| As the proportion of building energy consumption in the national energy consumption is increasing year by year,how to achieve energy conservation in buildings through effective energy management and reduce the impact of energy consumption on the environment has attracted widespread attention.As the central air-conditioning system is one of the most important energy-consuming equipment in the building,it is of great significance to study its load characteristics and optimize the system operation strategy for reducing building energy consumption and reducing operating costs.Adjusting the operation strategy according to the actual load demand can greatly reduce the energy consumption of the central air-conditioning system.However,the central air-conditioning system is a nonlinear and complex system.It is affected by many factors and there are interactions between the system equipment.Its operation optimization is very challenging.In this study,the machine learning algorithm was used to predict the cooling load of an air-conditioning system in a hospital in Shanghai,and data energy consumption modeling and operation strategy optimization were performed on the air-conditioning system based on historical operating data.First of all,the collected historical operating data is sorted and analyzed.After preprocessing such as abnormal point culling,wavelet analysis is used to reduce the noise.Three methods were used to learn the data set after noise reduction,and three different load prediction models were obtained.Taking the predicted load data as the goal to optimize the operation of the chiller,a multi-objective optimization method based on the Lagrange multiplier method was proposed,and the optimized operation plan of the chiller was obtained.Afterwards,the pumps and cooling towers were regression modeled.Considering the increase of equipment and constraints,the genetic algorithm was used to optimize the cooling water system.Finally,the cooling tower heat release was used as the constraint,coupling the cooling water system and the chilled water system.The optimized operation plan of the system.In order to predict the cooling load of central air conditioning,multiple nonlinear regression,support vector machine regression,and long and short memory neural networks were used to establish the load forecasting model.For the first two models,the energy coefficient was used to process the load data,which improved the accuracy of the model;The third model,using the Adam algorithm to optimize the neural network,improved the training speed of the model.After comparing the prediction accuracy and modeling time of the three models,the results show that,compared with the previous two models,although the training time of the long and short memory neural network model is much higher than the other two,it has better prediction accuracy.By analyzing the variable flow system of the secondary pump,the operation energy consumption model of the single chiller and the operating constraints were obtained.According to the regression,the energy consumption model parameters were obtained.The accuracy of the energy consumption regression models of the two chillers is above 99%.Then the Lagrange function was constructed,the predicted cooling load was used as input,and the optimal chiller operation scheme was obtained by solving the KKT condition.After that,the algorithm was used to optimize the five working conditions.Compared with the original operation plan,it was found that when the temperature difference is the minimum,the chiller can save up to 16%,and the greater the load,the greater the energy saving potential.The energy consumption regression model of the pump and cooling tower was established,and the model parameters were solved.The accuracy of the obtained model was about 97%.At the same time,the heat release model of the cooling tower was established.Due to the poor quality of the flow data,the accuracy of the model is only 82.4%.It is observed that the energy consumption of the chiller and other equipment are quite different.The energy consumption of the system is divided into two parts: the chilled water system and the cooling water system.The genetic algorithm was used to optimize the energy efficiency of the 5 sets of historical working conditions of the cooling water system.The results show that the optimization scheme can reduce the energy consumption of the cooling water system by up to 11.6%.Finally,the chilled water system and the cooling water system was coupled.When the optimal operation of the chiller is satisfied,the energy consumption of the cooling water system will increase accordingly,and the total energy saving efficiency of the system can reach 11.4%. |
