Open AccessDemand Response Model f or Duck Curve o n Pv Dominated System u sing Support Vector Machines Based Multistage Modelling
Author(s) -
C. R. Sarin,
Geetha Mani,
Albert Alexander Stonier
Publication year - 2019
Publication title -
international journal of engineering and advanced technology
Language(s) - English
Resource type - Journals
ISSN - 2249-8958
DOI - 10.35940/ijeat.a2957.109119
Subject(s) - photovoltaic system , demand response , cost of electricity by source , grid connected photovoltaic power system , electric power system , electricity generation , computer science , environmental science , power (physics) , maximum power point tracking , engineering , electricity , electrical engineering , physics , quantum mechanics , inverter
Solar Photovoltaic (PV) generation systems have a less Levelized cost of electricity (LCoE). As such, when solar energy is available, the demand response is scheduled in such a way that maximum utilization of solar energy is practised.But the power generation from a solar PV system is highly uncertain and unpredictable due to irregular solar irradiation. Also, the power generation is limited to a time fraction of a day.The impact of these negative traits in a power system is studied with the help of an analytical curve called “Duck curve”. “Solar Duck curve” is a graphical representation of time scaled imbalances between a SPV generation to peak demand. A steep or rugged part in a duck curve indicates sudden shortcoming of SPV generation with respect to the peak demand. Hence, during this period, the loads are shifted between solar PV sources and the main grid with respect to the insufficiency of solar power from peak demand. The proposed system is a machine learning-based multistage demand response system for meeting demand response of a SPV dominant duck curve. The model has four layers/stages.The primary layer is used to analyse the behaviour of the duck curve with the help of a Support Vector regression algorithm and the second layer is used for determining theoperating parameters based on the economic constraints imposed. The third layer is a demand response model based on the previous layer, and the fourth layer is aadaptive signal-processing model used to improve the stability of the system.The obtained demand response model is updated continuously in an adaptive manner so as to improve the stability of the system.A hardware experimental setup is made with eighteen numbers of 24V/2kW interconnected solar PV real-time system which is used for validating and analysing the method.