基于FLS-SVM的地铁车站照明能耗预测  

Prediction of Energy Consumption of Subway Station Lighting System Based on FLS-SVM (Fuzzy Least Squares Support Vector Machine)

作　　者：陈世栋[1] 段中兴[1] 

机构地区：[1]西安建筑科技大学信息与控制工程学院,西安71005

出　　处：《地下空间与工程学报》2014年第4期968-974,共7页Chinese Journal of Underground Space and Engineering

基　　金：国家科学自然基金资助项目(No.51178373)

摘　　要：地铁车站由于其特殊环境,照明主要依靠人工照明,照明工作时间长、功耗大是其主要特点,因此照明系统节能存在较大空间。对照明能耗的预测与分析是进行照明节能改造与设计的前提和基础。模糊最小二乘支持向量机因其有着学习速度快、跟踪性能好、泛化能力强、精度高等优点被广泛应用于能耗预测领域。本文利用模糊最小二乘支持向量机建立能耗预测模型,并且采用MATLAB的LS-SVM工具箱对预测模型进行了仿真研究,最后通过与RBF神经网络能耗预测模型仿真对比试验表明了基于最小二乘支持向量机的能耗预测模型拟合度好、精度高,是照明能耗预测的有效方法。Owing to the specific environment of subway station,where the lighting mainly relies on artificial lighting and the lighting systems have the characteristics of long working time,high running cost and so on. Thus there exists giant potential in the work of lighting system energy saving in the future. Prediction and analysis of lighting energy consumption is the prerequisite and basis of the lighting energy-saving design. The fuzzy least squares support vector machine which has fast learning speed,good tracking performance and generalization ability,is widely used in the field of energy consumption prediction. The prediction model is constructed of lighting energy consumption based on FLS-SVM in this paper; the prediction model is simulated with MATLAB LS-SVM lab Toolbox. By way of the contrast test for prediction model with the RBF neural network energy with FLS-SVM model to show that FLS-SVM model has high fitting degree and high precision and is an effective way of lighting energy consumption forecast.

关 键 词：能耗 预测模型 最小二乘支持向量机 模糊最小二乘支持向量机 RBF神经网络 

分 类 号：TU111.195[建筑科学—建筑理论]