Artificial intelligence power controller of fuel cell based DC nanogrid. (September 2020)

• Record Type: Journal Article
• Authors: Kumar, Saurabh; Krishnasamy, Vijayakumar; Neeli, Satyanarayana; Kaur, Rajvir
• Is Part Of: Renewable energy focus. Volume 34(2020)
• Page Start: 120
• Page End: 128
• Publication Date: 2020-09
• Subjects: Boost converter -- Binary genetic algorithm -- Fuel cell -- Nanogrid -- Real-coded genetic algorithm; Renewable energy sources -- Periodicals; Solar energy -- Periodicals; 333.79405
• DOI: 10.1016/j.ref.2020.05.004
• Languages: English
• ISSNs: 1755-0084

Abstract

Highlights: Standalone nano-grid is a simple and feasible solution to power the rural areas. Boost converter based interfacing unit with an intelligent controller is proposed. Intelligent controller improves the dynamic response of the system and regulates the voltage. PI parameters are tuned using a binary genetic algorithm and real-coded genetic algorithm. Abstract: The expansion of DC nanogrid is projected in rural and remote areas for electrification. Conventional way of powering the rural areas is to extend the central grid to that area. However, extending the network of central grid is a cumbersome process and it involves huge capital cost. Further, with the development of power electronics and evolution of renewable resource, a standalone nanogrid is a simple and feasible solution to power the rural areas. Therefore, a clean and sustainable fuel cell based nanogrid is proposed as a step towards rural electrification. The output voltage of the fuel cell is varying in nature which could damage the voltage sensitive electronic equipment of the house hold. Therefore, boost converter based interfacing unit with an intelligent controller is proposed in this paper to regulate the voltage. The intelligent controller improves the dynamic response of the system and regulates the voltage at the DC distribution bus irrespective of source side or load side disturbances. The proposed intelligent proportional integral and derivative controller parameters are tuned using a binary genetic algorithm and real-coded genetic algorithm by formulating an optimal control problem. The proposed controller is model independent, simple in implementation and gives an excellent dynamic response.A laboratory scaled hardware prototype is developed to validate the performance of proposed controller.