论文部分内容阅读
通过14个复杂函数对一种新型仿生群体智能算法——鸡群优化(CSO)算法进行仿真验证,并与粒子群优化(PSO)算法进行对比。针对支持向量机(SVM)学习参数难以确定的不足,利用CSO算法搜寻SVM最佳学习参数,提出CSO-SVM预测模型,并以云南省某水文站枯水期1—3月月径流预测为例进行实例研究。结果表明:CSO算法收敛精度完全优于PSO算法,具有较好的计算鲁棒性和全局寻优能力;CSO-SVM模型预测精度优于PSO-SVM模型,利用CSO算法寻优SVM学习参数可有效提高SVM模型的预测精度和泛化能力。
Fourteen complex functions were used to simulate a novel bionic group intelligence algorithm - flock optimization (CSO) algorithm and compared with Particle Swarm Optimization (PSO) algorithm. In view of the difficulty of determining the learning parameters of support vector machine (SVM), the CSO algorithm is used to search the best learning parameters of SVM and the prediction model of CSO-SVM is proposed. The monthly runoff prediction of January to March in a dry season of a hydrometric station in Yunnan Province is taken as an example the study. The results show that the accuracy of the CSO algorithm is better than that of the PSO algorithm, and it has better computational robustness and global optimization ability. The prediction accuracy of the CSO-SVM model is better than that of the PSO-SVM model. Using the CSO algorithm to optimize SVM learning parameters is effective Improve the prediction accuracy and generalization ability of SVM model.