基于改进差分进化算法的RBF神经网络在股指预测中的应用

Application of RBF neural network based on improved differential evolution algorithm to stock index forecasting

为了提高径向基函数(RBF)神经网络的预测性能,文章提出改进的差分进化算法(IDE),通过引入混合变异策略和局部算子来增强算法的收敛速率和局部搜索能力,用改进的差分进化算法对径向基函数神经网络的网络结构参数进行优化,建立了IDE-RBF神经网络股指预测模型,并以上证综指为例进行了实证分析。实证结果表明,IDE-RBF神经网络的预测效果明显优于其他预测模型。

To improve the forecasting performance of radial basis function （RBF） neural network ,an improved differential evolution （IDE） algorithm is proposed by introducing the hybrid‐mutation strate‐gy and the local operator ,w hich greatly enhance the convergence rate and the ability of local search of the algorithm .Then the IDE algorithm is used to optimize the parameters of the structure of the RBF neural network and an IDE‐RBF stock index forecasting model is built .Finally ,an empirical analysis is conducted by using Shanghai Composite Index .The empirical results show that the IDE‐RBF neural network performs better than other forecasting models .