The future of electronics is to simulate the human brain: we study how to replicate the functioning of biological neurons

Self-learning, self-adaptation and self-organization – these are the key characteristics missing in current computer technologies but present in the human neural system. This reflection marked the new beginning from which artificial intelligence techniques, such as expert systems, fuzzy logic, artificial neural networks (ANN) and genetic algorithms, recently widely applied in power electronics and motor drives, with the aim of building a control system, in machines, equipped with ahuman-like intelligence. While expert systems and fuzzy logic are mostly based on standard rules and procedures, ANNs, born in the 1940s, are of a more generic nature and tend to directly emulate the biological neural network, a characteristic that in the 2000s brought the sector artificial intelligence to see them with renewed interest.

The ANN can be seen as an automatic copy of a biological neuronan autonomous system capable of performing different functions at the same time, delegating them to the parts that compose it, and they are of two types: feedforward and feedback. Most applications in power electronics have adopted a feedforward ANN, but the feedback ANN is used to precisely control and monitor motors using motor drives. The fields of application of the ANNs, as reported by Powerelectronicnewsrange from the sector of renewable energies, such as grid-connected inverters and solar photovoltaic inverters, to intelligent charging systems for electric vehicles. In the first case, ANN networks are used to improve the design, operation and maintenance of photovoltaic cells. Traditional PV controllers use PI controllers or PR algorithms, which sometimes are slow in their response to sudden disturbances. In grid-tied operations, these interferences are quite frequent and the controllers show their weakness by not being able to handle these unexpected events and causing loss of efficiency and operating precision. When for AI algorithms are added to the control center, the response time to disturbances and the accuracy of the converter are greatly improved.

In electric vehicle charging stations, artificial neural networks optimize charging by monitoring efficiently the flow of current, the type of battery and other charging parameters to speed up the operation times; the continuous monitoring, moreover, will also help to monitor the state of degradation of the cells and to prevent breakdowns.