组合预测方法在建筑能耗预测中的应用

发布时间：2018-06-08 02:24

本文选题：建筑能耗预测 + 组合预测　；参考：《福州大学》2014年硕士论文

【摘要】：建立反映建筑能耗变化的预测模型,是进行建筑能耗分析的一个重要基础,也是对建筑能耗进行有效管理的必要前提。建筑能耗系统是一个动态的,不确定的,随机的非线性模型,传统方法很难实现建筑能耗的准确、快速预测。因此本文在分析建筑能耗特性的基础上,提出组合预测方法对建筑能耗预测进行研究,主要研究内容从三个方面展开：建筑能耗各影响因子之间存在严重的相互关联,传统建模方法无法消除各影响因子之间的冗余,预测结果不够理想。为了提高建筑能耗预测精度,本文提出主元分析和RBF神经网络相结合的建筑能耗预测方法。通过主元分析降低网络输入维数,使得网络的学习性能和泛化性能得到增强,从而提高了RBF神经网络的预测精度。实例表明,改进后的模型可以有效地提高建筑能耗的预测精度。针对建筑能耗数据存在噪声和变量间存在多重相关性问题,提出了KPCA-WLSSVM建筑能耗预测方法。首先对能耗数据样本进行核主元分析,降低样本数据间的噪声和多重相关性。然后根据每个训练样本对预测结果的影响程度不同,赋予每个样本不同的权重系数。最后采用粒子群算法对WLSSVM的正规则参数和核参数进行优化,建立了具有良好推广性能的组合预测模型。实例表明,本文所提出的KPCA-WLSSVM建筑能耗预测模型具有良好的预测效果。为了提高高校建筑的能耗预测精度,在比较传统灰色预测模型和神经网络预测模型优缺点的基础上,建立了灰色RBF神经网络能耗预测模型。该方法综合了灰色系统理论所需数据少以及神经网络自学习和自组织的优点,更加有效地利用样本数据的有效信息,提高了模型预测精度。仿真实例证明了该方法在高校能耗预测中的可行性和有效性。研究工作表明,本文所提出的组合预测方法均比单一的能耗预测方法具有更好的预测效果,对于改善建筑综合设计效率、能耗管理水平具有重要意义。
[Abstract]:It is an important foundation for building energy consumption analysis to establish a prediction model to reflect the change of building energy consumption. It is also a necessary prerequisite for effective management of building energy consumption. Building energy consumption system is a dynamic, uncertain, stochastic nonlinear model, the traditional method is difficult to achieve accurate and rapid prediction of building energy consumption. Therefore, based on the analysis of the characteristics of building energy consumption, this paper puts forward a combined forecasting method to study building energy consumption prediction. The main research contents are as follows: there is a serious correlation among the factors affecting building energy consumption. The traditional modeling method can not eliminate the redundancy among the influence factors, and the prediction results are not satisfactory. In order to improve the prediction accuracy of building energy consumption, a method of building energy consumption prediction based on principal component analysis and RBF neural network is proposed in this paper. By reducing the input dimension of the network by principal component analysis, the learning performance and generalization performance of the network are enhanced, and the prediction accuracy of the RBF neural network is improved. The example shows that the improved model can effectively improve the prediction accuracy of building energy consumption. Aiming at the problem of multiple correlation between noise and variables in building energy consumption data, a KPCA-WLSSVM building energy consumption prediction method is proposed. Firstly, the kernel principal component analysis is used to reduce the noise and multiple correlation between the energy consumption data samples. Then according to the different degree of influence of each training sample on the forecast result, each sample is given different weight coefficient. Finally, the particle swarm optimization algorithm is used to optimize the normal parameters and kernel parameters of WLSSVM, and a combination prediction model with good generalization performance is established. The example shows that the KPCA-WLSSVM model has good prediction effect. In order to improve the prediction accuracy of building energy consumption in colleges and universities, on the basis of comparing the advantages and disadvantages of traditional grey prediction model and neural network prediction model, a grey RBF neural network energy consumption prediction model is established. This method integrates the advantages of less data needed by grey system theory and neural network self-learning and self-organization. It makes more effective use of the effective information of sample data and improves the prediction accuracy of the model. A simulation example shows the feasibility and effectiveness of this method in the prediction of energy consumption in colleges and universities. The research results show that the combined forecasting method proposed in this paper has better prediction effect than the single energy consumption prediction method. It is of great significance to improve the efficiency of building comprehensive design and the level of energy consumption management.
【学位授予单位】：福州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU111.195

【相似文献】