Artificial Intelligence In Engineering Applications I

Learning Results

The specific course is an introduction to basic concepts of Artificial Neural Networks (ANNs) and Machine Learning (ML) techniques. As the time passes, the courses deepen in more specialized knowledge and concepts that require ability and skills development from the students, both understanding and use of new technologies.

This course is also a technological analysis on issues, problems and applications related to modeling techniques based on artificial intelligence.

The aim of the course is the understanding of the basic elements of ANNs and ML in issues related to applications of modeling and forecasting for Engineers.

Upon completion of the course, students will be able to:

• They distinguish, interpret and clearly explain concepts and issues related to ANNs & ML techniques.
• Be able to perceive, interpret and clearly explain issues related to ANNs & ML, to generalize the problem, to correctly appreciate in order to make right conclusions.
• Be able to use all the concepts related to ANNs & ML, to provide new calculations, to be able to correctly classify the causes of the various problems and generate new knowledge, while gaining implementation experience.
• Have proven ability to create and manage large data files, which are necessary for ANNs & ML models training and development.
• To make new calculations, to be able to correctly classify the issues that cause the various relevant problems and to generate new knowledge, while gaining experience in applying ANNs & ML modeling techniques.
• To have the ability to distinguish and analyze in their potential components the issues that will be modeled with the use and application of ANNs and ML, so that they can combine, design, develop and implement older and innovative technologies to deal with these problems/issues.
• Be able to review initial thoughts and views related to the development, use and implementation of ANNs & ML so that they can create, as far as possible, new knowledge and be able to compose and organize working groups and propose solutions.
• To have proven judgment, to be able to compare and evaluate different situations/proposals regarding the development, use and application of ANNs and ML, concerning the modeling of different magnitudes and parameters.
• To be able to properly plan the development and the training of ANNs and ML models in issues related to modeling/forecasting of parameters related to the science of Mechanical Engineering, such as energy production, buildings energy consumption and saving, indoor and outdoor human thermal comfort/ discomfort, indoor and outdoor air quality, air pollution management, etc.
• Have the ability to evaluate the accuracy and reliability of a developed ANNs or ML model using appropriate statistical evaluation methods.
• To be able to work with their fellow students to create and present, both individually and in groups, a case study from its initial stages to its final evaluation and proposal for solutions.

Skills acquired

More specifically, students will be able to:

1. Theoretical knowledge concerning ANNs and ML modeling techniques

2. Search, analysis and synthesis of data and information using and applying the required technologies

3. Practical ability of applications concerning ANNs & ML

4. Ability to evaluate the accuracy and the efficiency of developed ANNs & ML models

5. Decision making

6. Individual project

7. Teamwork

Theory

The core modules of the course include:

• Introduction to Machine Learning
• Introduction to artificial neural networks (ANNs)
• Advantages and disadvantages of ANNs
• ANNs types and classification
• Introduction to the MultiLayer Perceptron-MLP
• ANNs training algorithms and methods
• The error back propagation training algorithm
• Methods for improving the generalization ability of ANNs
• ANNs development and training with the use of Matlab ANNs Toolbox
• Evaluation of developed ANNs models using appropriate statistical methods
• Machine learning in Computer Aided Design modeling techniques for design optimization. AI-powered generative design tools to automatically create and evaluate multiple design options.
• Machine learning in Computer Aided Engineering (Neural Network Optimization (NNO) algorithm for constitutive material parameter identification based on inverse analysis of experimental tests of specific specimens under different mechanical loads).
• Machine learning in machining (prediction of surface roughness contrary to cycle time exploiting Computer Aided Manufacturing systems in 3-Axis up to 5-Axis CNC milling processes).
• Machine learning in Additive Manufacturing (prediction of surface roughness in relation to cycle time in multi-material metallic specimens).
• Data management techniques and forecasting applications for optimizing Computer Aided Design (CAD) learning
• Prediction of material & rock characteristics using Neural Networks
• Sensor optimization for Structural Health Monitoring
• Anomaly/Intrusion Detection using Neural Networks and Adaptive Modelling
• Inspection of large infrastructures using UAVs and AI
• Crack detection on Concrete & Pavements using Deep Learning
• AI methods on Multi-Criteria Decision Making, Operational Research and decision problems in engineering project management

Laboratory

The workshop includes the following laboratory exercises:

• Development of databases suitable for the appropriate ANNs training
• Statistical evaluation methodology of modeling/forecasting results
• Use of Matlab ANNs Toolbox for the development and training of ANNs models
• Application and use of ANNs for the electricity consumption forecasting in the domestic/building sector
• Application and use of ANNs for capacity factor prediction related to wind turbines energy production
• Application and use of ANNs for air pollution and air pollution management forecasting.
• Application and use of ANNs for human thermal comfort/discomfort forecasting concerning energy applications in the building sector
• Application and use of ANNs and ML Computer Aided Design modeling techniques for design optimization regarding stiff and at the same time lightweight structures.
• Application and use of ANNs and ML Computer Aided Engineering modeling techniques for constitutive material parameters identification.
• Application and use of ANNs and ML modeling techniques for machining parameters optimization in terms of surface roughness and cycle time in 3-Axis up to 5-Axis CNC milling processes.
• Application and use of ANNs and ML Additive Manufacturing modeling techniques for 3D-printing parameters optimization in multi-material metallic specimens.
• Application and use of ANNs for prediction of material & rock characteristics using Neural Networks
• Application and use of ANNs for sensor optimization for Structural Health Monitoring
• Application and use of ANNs and Adaptive Modelling for Anomaly/Intrusion Detection
• Application UAVs and AI for the inspection of large infrastructures
• Deep learning application for crack detection on concrete & pavements

1. Russell, R. (2018). Neural Networks. Easy Guide to Artificial Neural Networks. CreateSpace Independent Publishing Platform. ISBN-10: 1718898428, ISBN-13: 978-1718898424

2. Beale, R., & Jackson, T. (1990). Neural Computing: Αn Ιntroduction. Ν.Υ.: Adam Hilger.

3. Snehashish Chakraverty and Sumit Kumar Jeswal, (2021). Applied Artificial Neural Network Methods for Engineers and Scientists. World Scientific Publishing Co. Pte. Ltd. https://doi.org/10.1142/12097