论文部分内容阅读
文章结合人工蜂群算法的自适应度因子对传统SVR模型的循环求解进行改进和优化,提高传统模型收敛和求解精度,并将改进SVR模型用于新疆喀什某冰川河流的年径流预测中。研究结果表明:在AIC计算准则下,改进的SVR模型最小计算值最小,参数结果更为合理;相比于传统模型,改进模型在区域冰川河流年径流预测精度得到较为明显的改善和提高,其中预测的误差均值减少16.3%,年相关系数提高0.25;建立的自回归方程可对未来冰川径流的变化趋势进行预测。研究成果对于新疆冰川河流年径流的变化预测提供方法参考。
Based on the adaptive factor of artificial bee colony algorithm, the article improves and optimizes the traditional SVR model, and improves the convergence and accuracy of the traditional model. The improved SVR model is applied to the annual runoff prediction of a glacier river in Kashgar, Xinjiang. The results show that under the AIC calculation criterion, the minimum calculated value of the improved SVR model is the smallest and the parameter results are more reasonable. Compared with the traditional model, the accuracy of the improved model in the regional glacier river runoff prediction is improved and improved obviously. The predicted error mean decreases by 16.3% and the annual correlation coefficient increases by 0.25. The established autoregressive equation predicts the future trend of glacier runoff. The research results provide a reference for the prediction of the annual runoff change in the glacial and glacial rivers in Xinjiang.