Research On Coordinated Control Methods Of HVAC Systems And Personal Comfort Systems For Smart Office Buildings

With the development of the times and the progress of science and technology,people are increasingly pursuing a more comfortable office environment.In the thermal comfort control system of traditional office building,only experienced HVAC engineers can select the appropriate but mechanical temperature set-points according to specific office environment.The above method is often difficult to meet personal thermal comfort demands of office users and may cause a great waste of energy.This phenomenon promotes the innovation and development of smart office building thermal comfort control technology.With the rise of Internet of Things technology and artificial intelligence,some novel control methods that using personal comfort systems to refine the microenvironment of users for providing personalized thermal comfort show some advantages in the field of reducing energy and maintaining thermal comfort.Existing works show that the coordinated control of HVAC systems and personal comfort systems contributes to improve thermal comfort and reduce system energy consumption/energy cost.However,existing works need to know explicit building thermal dynamic models.In fact,due to the influence of complex and random factors(e.g.,building structures and materials,outdoor temperature and humidity,solar radiation intensity,thermal gain from users and devices),it is very challenging to obtain explicit building thermal dynamic models that are accurate and efficient enough for building control.Therefore,how to optimize the coordinated scheduling of two systems without knowing explicit building thermal dynamic models is worth studying.This thesis first formulates a dual-objective minimization problem of user temperature deviation and energy cost(i.e.,HVAC systems and personal comfort systems)in multi-user shared office area.Due to the difficulty of obtaining the explicit building thermal dynamic models,uncertainty system parameters,and a large discrete solution space,it is very challenging to solve the optimization problem.We reformulate the optimization problem as a Markov game and proposes a coordinated control algorithm of HVAC systems and personal comfort systems based on the framework of multiagent actor-attention-critic.Compared with several typical methods,the proposed algorithm can reduce the building energy cost by 5.99%-22.38% under the premise of enhancing thermal comfort of users.Moreover,this thesis further formulates a dual-objective minimization problem of user temperature deviation and energy cost(i.e.,smart blind systems,HVAC systems,and personal comfort systems)in smart office building.Due to the coupled constraints brought by blind systems,the difficulty of obtaining the explicit building thermal dynamic models,uncertainty system parameters,and a large discrete solution space,it is very challenging to solve the problem.To overcome above challenges,we adopt a smart office building multi-system thermal comfort control algorithm based on the framework of multi-agent deep reinforcement learning.The proposed algorithm can coordinately control multiple systems in office buildings and achieve an optimal tradeoff between reducing energy cost and improving thermal comfort of users.Simulation results based on real traces show that the proposed algorithm can reduce the total energy cost of office buildings under the premise of satisfying thermal comfort demands of users.Finally,this thesis summarizes the content of full text and points out the future research directions.