论文部分内容阅读
采用牛津大学Angeliki Xifara使用Ecotect系统模拟的768个不同建筑物数据,尝试将半参数中的部分线性单指标模型(PLSIM)用于住房建筑物负荷的预测研究中。同时采用BP神经网络以及迭代加权最小二乘法分别建立热负荷、冷负荷预测模型,将3种方法所得结果进行比较。研究结果表明部分线性单指标模型在建筑物负荷预测中相对误差均在0.00104以内且更直观,可以为国家调整住房结构、节约能源提供有力的模型支持。
Using the data of 768 different buildings simulated by Ecotect system by Angeliki Xifara of Oxford University, the partial linear single-index model (PLSIM) of semiparameters is tried to predict the load of housing buildings. At the same time, BP neural network and iterative weighted least square method are respectively used to establish thermal load and cooling load forecasting models. The results of three methods are compared. The results show that some of the linear single-index models have a relative error of less than 0.00104 in building load forecasting and are more intuitive, which can provide powerful model support for the state to adjust housing structure and save energy.