The processes of Smart Engineering and thus of Industry 4.0 use the Internet of Things (IoT) as one of their core technologies. Vast amounts of data are generated by sensors and actuators in the IoT. We also call this huge amount of data „Big Data“.
Big Data consists of extensive data sets in the characteristics of volume, variety, velocity, variability – the 4 Vsvital of data management. Particular system architectures are required for efficient storage, manipulation and analysis of such large amounts of data. Let’s look at the 4 Vs in detail: Volume means that we have large amounts of data to collect, store, retrieve, process and update. Typically, they are generated by social media or IoT sensors. Velocity means that the data is captured in real time and at high speed. Typical data sources are
E-business, machines, social media, human interactions. Here, the speed of processing is important for business success. Variety means that data comes from different sources. They are large in dissimilarity and complexity. Besides, they can be structured or unstructured. Structured data would be: Machine temperature, running performance information, operating hours of the turbine blades. Unstructured data could be social media with text, video streams, information on airbag deployment in the event of a car accident. Veracity means that the quality of the data is different. This affects the accuracy of data analysis.
There are uncertainties in the data. Data may contain irregularities, noise, or dirt. To increase veracity, the data must be cleaned up. The components of the Big Data lifecycle are collection, preparation, analysis and action. Collecting means storing raw data. Preparation involves cleansing and transforming raw data into organised information. Analysis Creates knowledge from information. Finally, Action generates added value for businesses from knowledge.
Data analyses that are most important for Smart Engineering include descriptive, diagnostic, predictive and finally prescriptive Analytics. What is the role of analytics in the Smart Engineering/IoT context? Analytics is important to IoT because making sense from an endless stream of data from sensors is practically impossible without data analytics.
Analytics is very critical to IoT and Smart Engineering. Reasons for this fact are: Rapidly changing real-time data are „in constant motion“. Since IoT data have a limited shelf life, they disappear quickly. And because an acceptable response is required, a just-in-time response is required.
What makes then IoT analytics really different? There are 2 important, major differences:
Edge analytics are distributed analytics. Means high-velocity data is processed near the sensors, The reason for this measure is, that regular cloud-based analytics is too slow for IoT
With IoT we have bidirectional communication and control of the endpoints because “true” IoT has to be bidirectional, e.g. in order to control actors.
How to applyArtificial intelligence in IoT / smart engineering analytics? First we should ask ourselves “What is Artificial Intelligence (AI)?” AI is the intelligence of machines and the branch of computer science that aims to create it. It is the study and design of intelligent agents. An intelligent agent is a system that perceives it’s environment and takes actions to maximize the chances of success. There are many different branches of artificial intelligence (see the attached image). Let us focuss on machine learning and its relevance to IoT and smart engineering.
Learning is any process by which system improves its performance from experience. Machine Learning is concerned with computer programs that automatically improve their performance through experience. It deals with the design of systems that can learn from data, rather than follow only explicitly programmed instructions. Here are 2 examples for learning by doing
Robot: A robot that can act in a world, receiving rewards and punishments and determining from these what it should do.
Chess: A (robot) master chess player makes a move using reinforcement learning.
What is then deep learning? It is a collection of algorithms used in machine learning in order to model high-level abstractions in data through the use of model architectures, which are composed of multiple nonlinear transformations. It is based on learning representations of data.
Artificial intelligence (AI) combined with IoT provides framework and tools for highly sophisticated loT, real-time decisions and automation use cases. An interesting example is an autonomous IoT vacuum cleaner that adopts machine learning. It learns the home layout and remembers it. It adapts to different surfaces or new items and Improves on movement pattern for efficiency. It knows when to recharge and automatically docks itself and employs piezoelectric, and optical sensors. And, it employs machine learning to adapt and improve.
Auszug aus einem Blogbeitrag von Prof. Leisenberg im Projekt SMeART