Increasing resource efficiency, safety, and economic efficiency are some of the main goals of Industry 5.0. Herein, wearable Internet of Things devices is parallel to humans to optimize human tasks and meet a new Industry’s requirements. Integrating artificial intelligence algorithms and IoT into wearable technologies and the progress of sensors has created significant innovations in many fields, such as manufacturing, health, sports, etc.. However, wearable technologies have faced challenges and difficulties such as security, privacy, accuracy, latency, and connectivity. More specifically, the increasingly massive and complex data volume has dramatically influenced the improvement of the limits. However, these challenges have created a new solution: the federated Learning algorithm. In recent years, federated learning has been implemented with deep learning and AI to enhance powerful computing with big data, stable accuracy, and ensure the security of edge devices. In this chapter, the first objective is to survey the applications of wearable Internet of Things devices in industrial sectors, particularly in manufacturing. Second, the challenges of wearable Internet of Things devices are discussed. Finally, this chapter provides case studies applying machine learning, deep learning, and federated learning in fall and fatigue classification. These cases are the two most concerning work efficiency and safety topics in Smart Manufacturing 5.0.

In this work, we illustrate the interest in the use of a digital twin for the online fault diagnosis in a manufacturing system with sensors and actuators delivering binary signals that can be modeled as Discrete Event Systems. This chapter presents an intelligent diagnostic solution to replace traditional solutions, which are often non-industrialized, with a new data-based method learned from the simulation of the plant behaviors and using recurrent neural networks (RNN) with short-term and long-term memory (Long short-term memory, LSTM).