Research And Implementation Of Online Optimization System Of Building Energy Consumption Based On GA And LSTM-RNN

Building energy saving has always been the key research direction in the field of energy saving.Compared with other energy consumption,the energy consumption of central air-conditioning system accounts for the largest part of the total energy consumption of buildings.Nowadays,the intelligent building system brings us a lot of convenience.When building automation system(BAS)is used to monitor and manage building equipment,a large number of daily operation data of central air conditioning are generated.Without any additional physical hardware and other additional costs,how to effectively use this part of data,mining hidden information from a large number of data itself,to achieve the purpose of central air conditioning energy saving.Most of the existing energy-saving methods of building central air-conditioning are based on building structure and air-conditioning equipment to establish a mechanism model,and then use the mechanism model to predict the energy consumption of airconditioning.Due to the interference of building structure,air-conditioning equipment and external environment,this kind of traditional mechanism model is established under many hypothetical conditions.There are many defects such as the difficulty of modeling,the inaccuracy of the model,the difficulty of reflecting the real situation of the system and so on.It is difficult to apply it to the actual central air-conditioning system.At the same time,energy consumption optimization involves multiple objectives.How to effectively transform the uncertainty of complex multi-objective optimization into a practical single objective optimization problem is an urgent problem in engineering practice.Combined with the practical application of the project,this thesis mainly focuses on the problems of building central air-conditioning energy consumption prediction modeling,target energy consumption optimization and so on.The main contents are as follows:(1)In view of the lack of characteristic attributes and modeling accuracy.This thesis proposes a data-driven modeling method based on LSTM-RNN for energy consumption prediction.This method first preprocesses the data,adds and extracts the key data attributes.Then combined with LSTM-RNN algorithm,the energy consumption prediction model of central air conditioning is obtained.Compared with other algorithms.The method proposed in this thesis has greatly improved the training and testing accuracy.After testing by three practical engineering data the error rate of training set and testing set of the method is controlled at about 1%.(2)Aiming at the problems of unpredictable cooling demand and the uncertainty of multi-objective optimization in the existing cooling water system.This thesis proposes a method to optimize the energy consumption of air conditioning by combining genetic algorithm with air conditioning energy consumption prediction model.Firstly,the cooling water system of central air-conditioning and the feasibility of optimization are analyzed.Then,according to the shortcomings of genetic algorithm and the characteristics of airconditioning,three genetic algorithm strategies are integrated and constraints are added to make the optimization more in line with the actual engineering needs.Compared with other algorithms,the fusion genetic algorithm can not only provide better optimization results,but also has certain advantages in speed,which is more suitable for the actual needs of engineering precision and real-time.(3)This thesis develops a set of human-computer interaction system software for the optimization of central air conditioning energy consumption.The system is seamlessly connected with BAS,and consists of model management,model training,model prediction and other modules.Through the ECS,it integrates air conditioning modeling,scheme optimization,energy consumption prediction and other functions.This system can realize cross regional energy-saving scheduling of central air conditioning through ECS,which not only solves the problem of limited computing power of local computer,but also provides effective energy consumption optimization for central air conditioning across regions.