Machine learning: A key towards smart cyber-physical systems

Machine learning is one of the important components of cyber physical systems. Either development or implementation about cloud computing or edge computing, machine learning, deep learning and AI techniques are key to develop a smart cyber physical system. Machine learning has applications in almost all streams of engineering. It is considered as a subset of Artificial Intelligence (AI) but can be seen as an extension of AI that has broadened the application areas of AI. AI was initiated to make computers mimic human behaviour, considering human to be most intelligent creature on the earth. Although, later when discussions concluded that human behaviour cannot always be considered as intelligent, modified to "mimicking ideal human behaviour". Modern AI techniques are extracting intelligence not only from human but also from many other creatures like Ant, Bee, Monkey, birds, fishes and many more. Machine learning added a new feature to AI by trying to make our computer system learn from the data they are receiving. By using Machine Learning techniques, the computers are now not only processing the data, but also extracting the information from that data to give us better results in every next iteration. Two major thrust research areas in Electrical engineering are Smart Grid and Electric Vehicles. Both these areas are applied to make the power systems and power electronic converters smarter with help of smart inter-disciplinary techniques like machine learning, deep learning, different AI optimization tools etc. So, it is the need of the day for the researchers from different streams to equip themselves with these modern tools. Machine learning can be applied to various classification applications in electrical engineering like detecting power system faults, transformer faults, machine health monitoring etc. It can also be applied for regression applications like solar radiation prediction, selective harmonic elimination for multi-level inverters, electrical load forecasting etc. In this chapter basics of machine learning, different machine learning algorithms, stages in machine learning based implementations are discussed. At last Electrical engineering related applications of this mezzanine technology are discussed. Two, end to end applications one in power system and another in power electronics are also covered. Implementation on MATLAB as well as Python platforms are demonstrated through simple examples. Various Hardware's that supports machine learning in a cyber-physical system are also discussed and Raspberry-Pi, as a tool for development of machine learning based cyber physical systems is also demonstrated through example.