Industry 4.0 - UoB
Data Science workflow Stage 2 and key stages (Understand the data )
'Playing' with data and trying to understand its meaning Interrogate the Data - know better the dataset you are handling, realise potential issues Data Wrangling - converting to a format for analysis Exploratory analysis - ¨descriptive statistics¨ and typically involve the computation and/or visualisation of some statistical measures to characterise the spread, range and shape of the dataset.
Describe the data science workflow
1. Frame the problem 2. Understand the data 3. Extract features 4. Model And analyse -> 1 5. Present results OR Deloy code
What is data science
Deals with both quantitative and qualitative data (e.g. images) and focus on prediction and action (i.e. decision making)
Difference between Machine Learning and Deep learning
Deep Learning algorithms can extract features from data automatically - in ML we need to specify them manually
What is the meaning of Blockchain in detail
Digital platform that records and verifies transactions in a public and secure manner Distributed (i) data are stored in a decentralised way, using a peer-to peer network (ii) data can be written and accessed from any of the authorised locations - called nodes - that belong to network. Secure - a peer in the network cannot modify unilaterally a ¨block¨ once it has been added to the ¨chain¨. Ledger - it stores data/information.
What is smart manufacturing?
Digital solutions enable enhanced system connectivity, data analytics, and functional automation that improve operational agility, adaptability, and profitability
Key Features of a smart factory
Digitised and integrated processes, operations Improves machine utilisation and reduced maintenance costs. Update or receive data on the go, therefore complete data for faster decision making; data-driven decision making. Inter of Things (IoT), sensor, mobile app, read frequency identification(RFID) enabled. Increased transparency, visibility on operations and production data. Smart and intelligent products. Accurate asset tracking using IoT, RFID; improves resource utilisation. High interoperability. When switching between different types of products, the needed resources and the route to link these resources should be reconfigured automatically and online.
What is a Smart Factory?
Disruptive digital technologies applied to industrial manufacturing
Elements of blockchains
Distributed ledger technology Immutable records Smart contacts
Why is automation & robotics used in I4.0
Do not require human control Able to make their own decisions based on their specific programming Ideal for manufacturing tasks that are risky for human operators Repetitive actions
Examples of digital twin application
Engineering and manufacturing fields - performance optimisation of engines, pumps, and turbines via real time simulations Food sector - variability of the raw materials and the complexity of the physicochemical phenomena Healthcare - in silico personalised copies of organs
Data Science workflow Stage 3
Extract features Feature is just a number or a category that is extracted from your data and describes some entity Iterative process
Data Science workflow Stage 1 and its features
Frame the problem Understand the need of our company/project, the goals to achieve and the resources Craft a well‐defined analytics problem (or problems) out
Key elements of a Smart factory
Fully automated and inter connected modular facility Sensors monitoring physical processes Large data collection and analysis via advance analytical capabilities or shared in the cloud
Examples of AI applications
Predictive maintenance Quality Control Predictive analytics - Forecasting
What is PAT
Process Analytical Technology Measurement-based methods that provide dynamic information about product properties, material flow properties, or operating conditions
What is PAT used for
Process control - Identify Key process parameters Quality Assurance Greater flexibility & cost efficiency
What are the types of Blockchains
Public - data and records are accessible and visible to the public, and any person or peer can provide data as well as verify it Consortium (hybrid blockchains) - Owner decides who has access = Transparency & customisation Private - Single centralised authoritative member (Organisation)
Types of Tasks in data science (9) (AACCDF RRT)
Regression - make predictions based on the fitting of a model to the dataset points or features. Classification - used also to make predictions, but this time with categorical/nominal variables Deep learning - it is a more advance technique based on neural networks that is also used to solve classification and regression tasks. Clustering - identify ¨clusters¨ groups in the dataset. Association analysis - identify items/features that are bundled together. Recommendation engines - systems that recommend items based on individual user preference Anomaly detection - used to identify outliers in datasets. Forecasting - used to predict the performance or behaviour of a variable/feature based on previous historical records Text mining - this task converts text (the input) into categorical/nominal data/variables (the outputs)
What Pillars enable new ways of performing tasks (e.g. autonomously, remotely, distributed).
