关键词: 风电预测, 神经网络, 花朵授粉算未能, 混沌优化, 风电优化决策

Abstract: Aiming at the disadvantages of conventional Elman neural network such as easily falling into local optimum and insufficient generalization ability, a novel wind power forecasting method combining the flower pollination algorithm and Elman neural network was proposed. By using the logic self-mapping function, the chaotic sequence was constructed, and the chaotic variables were mapped to the solution space of the problem, which made the pollen grains without mutation mechanism had a strong adaptive ability and effectively prevented the convergence of the optimal solution in the later stage of the algorithm, and reduced the probability of the algorithm falling into local extremum by dynamically shrinking the range of the variable coefficients. Combining the forecast demand and network characteristics, the parameters of pollen grains were coded to determine the optimal weights and thresholds of Elman neural network. The numerical example shows that the proposed neural network model has a better predictive effect under the condition of guaranteeing the accuracy of probabilistic prediction, which provides a feasible solution for short-and-medium term wind power forecasting.

Key words: font-size:10.5pt, mso-fareast-font-family:宋体, mso-fareast-theme-font:minor-fareast, mso-ansi-language:EN-US, mso-fareast-language:ZH-CN, mso-bidi-language:AR-SA, ">wind power prediction；neural network；flower polination algorithm；chaos optimization；wind power optimization decision