Industrial Artificial Intelligence Technologies and Applications

✍ Scribed by Ovidiu Vermesan, Franz Wotawa, Mario Diaz Nava, Björn Debaillie

• Publisher: River Publishers
• Year: 2022
• Tongue: English
• Leaves: 244
• Series: River Publishers Series in Communications and Networking
• Category: Library

✦ Synopsis

The advances in industrial edge artificial intelligence (AI) are transforming the way industrial equipment and machine interact with the real world, with other machines and humans during manufacturing processes. These advances allow industrial internet of things (IIoT) and edge devices to make decisions during the manufacturing processes using sensors and actuators data.
Digital transformation is reshaping the manufacturing industry, and industrial edge AI aims to combine the potential advantages of edge computing (low latency times, reduced bandwidth, distributed architecture, improved trustworthiness, etc.) with the benefits of AI (intelligent processing, predictive solutions, classification, reasoning, etc.).

The industrial environments allow the deployment of highly distributed intelligent industrial applications in remote sites that require reliable connectivity over wireless and cellular connections. Intelligent connectivity combines IIoT, wireless/cellular and AI technologies to support new autonomous industrial applications by enabling AI capabilities at the edge and allowing manufacturing companies to improve operational efficiency and reduce risks and costs for industrial applications.

There are several critical issues to consider when bringing AI to industrial IoT applications considering training AI models at the edge, the deployment of the AI-trained inferencing models on the target reliable edge hardware platforms and the benchmarking of the solution compared with other implementations.

The next-generation trustworthy industrial AI systems offer dependability by design, transparency, explainability, verifiability, and standardised industrial solutions to be implemented into various applications across different industrial sectors.

New AI techniques like embedded machine learning (ML) and deep learning (DL) capture edge data, employ AI models and deploy them to hardware target edge devices from ultra-low-power microcontrollers to embedded devices, gateways, and on-premises servers for industrial applications. These techniques reduce latency, increase scalability, reliability, and resilience, and optimise wireless connectivity, greatly expanding IIoT capabilities.

The book overviews the latest research results and activities in industrial artificial intelligence technologies and applications based on the innovative research, developments and ideas generated by the ECSEL JU AI4DI, ANDANTE and TEMPO projects.

The authors describe industrial AI's challenges, the approaches adopted, and the main industrial systems and applications to give the reader a good insight into the technical essence of the field. The articles provide insightful material on industrial AI technologies and applications.

The book is a valuable resource for researchers, post-graduate students, practitioners, and technology developers interested in gaining insight into the industrial edge AI, IIoT, embedded machine and deep learning, new technologies, and solutions to advance the intelligent processing at the edge.

