Research On Energy Saving Optimization Of Air Conditioning Chilled Water System In Hospital Based On Load Forecast

Due to many of the factors,such as the building structures,patients and family flows,as well as staff flows,large power consuming instruments,seasonal,weather changes,and holidays,the changes in air-conditioning load in hospital buildings exhibit significant randomness and nonlinear chaotic properties.Normally,the energy consumption of the chilled water system in air-conditioning accounts for over 70%,and reducing the energy consumption of the chilled water system plays a decisive role in the energy-saving optimization of the entire air conditioning system.For this reason,on the basis of accurate prediction of air conditioning load in outpatient and inpatient departments of tertiary hospitals,in this study,the aims is to save energy and optimize the control of chilled water systems,achieving energy-saving in hospital air conditioning systems.The main contents are as follows:Firstly,This paper introduces the basic theory of Long Short Term Memory(LSTM)based on Deep Neural Network(DNN)framework and the air-conditioning load forecasting method of LSTM.Based on the energy consumption characteristics of hospital buildings,the air-conditioning load is divided into functional zones,and the key factors of each functional zone are identified by the method of Grey Relation Analysis(GRA).Subsequently,based on the phase space reconstruction theory of chaos theory,a hybrid forecasting model of chaos and long-term and short-term memory network(CLSTM)for air conditioning load time series forecasting is proposed.In this model,the maximum Lyapunov exponent is used to determine the chaotic characteristics of the system,and the mutual information function method and CAO method are used to determine the delay time and embedding dimension and reconstruct the phase space respectively.In order to further improve the forecasting accuracy,the load forecasting model is optimized by particle swarm optimization and compared with other models.The experimental results show that the average absolute error(MBE)of the particle swarm optimization LSTM prediction method is lower than that of the un-optimized LSTM prediction method for the air conditioning load of the top three medical department buildings with obvious chaotic characteristics.The MBE of the LSTM prediction method based on particle swarm optimization is reduced compared with that of the SVR method based on particle swarm optimization.The results show that PSOC-LSTM model is better than PSO-C-SVR model in predicting the air conditioning load of the top three medical buildings.Finally,according to the energy consumption model of the relevant equipment of the chilled water system,the minimum operation energy consumption of the chilled water system is taken as the objective for optimization,and the energy consumption and operation parameters of the chilled water system are optimized by means of the improved particle swarm optimization algorithm,so that the optimal energy consumption of the chilled water system and the temperature difference value of the supply and return water under the optimal energy consumption are obtained.At the same time,the control transfer function of the chilled water system is constructed based on the temperature difference between supply and return water,considering that it is difficult for the system to quickly and stably enter a new optimal state when the temperature difference setting changes,a model free adaptive control(MFAC)algorithm based on improved particle swarm optimization is proposed.Compared to traditional PID control,the rise time and accommodation time of the control system are greatly reduced,MFAC control effect of particle swarm optimization algorithm is better.