An air balancing method using support vector machine for a ventilation system. (1st October 2018)

• Record Type: Journal Article
• Authors: Jing, Gang; Cai, Wenjian; Chen, Haoran; Zhai, Deqing; Cui, Can; Yin, Xiaohong
• Is Part Of: Building and environment. Volume 143(2018)
• Page Start: 487
• Page End: 495
• Publication Date: 2018-10-01
• Subjects: Ventilation system -- Air balancing -- Support vector machine (SVM) -- Machine learning -- Model-based method -- Parameter identification -- Optimization -- Experiment; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2018.07.037
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: In this paper, support vector machine method is introduced to solve air balancing problem for ventilation systems and a novel non-iterative air balancing method is presented. The proposed method provides a well-formulated form of air balancing problem for air ventilation system by i) building a mathematical model for duct system based on steady-state pressure balance; ii) constructing a pressure prediction model based on the system model and theory of support vector machine; iii) identifying the significant model parameters by support vector machine regression supervised machine lssearning; and iv) determining the damper positions based on the formulated model. The performance of the method is validated experimentally in a duct testbed with five terminals. The results demonstrate that the method can achieve the desired values within 4.6% relative error and is a promising approach for air balancing. Highlights: A novel non-iterative air balancing method is presented. The proposed method provides a well-formulated form of air balancing problem for air ventilation system. The method can achieve the desired values within 4.6% relative error.