Complexity analysis and verification of real-time operation for a semi-Markov model of photovoltaic intermittency

Photovoltaic (PV) penetration levels in the power grid have significantly increased during the last few years. However, issues such as cloud-induced intermittency in PV generation forces equipment on the electrical grid to cycle excessively, preventing PV generation from being a reliable or dispatchable source of power, particularly for utilities. To mitigate this intermittency, the PV must be coordinated with either dispatchable generation, energy storage, or demand response. In order to do so economically, it is necessary to develop a model of PV power that can be included in design and control problems. A semi-Markov process model is proposed for PV power. Unlike existing models of PV power, the proposed model has a wide range of applicability across both small and large timescales. These applications include simulating PV power, short-term forecasting of PV power, design of rule-based controllers for energy storage units (ESU), and stochastic scheduling of ESU in conjunction with other resources. This paper investigates the suitability of the model implemented in algorithms in real-time. To this end, the complexity of four algorithms involving the model is analyzed in terms of the number of operations required vs. number of samples processed. This is validated by implementing the algorithms on a representative target platform, demonstrating that they can complete within reasonable deadlines.