A neural network control strategy for Multi-energy common dc bus hybrid power supply

Abstract

Solar energy, wind energy, ocean energy and geothermal energy are the reproducible energy source. The renewable energy, which is used to generate electricity, not only is clean and pollution-free, but also has a large reserve. Therefore, it becomes an important energy in the world. Multi-energy hybrid power supply system based on solar energy, wind energy, ocean energy and geothermal energy is a tendency. But in most cases, a multi-energy hybrid power supply system is bothered by the uncertainty of wind resources and non-continuity of solar power without high-powered control strategy. To provide a sustained, stable and continuous power to customers, a structure of multi-energy hybrid power supply system is established and a neural network model is developed in this paper. A control strategy for multi-energy hybrid power supply systems is proposed by analyzing the characteristics of solar energy, wind energy, ocean energy and geothermal energy. In order to accelerate the convergence and to prevent oscillation, Levenberg-Marquaret algorithm is used and the momentum factor α is introduced in the training. Simulation shows that this strategy can make the voltage sustained, stable and continuous in the maximum use of renewable energy, which can be satisfied to customers’ electricity need.