Research On Central Air Conditioning Load Prediction And Optimal Control Based On Deep Learning

With the rapid development of society and urbanization,the number of buildings is increasing,and central air conditioning,as an essential component of buildings and a major energy-consuming device,is receiving widespread attention and research.There are many devices inside the central air conditioning system and they are coupled with each other,so it is important to study how to save energy and optimize control of this complicated non-linear system.Thus,this article takes the cooling water system of central air conditioning as the target,and projections of air conditioning load by deep learning to achieve energy saving and control optimization of cooling water system.First,this paper explains the significance of research on cooling water systems in central air conditioning systems,reviews the development of air conditioning load prediction technology and the research field of cooling water systems,and analyzes the current research status of domestic and foreign scholars in this field,briefly introduces the concept and development history of deep learning,and explains the principles of several of the classical models to lay the theoretical foundation for the later load predictionSecond,this paper simulates load data by TRNSYS software and performs load prediction based on deep learning techniques.The problem of low prediction accuracy caused by manual empirical selection of network model parameters is solution by intelligent optimization algorithm,which is applications for the parameter search optimization of the model.In this paper,an improved beetle antennae search algorithm is proposed for its shortcomings of low search accuracy,easy to fall into local optimum and not applicable to high-dimensional space,which is small in code size and easy to implement.Through experimental tests,it is proved that the improved beetle antennae search algorithm has better search performance.The improved algorithm is applied to the parameter search of the long and short-term memory network,and the air conditioning load prediction is carried out.It is validated that the improved beetle antennae search algorithm has higher accuracy in load prediction than the recurrent neural network,the long and short-term memory neural network and the particle swarm optimization algorithm.Finally,this paper focuses on the optimal control of cooling water systems.Establish a mathematical model of its energy consumption,use the improved Archimedes optimization algorithm to find the optimal working point with the objective of minimizing the energy consumption of the entire cooling water system,and achieve the energy-saving effect of the air-conditioning cooling water system.Matlab is used to model the controlled objects of the cooling water system,and PID is applied to the cooling water flow control and MFAC is applied to the cooling tower discharge water temperature control,and the improved Archimedes optimization algorithm is used to optimize the controller parameters for the difficult control problem.The experimental results show that the controller has better control performance after parameter optimization.