Applied AI and Multimedia Technologies for Smart Manufacturing and CPS Applications

In the past decade, artificial intelligence (AI), data analytics, multimedia technologies and methods of integrating smart systems in manufacturing industries and in Industry 4.0 applications have been steadily growing in availability and diversity. However, for industrial leaders, finding cost effective AI and robust multimedia technologies that are easily implementable, remains a challenging endeavor. In addition, due to the lack of uniform standards for Industry 4.0 and cyber-physical systems (CPS), different types of data sets in different formats are always generated from differently standardized devices at different layers of manufacturing CPS. Thus, finding means and methods of reliable data governance, data integration, inter-operable data generation and data analytics for CPS and for general industrial applications is also quite challenging. These problems are even more pervasive since the ecosystem of industrial and CPS devices and applicable technologies keeps diverging. Applied AI and Multimedia Technologies for Smart Manufacturing and CPS Applications provides an in-depth review of the challenges and applicable state-of-the art multimedia, deep learning, data conversion and data integration solutions for IIoT and industrial CPS. In the book, challenges relating to the integration of data, machine learning (ML) and multimedia technologies for smart manufacturing applications and CPS are discussed in detail. In-depth and wide-ranging hands-on examples and solutions for ML, data engineering, data conversion and data integration strategies for industrial CPS, and for academic and industrial captains are also provided in form of algorithms and Python software implementations. The book is valuable and quite timely due to its substantive coverage of data engineering, schema and metadata generation strategies for common time series data types that are found in many manufacturing and medical CPS. From extensive literature searches, the book provides the most expansive coverage of Python-based time series data conversion strategies in literature. Algorithms and Python-based data conversion, data engineering, schema and metadata generation solutions that are provided in the book can be used to integrate or convert YAML, CSV, Tensors, HDF5, JSON, Avro, Feather, Parquet etc., data formats to other formats. These data conversion solutions can provide means through which different data formats and data types that are generated from many subsystems of CPS can be collectively useful for providing needed AI solutions for CPS. Methods of making different data formats that are generated from different parts of a CPS to be more inter-operable and readily convertible from one format to another are also extensively provided with the aid of hands-on Python implementations. Moreover, being able to source data from different parts of CPS and being able to convert generated data sets from one format to another can provide avenues for implementing very reactive and scalable distributed deep learning (DL) solutions for CPS. Time series semantic data interpretation described in detail in one of the book chapters can also contribute to providing needed meaning and context for better understanding of the intrinsic interactions of different parts of CPS. The suite of algorithms and Python solutions provided in the book can also assist experienced data engineers, cloud engineers, IIoT and DevOps practitioners to provide needed data conversion solutions to most problems that are always encountered in industry 4.0 and CPS projects. Other likely beneficiaries of the algorithms and Python implementations presented in the book include applied AI and ML scholars, practitioners, researchers, educators, ML enthusiast and industrial captains. Topics covered in the book include: Applied DL for industrial fault diagnosis, DL methods in the concept of Industry 4.0, in-depth study on data engineering for Factory of the Future (FoF), performance evaluation of state-of-the-art data formats for time series applications, algorithms and Python-based software development for time series data formats conversion, data engineering for multimedia applications and cyber-physical systems, an overview of the characteristics and importance of next generation industrial robotics, a review of Big Data analytics for the Internet of Things (IoT) applications in supply chain management, cognitive load measurement based on electroencephalogram (EEG) Signals, nonlinear filtering methods in conditions of uncertainty for mobile CPS such as UAVs and airplanes, ML based approach for suspicious activity detection using agents’ facial analysis, etc.