دانلود کتاب فناوری ها و کاربردهای هوش مصنوعی صنعتی

Digitalisation and Industry 5.0 are changing how manufacturing facilities operate by deploying many sensors/actuators, edge computing, and IIoT devices and forming intelligent networks of collaborative machines that are able to collect, aggregate, and intelligently process data at a network’s edge. Given the vast amount of data produced by IIoT devices, computing at the edge is required. In this context, edge computing plays an important role – the edge should provide computing resources for edge intelligence with dependability, data management, and aggregation provision in mind. Edge intelligence – for example, AI technologies with edge computing for training/learning, testing, or inference – is essential for IIoT applications to build models that can learn from a large amount of aggregated data. Edge computing is a distributed computing paradigm that brings computation and data storage closer to a device’s location. AI algorithms process the data created on a device with or without an internet connection. These new AI-based algorithms allow data to be processed within a few milliseconds, providing real-time feedback. The AI models operate on the devices themselves without the need for a cloud connection and without the problems associated with data latency, which results in much faster data processing and support for use cases that require real-time inferencing. Major challenges remain in achieving this potential due to the inherent complexity of designing and deploying energy-efficient edge AI algorithms and architectures, the intricacy of complex variations in neural network architectures, and the underlying limited processing capabilities of edge AI accelerators. Industrial-edge AI can run on various hardware platforms, from ordinary microcontrollers (MCUs) to advanced neural processing devices. IIoT edge AI-connected devices use embedded algorithms to monitor device behaviour and collect and process device data. Devices make decisions, automatically correct problems, and predict future performance.