基于改进和声搜索算法的支持向量机参数优化

SVM Parameter Optimization Based on Improved Harmony Search Algorithm

提出了一种改进和声搜索算法以优化支持向量分类机的参数,该方法可以随着迭代次数的增加自适应动态调整参数。在核电站电动隔离阀门故障分类预测的仿真实验中,将该方法与经典智能算法优化参数的支持向量分类机对比,基于改进和声搜索算法的支持向量分类机具有更高的分类精度和泛化能力。

The method of improving harmony search algorithm to optimize the parameters of support vector classification machine was proposed. This method can respond to the dynamic-regulation parameters with the increase of iterations. In simulation experiment of the dynamoelectric isolation valve's fault classification and prediction in the nuclear power plant,having this method compared with the classical parameter optimization algorithm shows that this support vector machine based on the improved harmony search algorithm has better classification accuracy and generalization ability.