基于海鸥优化算法的相关向量机模型在径流预测中的应用

Relevance Vector Machine Model Based on Seagull Optimization Algorithm and Its Application in Runoff Prediction

为提高径流预测精度,提出了基于海鸥优化算法(SOA)的相关向量机(RVM)径流预测模型(SOARVM)。选取4个标准测试函数对SOA进行仿真验证,并与PSO算法的仿真结果进行比较;通过主成分分析(PCA)对数据样本进行降维处理,利用SOA优化RVM核宽度因子和超参数,建立SOA-RVM径流预测模型,利用云南省龙潭站年径流及枯水期1~3月月径流预测对SOA-RVM模型进行验证,并将预测结果与RVM、SOA-SVM、SVM、SOA-BP、BP模型进行比较。结果表明,SOA在不同维度条件下仿真效果优于PSO算法,具有较好的寻优精度和全局搜索能力;SOA-RVM模型对实例年径流和1~3月月径流预测的平均相对误差分别为1.77%、4.46%、5.40%、4.03%,预测精度优于RVM、SOA-SVM、SVM、SOA-BP、BP模型。可见SOA-RVM模型具有更好的预测精度,可用于径流预测研究。

To improve the accuracy of runoff prediction,a relevance vector machine(RVM) runoff prediction model based on seagull optimization algorithm(SOA) is proposed.Four standard test functions are chosen to simulate and verify SOA,and compare them with the simulation results of PSO algorithm.The principal component analysis(PCA) is used to reduce dimensionality of the data samples.The kernel width factors and hyper parameters of RVM are optimized by SOA.And then the SOA-RVM runoff prediction model is established.The annual runoff and monthly runoff from January to March during the dry season at Longtan Station in Yunnan Province are used to verify the SOA-RVM model.Compared the prediction results with the RVM,SOA-SVM,SVM,SOA-BP,BP models,the results show that the simulation effect of SOA is better than the PSO algorithm under different dimensional conditions,and it has better optimization accuracy and global search ability;The average relative errors of the SOA-RVM model for the annual runoff and JanuaryMarch monthly runoff are respectively 1.77%,4.46%,5.40%,4.03%,and the prediction accuracy is better than RVM,SOA-SVM,SVM,SOA-BP,BP model.Thus,the SOA-RVM model has better prediction accuracy,and can be used for runoff prediction research.