One of the biggest milestones in the history of human civilization was the Industrial Revolution which started with the invention of the steam engine in the 1780s. The 19th Century saw the electrification of industries and the use of electrical machines which made the second generation of industries more productive. The third most significant development in the history of industrialization was the invention of computers and the industry realized the potential of software to make machines intelligent empowering them to follow a set of pre-defined instructions with minimal or no human intervention which increased the productivity, reliability, repeatability and quality immensely.
And now we are entering the fourth generation of Intelligent Industries or “Smart Factories” which leverage Cyber Physical Systems, Internet of Things, Connectivity, Cloud Computing, Analytics and Machine Learning to reach an entirely next level of automation and information exchange.
The term “Industrie 4.0” owes its origin to a project from the German Government with the aim to use cutting edge technologies to promote the automation of manufacturing.
Smart factories, leverage connected systems and related technologies to ensure that the machines in the factory work collectively and assist factory operators to increase the productivity of factories, while maintaining best quality outputs. Imagine industrial systems that can continuously monitor and report their own health! Instead of running to failure, machines schedule their own maintenance or, better yet, adjust their control algorithms dynamically to compensate for a worn part and then communicate that data to other machines and the people who rely on those machines.
A Smart Factory uses Cyber Physical Systems which can sense the environment and take decisions on the basis of the sensed stimuli through some kind of control alogithm. These subsystems have their own decentralized processing.After sensing the physical parameters around them, this large chunk of data is processed to extract the required features or information and controls an actuator.The extracted features are communicated to other subsystems and a central supervisory system over the networking infrastructure.The centralized supervisory system receives this information from thousands or millions of end nodes and processes them using powerful analytics and machine learning algorithms running on either high end processors or cloud computing services to recognize patterns, faults, defects or predict a failure in the production line. This helps the management take business decisions on increasing further productivity by studying various patterns as well as mitigate the downtimes caused because of failures.
The Four Pillars of Industry 4.0
- High Quality Measurement Data: Just like the human brain requires information about the environment from the sensory organs, a computer algorithm in an intelligent system needs a large chunk of accurate sensory information like vibration, temperature, strain, pressure, images, voltage, current, power measurements, etc. to take a decision. The end nodes are also responsible for analyzing this huge chunk of analog data and extracting useful parameters to limit the amount of data throughput on the communication bus maintaining its determinism. For instance, the raw vibration data is processed to extract features like true peak, derived peak, RMS magnitude, fundamental frequency, harmonics, rate of change, orders of vibrations, etc.
- De-centralized Decision Making and Control: De-centralized decision making and control enables the industry to avoid the latency involved in passing the operational data from the edge to the centralized processor and back thus increasing the speed of production. This is done using embedded systems trusted with the dedicated task to take an input, process the information, run an algorithm, and generate an actuation deterministically and reliably over and over again.
- Connectivity: The end nodes or cyber physical systems need to inter-operate and hence there needs to be the requirement of a network infrastructure that connects these sub-systems and the central system through reliable and time deterministic wired or wireless standards. Proprietary field buses are one such option that allow data to be transferred from one end to another end in a time bound manner but the problem with these proprietary buses is that it cannot be integrated over the existing IT infrastructure, i.e., ethernet available in the industries. An alternative can be the usage of Time Synchronized Networks (TSN) technology which is a set of standards under development by the Time-Sensitive Networking task group of the IEEE 802.1 working group. There is also a need to publish the reports or the data remotely over the web to allow remote monitoring through a computer, tablet, or a mobile phone around the world.
- Big Data Analytics and Machine Learning: The concept of intelligence inherently means the ability to self-learn and diagnose problems. These intelligent central systems should act as the Central Nervous System of the human body and get processed information from thousands or millions of neurons (cyber physical systems) existing at the edge and be a power house for powerful analytics and machine learning algorithms to learn from the patterns and do prognostics on the data and detect potential problems before they occur or take business decisions. This is primarily done using powerful processors or Cloud Computing.