Artificial Intelligence in Design Engineering.

1
Artificial Intelligence in Design Engineering
MEC
2
Contents

• Definitions.
• Key Engineering Design Processes.
• Key Technologies.
• Weak and Strong AI.
• Applying AI to Engineering Designs.
• Benefits of AI.
• Merits and Challenges.
• Ethical Issues.

3
Engineering Design

• Application of engineering principles to design
  and develop products, systems, and processes that
  meet specific requirements.
• Encompasses various disciplines - mechanical,
  electrical, software, industrial engineering etc.
• Use of tools like computer-aided design,
  software, simulation, and prototyping. 

4
ABET Definition

• Process of devising a system, component, or
  process to meet desired needs.
• A decision-making process (often iterative).
• Application of basic sciences, mathematics, and
  engineering sciences to convert resources
  optimally to meet a stated objective. 
• Fundamental elements of design process include
  establishment of objectives and criteria,
  synthesis, analysis, construction, testing, and
  evaluation.

Accreditation Board for Engineering and
Technology, October 2017
5
Engineering Design

• Identifying opportunities.
• Developing requirements.
• Performing analysis and synthesis.
• Generating multiple solutions.
• Evaluating solutions against requirements,
  considering risks.
• Making trade-offs to obtain high-quality solution.

Approaches to Addressing ABET Engineering Design
Requirements Jeffrey W. Fergus
6
Design Engineering

• Iterative, systematic process for solving
  problems.
• Involves creativity, experience, and accumulated
  disciplinary knowledge. 
• Dynamic process, not a rigid method.
• Result of engineering design process not always a
  product, can be a process or a computer program.

7
Design Engineering

• Design engineers employed in a wide range of
  industries
• - Aerospace.
• - Automotive.
• - Consumer electronics.
• - Biomedical engineering.
• - Robotics/IT.
• - Manufacturing.
• - Construction.
• - Electronics/Telecommunications.

8
Key Engineering Design Phases

• Ideation and conceptual phase identify the
  engineering problem and develop a concept.
• Creation design (fabricate) a prototype of the
  concept (usually a CAD model).
• Redefine and enhance the design.
• Validate the design test with CAE.
• Build develop optimal production processes for
  the design.

9
Artificial Intelligence

• Enables computers and machines to simulate human
  learning, comprehension, problem solving,
  decision making, creativity and autonomy.
• Machine learning -creating models by training an
  algorithm to make predictions or decisions based
  on data.

10
Artificial Intelligence

• Deep Learning use of multilayered neural
  networks (deep neural networks) to closely
  simulate complex decision-making power of human
  brain.
• Generative AI (gen AI) - deep learning models to
  create complex original content long-form text,
  high-quality images, realistic video or audio
  etc. in response to a users prompt or request.

11
Artificial Intelligence
12
How can AI help?

• Gather and analyze reference information.
• Generate design ideas and alternatives.
• Optimize parameters and complex combinations.
• Create more efficient designs.
• Design more quickly than humans.
• Create layouts and 3D models.
• Program design tools.
• Quality reviews of designs.

13
How can AI help?
A 3D rendering service developed by AI
14
Key Technologies

• Machine Learning
• Algorithms analyze data and learn patterns to
  predict outcomes or optimize designs.
• Techniques include supervised, unsupervised, and
  reinforcement learning.
• Computer Vision
• Enables AI systems to interpret and process
  visual design data.
• Useful in quality control, defect detection, and
  reverse engineering.

15
Key Technologies

• Natural Language Processing (NLP)
• Assists in processing and analyzing engineering
  documents, standards, and user feedback.
• Neural Networks and Deep Learning
• Mimic human cognition to solve complex design
  problems.
• Effective in recognizing patterns in
  multidimensional datasets.

16
Traditional vs AI Algorithms
17
Weak and Strong AI

• Weak AI or narrow AI, - AI systems designed to
  perform a specific task or a set of tasks. eg
  Voice Assistant Apps Alexa, Co-pilot.
• Strong AI or artificial general intelligence
  (AGI) or general AI, with ability to understand,
  learn and apply knowledge across a wide range of
  tasks at levels equal to or more than human
  intelligence.

