基于自适应进化相关向量机的城市需水量预测模型研究  

Research on urban water demand forecast model based on adaptive evolution relevance vector machine

作　　者：徐继红[1] 

机构地区：[1]新疆塔里木河流域管理局,新疆库尔勒841000

出　　处：《水资源开发与管理》2016年第1期45-48,52,共5页Water Resources Development and Management

摘　　要：为改进城市需水量预测模型,将相关向量机与差分进化优化算法进行融合及改进,提出基于自适应进化相关向量机的需水量预测模型。以新疆阿克苏市为例,建立基于自适应进化相关向量机的城市需水量预测模型,并与多元线性回归、BP神经网络、支持向量机算法在精度与可靠性方面进行对比分析。结果表明:新模型预测精度大约是上述其他方法的2倍以上;测试数据的实际需水量均在自适应进化相关向量机估计的95%置信度的置信区间内,并且由后验差比、小误差概率判定模型等级属于"好"级别。The relevance vector machine and differential evolution optimization algorithm are converged and improved in order to improve urban water demand forecast model. Water demand forecast model based on adaptive evolution relevance vector machine is proposed. Aksu in Xinjiang is adopted as an example. Urban water demand forecast model based on adaptive evolution relevance vector machine is established. It is comparatively analyzed with multiple linear regression,BP neural network and support vector machine algorithm in terms of accuracy and reliability. The results show that new model forecast accuracy is about more than 2 times compared with other above-mentioned methods. Actual water demand of test day is in the confidence level of 95% confidence estimated by adaptive evolution relevance vector machine. It is determined that the model level belongs to ‘good'level through posteriori difference ratio and small error probability.

关 键 词：城市需水量 预测 自适应进化 相关向量机 

分 类 号：TU991.31[建筑科学—市政工程]