Robotics Additive manufacturing Augmented reality
What is Machine Learning
The automated detection of meaningful patterns in datasets and the ability to develop algorithms that are capable to learn based on information provided by (big-)data analytics Learns from provided examples via classification, clustering and regression task with large datasets.
What are smart sensors and why are they useful?
They have the integration of sensing, computing and IoT Used with monitoring and measurement purposes, in particular real time scenarios, with the potential of taking corrective actions in an instant
Why is Data Science useful?
Tool to analyse complex and heterogenous data in a systematic way, helping us to identify patterns, to detect anomalies and to predict behaviours, this is, helping us to extract knowledge from data.
What is Big data
Trends and protocols for storage and processing of large datasets that require veryspecific software engineering tools and skills
Benefits of Blockchain
Trust - network validates every transaction & holds a single version of the truth Security - combination of cryptographic signatures, automated verification, and decentralised storage. Automation - set up a wide range of smart contracts so repetitive processes such as billing and shipping can be automated. Also triggers from measurable aspects Resilience - distributed, they can continue to operate even if part of the network fails.
Types of data analytical techniques
Univariate Analysis - Involves analysis for a single feature/variable at a time Multivariate Analysis - Analysis of two or more variables/features simultaneously
What is Technical assistance
Use of digital automated systems to assist human operated tasks and even to avoid difficult or unsafe ones.
What's is a Digital Twin
Virtual replica of a product, a machine, a process or of a complete production facility.
What makes data science different from big data
Volume and structure of the datasets as well as the software engineering tools
What is cybersecurity?
preventing the unauthorized access to data and information systems
What are Microsoft's 6Vs Big Data concept
volume: i.e., scale of data velocity: for the analysis of streaming data variety: i.e., different forms of data veracity: i.e., trustworthiness of data sources variability: i.e., the complexity - as the number of variables in the data. visibility: i.e., need for a full understanding of the data to make an informed decision.
What types of problems can be defined with the Data Science workflow
1. Supervised problems - predict the value of output variables/features based on a set of input variables/features 2. Unsupervised problems - aim at revealing patterns in datasets based only on the data points themselves.
When was Industry 1.0 and its key elements?
18th Century Built on machinery for water and steam production.
When was Industry 3.0 and its key elements?
20th Century Introduction of electronics and information technology to increase automation.
When was Industry 4.0 and its key elements?
21st Century Introduction of cyber-physical systems in our factories.
Where does Industry 4.0 Stem from?
A process that leads to fully automated and interconnected industrial production, by amplifying the digital connectivity of product, process, and factory
What is an algorithm?
A set of well-defined instructions, typically used to solve a class of specific problems or to perform a computation. Algorithms are used as specifications for performing calculations and data processing (Wikipedia).
What is Decentralisation of decision making
Ability of cyber physical systems to make decisions on their own and to operate in an autonomous way.
What is Interconnection?
Ability of machines, devices, sensors, and people to connect and communicate with each other via the IoT.
What is artificial intelligence?
Algorithms and machines capable of learning and solving tasks that usually require human intelligence, such as learning and problem solving
Example of Blockchain applications
Allow security, visibility, transparency along supply chains Traceability and safety in food supply chains Responsible sourcing in supply chains Sustainability in supplychains
What's is Information transparency
Being able to access data, data we collect through sensors or stored in the cloud, from all points across the manufacturing process.
What are the 9 Pillars of I4.0 (BrivaCacs)
Big Data & Analytics Robotics Internet of Things Vertical & Horizontal integration Augmented Reality Cloud Computing Additive manufacturing Cybersecurity Simulation
What Pillars enable twinning and computer-aided predictive capabilities
Big Data & Analytics Simulation Cloud computing
Example of Automation in industry
Car manufaturing Rolls-Royce Engine manufacturing Packaging lines Simple pick and place activities
What is Cobot?
