Improved particle filter based soft sensing of room cooling load

• Published in: Energy and buildings Vol. 142; pp. 56 - 61
• Main Authors: Zhanpei, Li, Tingzhang, Liu
• Format: Journal Article
• Language: English
• Published: Lausanne Elsevier B.V 01.05.2017; Elsevier BV
• Subjects: Air conditioners; Air conditioning; Computer simulation; Cooling; Cooling load; Decomposition; Deep learning; Detection; Energy balance; Energy conservation; Energy management; Energy measurement; Machine learning; Parameter identification; Particle filter; Real time; Soft sensing; State estimation; Deep learning; Soft sensing; Energy conservation; Cooling load; Particle filter
• ISSN: 0378-7788; 1872-6178
• DOI: 10.1016/j.enbuild.2017.03.010

Accurate value of room cooling load is the basis data for energy conservation and demand response in the air conditioning operating process. Unfortunately, traditional calculation methods for room cooling load are too complex and time-consuming to meet the demand of real-time control. Soft sensing is an attractive technology, however the room cooling load cannot be measured directly, so the parameter identification for soft sensing model cannot be realized based on the sample data. Aiming at the bottleneck problem in room cooling load soft sensing, an improved particle filter based method is presented in this article. Firstly, a model for room cooling load soft sensing is built with analysis of room energy balance equation. Then frequency domain decomposition is employed for rough measurement, and deep learning is employed for prediction. Finally, the improved particle filter is employed to realize the real-time state estimation of room cooling load. In the process of particle filter, the artificial fish swarm algorithm is introduced to overcome the sample impoverishment problem of traditional Re-sample method. The simulation experiments and real-test experiments show that the proposed method can realize the soft sensing of room cooling load quickly and efficiently, which method also provide references for the soft sensing of other un-measurable variables.