18
Virtual Assistants

• A large language model (LLM) that interacts in a
  human-like manner.
• Can perform a range of tasks or services based on
  user input such as commands or questions.
• Typically utilizes online chat (chatbot, eg Chat
  GPT) capabilities to simulate human conversation,
  also called a visual dialog model.
• Some virtual assistants can interpret human
  speech and respond via synthesized voice.
• Assist engineers in finding information and
  making informed decisions.

19
Chatbot Answering Questions
20
Chat GPT Design of a Biomimetic Fan Regulator
21
Digital Twins

• Combines human expertise and machine intelligence
  to permit evolution of work in new and unexplored
  ways. 
• To design virtual replicas of products in a
  virtual world, simulate processes and improve
  operations over time.
• Allows design engineers to test products before
  expending resources to produce them.
• Anticipates problems, prevents mistakes before
  occurence.
• Detects anomalies, automates repair processes.

22
AI in Transportation Design

• Intelligent Transportation Systems (ITS) leverage
  AI to enhance traffic management, safety, and
  efficiency.
• Data collected and analyzed through sensors and
  cameras to identify congestion, areas of
  speeding, predict traffic patterns, and optimize
  traffic flow.
• A proactive approach to traffic management.

23
Intelligent Transportation Systems

• Can adjust traffic lights and reroute vehicles to
  minimize congestion.
• Reduce travel time.
• Minimize the likelihood of accidents.
• Create a safer travel environment.
• Provide alerts for emergency response.
• Provide traffic insights for design purposes.

24
Automated Highway Systems

• AI technologies to control vehicle movements
  based on real-time information on surroundings.
• Connected vehicles to make driving more reliable
  and efficient.
• V2V (Vehicle to Vehicle).
• V2I (Vehicle to Infrastructure).
• V2P (Vehicle to Pedestrian).
• V2G (Vehicle to Grid).
• V2D (Vehicle to Device like smartphone).

25
Smart Highways

• Interconnections between several items.....
• Dynamic real time responses to changing traffic
  and weather.
• Instant updates to road conditions ahead.

26
Road Design

• Software to design roads.
• Civil 3D, OpenRoads, and other civil software
  will soon have AI toolkits released.
• Softree RoadEng Optimal has an AI toolbar ton
  help make design decisions and make adjustments
  to a previous design to achieve project
  objectives.
• Software can quickly optimize routes between
  points, keep grades within maximum and minimum
  ranges, minimize cut and fill quantities, avoid
  no-go zones, and adjust for crossings.

27
Traffic Simulation

• Trafficware by Cubic and PTV Vissim with AI.
• Use of AI to test traffic flow scenarios and
  alternative arrangements for infrastructure
  projects.
• Enables engineers to visualize traffic issues and
  make informed decisions to optimize traffic
  management.

28
Structural Design

• Software with AI capabilities for structural
  design.
• Can optimize a structural design based on desired
  objectives such as minimizing cost, weight, or
  footprint.
• Can create or modify a design based on project
  limitations.
• Can identify problems/failures in the design and
  modify the design.
• Programs with feedback or reinforced learning.
• User can identify areas where the program did not
  make a desirable design decision.

29
Design Optimization
Topology, Shape and Sizing Optimization possible.
30
Automotive Design

• Design and optimize the shape of vehicles.
• Helps to maximize aerodynamic efficiency and
  reduce drag forces.
• Used in a variety of aerospace applications to
  reduce weight while still meeting structural
  strength and deflection requirements.

31
Design of Structures

• Design from ground up based on initial directions
  such as building purpose, number of occupants,
  location, space available, height restrictions,
  budget, etc.
• Identify horizontal and vertical irregularities
  and evaluate modification options to remove them.
  
• STAAD Pro utilizes AI for complex load analysis,
  including running various seismic and wind load
  combinations and analyzing the results for each
  structural member.

32
CFD Modelling

• Computational fluid dynamics (CFD) modeling helps
  civil, process, mechanical, and biomedical
  engineers simulate designs involving moving
  liquids and gases.
• AI allows engineers to simulate more iterations
  in shorter times and view the results very
  quickly.
• Software such as Solid Works, Ansys Fluent, Xflow
  etc.

