Parameter estimation and power prediction for pv power generation using a multi-agent algorithm

The rapid development of smart grids signifies that accurate prediction of photovoltaic (PV) power generation allows efficient load demand scheduling and reduces the impact of uncertainty for power dispatch. The PV power output, however, is not easy to accurately predict because of its intermittent and random characteristics. Parameters such as the generated photocurrent, the reverse saturation current, the series resistance, the shunt resistance and the diode ideality dominate the power output of solar cells. These parameters have a nonlinear relationship with power output and can vary when the PV cells age. To improve this problem, this paper first uses Matlab/Simulink to establish the PV power generation model using a single-diode circuit. A multi-agent algorithm that uses an enhanced charged system search (ECSS) is then used to estimate the optimal parameters and to predict the PV power output. The proposed method is tested on a 32kWp PV power generation system. To verify that the proposed method is optimal, the results are compared with those for the traditional differential evolution (DE) and particle swarm optimization (PSO) methods.