Abstract: Under the partial shading condition (PSC), the power-voltage (PV) characteristic output curve of photovoltaic system will have multiple peak points. In order to solve this problem, the traditional maximum power point tracking (MPPT) algorithm is no longer applicable because it is easy to fall into the local optimal point and has poor stability.In order to achieve maximum power point tracking under partial shadow shadowing, a dynamic surrogate model based optimization (DSMO) method for photovoltaic system is designed in this paper. In order to avoid blind search, the radial basis function network is adopted to construct the dynamic agent model of input/output feature base on the real-time data of PV system. Based on the dynamic agent model, greedy search is used to accelerate convergence. In this paper, the practicability and superiority of the method are evaluated by three examples, including constant temperature and constant light intensity start-up experiment, constant temperature light intensity step change and variable temperature light intensity. Compared with ant colony algorithm (ASO), gray wolf optimizer (GWO), perturbation and observation method (P&O) and particle swarm optimization algorithm (PSO), the DSMO method proposed in this paper can quickly and stably generate more energy and smaller power fluctuation in the photovoltaic system under partial shading conditions.

Key words: photovoltaic system, partial shading conditions, maximum power point tracking, dynamic surrogate model, greedy search