33
Machine and Engine Design

• AI has the potential to handle the complexity and
  vast amount of information involved in a full
  engine simulation.
• Mechanical Software packages such as Engine
  Builder, Engine Analyzer Pro, Fusion 360.
• Simulations to anticipate weaknesses and make
  design modifications to make engines more robust.

34
Electrical Design

• AI models can simulate complex electrical
  systems, predict performance, and provide
  insights and tools that guide the design of more
  efficient and robust systems.
• Software for electrical design with AI such as
  Altair HyperWorks, Ansys Electronics Desktop,
  MATLAB Simscape Electrical, Siemens NX with AI
  capabilities

35
Circuit Design

• AI to automatically generate circuit schematics
  and layout designs, and how to use simulation
  tools to test and evaluate circuit performance.
• Tools such as Cadence, Snapmagic Copilot,
  Synopsis.ai copilot with AI capabilities.
• Use of Chat GPT to choose a circuit diagram.

36
Predictive Maintenance

• Proactive approach.
• Avoids surprising failures, extends the lifespan
  of equipment, minimizes downtime, optimizes
  performance, and reduces resources.
• AI driven software can continuously analyze
  electrical installations to predict potential
  malfunctions before they occur.

37
Smart Grid Management

• AI to oversee and optimize electricity
  distribution.
• Remote terminal units installed on various
  electrical lines in the distribution system to
  collect real-time data from the electrical grid.
• AI software with predictive analytics and machine
  learning to achieve forecast demand, adapt to
  supply changes, prevent outages and resource
  reduction.

38
Smart Grid Management
Smart Grid Arrangement for a Solar Panel
39
Artificial Intelligence in Engineering Design

• Generative Design
• - To explore design options based on
• constraints and objectives.
• Use techniques like topology optimization
• to create lightweight, high-performance
  designs.
• Use in aerospace, automotive, and architecture.

40
Artificial Intelligence in Engineering Design

• Predictive Analytics and Optimization
• - Machine learning models to predict design
  
• performance under various conditions.
• - Parameter optimization for better
• efficiency, durability, and
  cost-effectiveness.

41
Artificial Intelligence in Engineering Design

• Simulation and Modeling
• - Predicting real-world behavior of
  designs.
• Reducing the need for physical prototypes.
• Savings in time and costs.
• Automation of Repetitive Tasks
• - Automating CAD modeling, component
  selection
• and other routine tasks.
• - Frees up engineers to focus on creative
  and
• strategic problem-solving.

42
Artificial Intelligence in Engineering Design

• Design Customization
• - analyzing user preferences, generating
• personalized designs.
• - Useful in consumer products and medical
• applications.

43
Artificial Intelligence in Engineering Design

• Failure Analysis and Risk Assessment
• - To identify potential design flaws and
• predict failures. -
• - Analyzing historical data and real-time
• inputs to enhance safety and reliability .

44
Benefits of Artificial Intelligence

• Improved Efficiency Reduces time and costs in
  the design cycle.
• Innovation Encourages out-of-the-box solutions
  by exploring unconventional design spaces.
• Accuracy Enhances precision in simulations,
  predictions, and optimizations.
• Sustainability Optimizes designs for reduced
  material usage and energy consumption.

45
Benefits of Artificial Intelligence

• Automation of repetitive tasks - automate
  routine, repetitive and often tedious tasks.
• More and faster insight from data - generate and
  evaluate various design possibilities
  automatically.
• Enhanced decision-making accurate and reliable.
• Fewer human errors flagging before they occur.
• 24x7 availability no machine fatigue!.
• Reduced physical risks automation of dangerous
  works!.

46
Challenges

• Data Dependence
• Requires large, high-quality datasets for
  training AI models.
• Interpretability
• Difficult to understand and trust AI-generated
  solutions in critical applications.
• Integration
• Requires compatibility with existing engineering
  workflows and tools.
• Ethical and Social Concerns
• Balancing automation with the human role in
  decision-making and creativity.
• Job Security?..... unskilled?