Collaborative robots Not programmed for a specific task, they are trained by humans manipulating the robot arms/parts
How is IoT used in conjunction with Smart factories
Communicate and cooperate with each other in real time across the world, bringing flexibility and resilience
Benefits of IoT in manufacturing industry
Cost Reduction Shorter Time to Market Mass customisation Improve Safety
What is a smart factory?
Highly digitalised, responsive, flexible and interconnected production/manufacturing environment that integrates both physical technology and cyber technology
How does PAT link to smart processing?
Holistic approach to interconnection and digitalisation of the workflows - from product design and development to clients/retail Online control made possible by IOT, smart sensors, online predictive capabilities
Data Science workflow Stage 4
How variables in the data are related to other variables/features via fitting the features/variables around a certain family of models Training datasets = build model Testing dataset = validate model
Why is AR beneficial?
Improving human operators performance in safer environments Providing expert support or contributing to a more effective maintenance and quality assurance process
What are the main 4th Industrial Revolution principles?
Interconnection Information transparency Technical assistance Decentralisation of decision making
What Pillars create a big interconnected network, embedded with sensors for data collection and that enables secure data sharing.
Internet of Things Vertical & Horizontal integration Cybersecurity
When was Industry 2.0 and its key elements?
Late 19th Century Built on mass production, by dividing labour and using electricity.
How are digital twins build
Linking Physical systems with digital twin via data and optimisation by utilising sensors
AI, Machine Learning and Deep learning - How are they related?
Machine Learning can be considered a subset of AI, and Deep Learning a subset of Machine Learning
Key features of a traditional factory
Manual and isolated processes, operations; no integration with different systems and tools. Frequent machine failures and increase maintenance costs. Tied to systems or machines for data, therefore zero or limited data for decision making; process driven-decision making. Limited technology involvement. Zero or limited visibility on operations, productivity data. Limited innovation in production development. Inaccurate asset tracking process and poor resource utilisation. Poor interoperability. The production line is fixed unless manually reconfigured by people with system power down.
Examples of Big data application in industry
Manufacturing - Links with IoT & Cloud Food - Creative recipes Health - Public health sector Bioprocessing - Personalised medication
Data Science workflow Stage 5: Present results
Methods description, Data visualisation Conclusions from the analysis.
What are the main Features of AR?
Mix real-world and digital content Work in real time Use a display device: mixing of the two layers (i.e. real world and digital content)
What is Internet of Things (IoT)?
Network that connects everything, the enabler for all the other technological pillars linked to I4.0 methods Pervasive interconnectivity, allowing data to be shared and send between devices.
What does I4.0 provide?
New levels of flexibility and resilience via Internet of Tings (IoT)
Types of Data
Numeric or continuous - Numbers Categorical or nominal - Attributes or features treated as symbols or names (Colour / Flavours)
What is Augmented Reality?
Overlay information and digital content on the real world, in real time, using a display piece
Elements of the IoT architecture
Sensing layer - perceives and monitors environmental changes (radio-frequency and/or wireless sensors (Antenna)) Network layer - transforms, processes and shares that data gathered via gateway/hub (connectivity and software tools) Application layer - offers intelligent services via Smartphone or dashboard- (monitoring, control or optimisation)
How does digital twin work?
Sensing: sensors allocated in the physical part collect real-time data, that is typically stored in the cloud on in local servers. Simulating/analysing: the twin is then used to evaluate that information, through advanced analysis or simulation. Acting: simulation/analysis results are read and translated into actions, which are sent to the physical system through actuators (valves, temperature controls, etc).
Data Science workflow Stage 6: Deploy Code
Sharing the code and making it accessible for checking/validation
What are the key features in Smart manufacturing?
Smart Factories & Smart Processing Increased level of integration and flexibility across all the steps in the manufacturing supply chain
Examples of IoT applications
Smartphones Smartwatches Smart greenhouses Remote production control
What is Deep learning?
Specialised form of ML that is based on the use of artificial structures know as Artificial Neural Networks (ANN) Mimicking the way human brain works and adapting to new data and obtaining the best outcome without external instructions
