论文部分内容阅读
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。
In order to improve the prediction model of urban water demand, the relevant vector machine and the differential evolutionary optimization algorithm are integrated and improved, and the water demand forecasting model based on adaptive evolution-related vector machine is proposed. Taking Aksu City of Xinjiang as an example, a prediction model of urban water demand based on adaptive evolution-related vector machines was established and compared with multiple linear regression, BP neural network and support vector machine in accuracy and reliability. The results show that the prediction accuracy of the new model is about twice that of the other methods above. The actual water requirement of the test data is within 95% confidence interval of the adaptive evolution-related vector machine, Error probability judgment model level belongs to “good” level.