47
AI Driven Designs

• Not as a replacement for engineers.
• As a tool that augments their capabilities.
• More efficient workflows.
• Frees engineers to focus on more creative and
  strategic aspects of their work.
• New opportunities for innovation and
  problem-solving.
• Integrating human and artificial intelligence to
  achieve better outcomes Collaborative
  Intelligence.

48
Design by Morphing

• Optimal design for objects considering their
  aerodynamic, hydrodynamic, thermal, and/or
  structural performance. 
• Drawbacks of currently used methods of optimal
  design based on Trial and Error approaches or
  Gradient-based methods overcome. ?
• Dependence on designer heuristics, complexity and
  computational costs reduced.

49
Design by Morphing
50
Ethical Issues
51
Deep Fakes

• Synthesized images, videos, or audio edited or
  generated using artificial intelligence tools.
• May depict real or non-existent people, satellite
  images or buildings. 
• Can scramble our understanding of truth.
• Raises a set of challenging policy, technology,
  and legal issues.
• AI algorithms called encoders used in
  face-replacement and face-swapping technology.
• Decoder retrieves and swaps images of faces,
  which enables one face to be superimposed onto a
  completely different body.

52
Deep Fakes

• Generative Adversial Network  neural network
  technology uses generator and discriminator
  algorithms to develop all deepfake content.
• Convolutional neural networks  - analyze patterns
  in visual data, used for facial recognition and
  movement tracking.
• Autoencoders  - a neural network technology
  identifies the relevant attributes of a target
  such as facial expressions and body movements,
  and imposes these attributes onto the source
  video.

53
Deep Fakes

• Natural language processing  - used to create
  deepfake audio. NLP algorithms analyze attributes
  of a target's speech and generate original text
  using the attributes.
• High-performance computing - a type of computing
  that provides the significant necessary computing
  power deepfakes require.
• Video editing software - not always AI-based, but
  frequently integrates AI technologies to refine
  outputs and make adjustments that improve
  realism.

54
Deep Fake Architecture
https//www.researchgate.net/figure/General-DeepFa
ke-Architecture_fig1_382579812
55
Deep Fake
https//cacm.acm.org/research/beyond-deep-fakes/
56
(No Transcript)
57
Deep Fakes
Left Original Photograph Right Image
generated in Midjourney with word prompts alone
(www.linkedin.com). Midjourney is an AI based
text to image converter.
58
AI Hallucinations

• Large language model perceives patterns or
  objects that are nonexistent or imperceptible to
  human observers.
• Creates nonsensical or altogether inaccurate
  outputs.
• AI algorithms produce outputs that are not based
  on training data.
• Incorrectly decoded by the transformer / do not
  follow any identifiable pattern.

59
AI Hallucinations

• AI model trained on dataset comprising biased or
  unrepresentative data.
• Adversarial attack by adding small amounts of
  specially-crafted noise to an image.
• Leads to unnecessary medical interventions.
• Contributes to the spread of misinformation. 

60
Solving AI Hallucinations

• Establish chosen AI systems responsibilities and
  limitations.
• Adversarial training to counter adversarial
  attack.
• AI models trained on diverse, balanced and
  well-structured data.
• Data templates to ensure output consistency and
  reduce the likelihood of faulty results.
• Define boundaries for AI models using filtering
  tools. 

61
Solving AI Hallucinations

• Test and refine the system continually.
• Human intervention to validate and review AI
  outputs. 
• Stop them before they happen!

62
Conclusion

• Artificial Intelligence (AI) is transforming
  engineering design by automating processes,
  enhancing creativity, and improving
  decision-making. AI tools can analyze vast
  datasets, optimize complex systems, and simulate
  real-world scenarios, enabling engineers to
  develop innovative solutions more efficiently.

63
References

• Artificial Intelligence in Engineering Design -
  SunCam online continuing education course
  material.
• https//www.ibm.com/think/topics/artificial-intell
  igence
• https//www.arcweb.com/industry-best-practices/und
  erstanding-role-ai-generative-engineering-design
• https//indiaai.gov.in/article/the-art-and-algorit
  hms-of-deepfake-ai-a-comprehensive-study
• https//funginstitute.berkeley.edu/capstone-projec
  t/design-by-morphing/
• Chat GPT and other Internet Sources.

64
AI Ready ?