✦ Table of Contents

Front Cover
Industrial Artificial Intelligence Technologies and Applications
Dedication
Acknowledgement
Contents
Preface
List of Figures
List of Tables
List of Contributors
1 Benchmarking Neuromorphic Computing for Inference
1.1 Introduction
1.2 State-of-the-art in Benchmarking
1.2.1 Machine Learning
1.2.2 Hardware
1.3 Guidelines
1.3.1 Fair and Unfair Benchmarking
1.3.2 Combined KPIs and Approaches for Benchmarking
1.3.3 Outlook : Use-case Based Benchmarking
1.4 Conclusion
References
2 Benchmarking the Epiphany Processor as a Reference Neuromorphic Architecture
2.1 Introduction and Background
2.2 Comparison with a Few Well-Known Digital Neuromorphic Platforms
2.3 Major Challenges in Neuromorphic Architectures
2.3.1 Memory Allocation
2.3.2 Efficient Communication
2.3.3 Mapping SNN onto Hardware
2.3.4 On-chip Learning
2.3.5 Idle Power Consumption
2.4 Measurements from Epiphany
2.5 Conclusion
References
3 Temporal Delta Layer: Exploiting Temporal Sparsity in Deep Neural Networks for Time-Series Data
3.1 Introduction
3.2 Related Works
3.3 Methodology
3.3.1 Delta Inference
3.3.2 Sparsity Induction Using Activation Quantization
3.3.2.1 Fixed Point Quantization
3.3.2.2 Learned Step-Size Quantization
3.3.3 Sparsity Penalty
3.4 Experiments and Results
3.4.1 Baseline
3.4.2 Experiments
3.4.3 Result Analysis
3.5 Conclusion
References
4 An End-to-End AI-based Automated Process for Semiconductor Device Parameter Extraction
4.1 Introduction
4.2 Semantic Segmentation
4.2.1 Proof of Concept and Architecture Overview
4.2.2 Implementation Details and Result Overview
4.3 Parameter Extraction
4.4 Conclusion
4.5 Future Work
References
5 AI Machine Vision System forWafer Defect Detection
5.1 Introduction and Background
5.2 Machine Vision-based System Description
5.3 Conclusion
References
6 Failure Detection in Silicon Package
6.1 Introduction and Background
6.2 Dataset Description
6.2.1 Data Collection and Labelling
6.3 Development and Deployment
6.4 Transfer Learning and Scalability
6.5 Result and Discussion
6.6 Conclusion and Outlooks
References
7 S2ORC-SemiCause: Annotating and Analysing Causality in the Semiconductor Domain
7.1 Introduction
7.2 Dataset Creation
7.2.1 Corpus
7.2.2 Annotation Guideline
7.2.3 Annotation Methodology
7.2.4 Dataset Statistics
7.2.5 Causal Cue Phrases
7.3 Baseline Performance
7.3.1 Train-Test Split
7.3.2 Causal Argument Extraction
7.3.3 Error Analysis
7.4 Conclusions
References
8 Feasibility ofWafer Exchange for European Edge AI Pilot Lines
8.1 Introduction
8.2 Technical Details and Comparison
8.2.1 Comparison TXRF and VPD-ICPMS Equipment for Surface Analysis
8.2.2 VPD-ICPMS Analyses on Bevel
8.3 Cross-Contamination Check-Investigation
8.3.1 Example for the Comparison of the Institutes
8.4 Conclusiion
References
9 A Framework for Integrating Automated Diagnosis into Simulation
9.1 Introduction
9.2 Model-based Diagnosis
9.3 Simulation and Diagnosis Framework
9.3.1 FMU Simulation Tool
9.3.2 ASP Diagnose Tool
9.4 Experiment
9.5 Conclusion
References
10 Deploying a Convolutional Neural Network on Edge MCU and Neuromorphic Hardware Platforms
10.1 Introduction
10.2 Related Work
10.3 Methods
10.3.1 Neural Network Deployment
10.3.1.1 Task and Model
10.3.1.2 Experimental Setup
10.3.1.3 Deployment
10.3.2 Measuring the Ease of Deployment
10.4 Results
10.4.1 Inference Results
10.4.2 Perceived Effort
10.5 Conclusion
References
11 Efficient Edge Deployment Demonstrated on YOLOv5 and Coral Edge TPU
11.1 Introduction
11.2 Related Work
11.3 Experimental Setup
11.3.1 Google Coral Edge TPU
11.3.2 YOLOv5
11.4 Performance Considerations
11.4.1 Graph Optimization
11.4.1.1 Incompatible Operations
11.4.1.2 Tensor Transformations
11.4.2 Performance Evaluation
11.4.2.1 Speed-Accuracy Comparison
11.4.2.2 USB Speed Comparison
11.4.3 Deployment Pipeline
11.5 Conclusion and Future Work
References
12 Embedded Edge Intelligent Processing for End-To-End Predictive Maintenance in Industrial Applications
12.1 Introduction and Background
12.2 Machine and Deep Learning for Embedded Edge Predictive Maintenance
12.3 Approaches for Predictive Maintenance
12.3.1 Hardware and Software Platforms
12.3.2 Motor Classification Use Case
12.4 Experimental Setup
12.4.1 Signal Data Acquisition and Pre-processing
12.4.2 Feature Extraction, ML/DL Model Selection and Training
12.4.3 Optimisation and Tuning Performance
12.4.4 Testing
12.4.5 Deployment
12.4.6 Inference
12.5 Discussion and Future Work
References
13 AI-Driven Strategies to Implement a Grapevine Downy MildewWarning System
13.1 Introduction
13.2 Research Material and Methodology
13.2.1 Datasets
13.2.2 Labelling Methodology
13.3 Machine Learning Models
13.4 Results
13.4.1 Primary Mildew Infection Alerts
13.4.2 Secondary Mildew Infection Alerts
13.5 Discussion
13.6 Conclusion
References
14 On the Verification of Diagnosis Models
14.1 Introduction
14.2 The Model Testing Challenge
14.3 Use Case
14.4 Open Issues and Challenges
14.5 Conclusion
References
Index
About the Editors
Back Cover