数据驱动的空调系统故障诊断算法综述

A Review of Data-driven Fault Diagnosis Algorithms for Air Conditioning Systems

随着经济发展和人们对生活环境品质要求的提高,越来越多的建筑安装了中央空调系统,而空调系统的故障会造成能耗浪费和室内环境品质下降,无法满足室内人员的热舒适需求,产生热抱怨。同时,随着自控技术的越来越多地应用于中央控制系统,传感器数量以及数据海量增加,使得基于白箱的故障诊断算法难以满足空调系统故障诊断的需求。以机器学习为主的人工智能技术不断地发展,为故障诊断算法带来了新的思路。本文全面综述了近二十年来基于数据驱动的故障检测和诊断(FDD)算法在暖通空调(HVAC)系统中应用的研究。由于数据驱动存在依赖与数据且模型解释性不足的问题。对此,最近越来越多的研究开始融合物理模型和数据驱动方法,旨在提升解释性、准确性并减少对大数据依赖,这将成为未来故障诊断研究的新方向,以实现更高效准确的诊断。

With the economic development and the growing demands for better living environment quality,more and more buildings have installed with central air conditioning systems.However,faults in these systems lead to energy waste,decline in indoor environmental quality,and even failure to meet the thermal comfort needs of occupants,resulting in thermal complaints.Concurrently,the increasing application of automatic control technology in central control systems,along with the surge in the quantity of sensors and voluminous data,have made it challenging for traditional white-box fault diagnosis algorithms to fulfll the requirements for air conditioning system fault diagnosis.The continuous development of artificial intelligence technology,particularly machine learning,has brought about novel approaches to fault diagnosis algorithms.This paper provides a comprehensive review of the research conducted over the past two decades on data-driven fault detection and diagnosis(FDD)applied to HVAC systems.Due to the issues of data-driven methods,such as their dependency on large-scale data and insufficient interpretability of their models,there is a growing trend to integrate physical models with data-driven methods.The aim of this integration is to enhance interpretability,improve accuracy,and reduce the dependency on large-scale data.It is expected that this will emerge as a new direction for future research on fault diagnosis,with a view to achieving more efficient and accurate diagnoses.