Research On Energy Saving Optimization Algorithm Of Air Conditioning In Communication Base Station Based On Streaming Data

With the arrival of 5G era,the high energy consumption of communication base station has caused widespread concern in society.In order to ensure the stable operation of communication equipment in the base station,the air conditioning in base station usually adopts the setting scheme of constant temperature for 24 hours without interruption,which makes the air conditioning energy consumption in the whole communication base station energy consumption is relatively large.Although energy-saving technology for base station air conditioning itself has been extensively studied,and good results have been achieved.However,due to the insufficient management,most rooms still use the constant temperature setting of air conditioning,which makes the energy saving of air conditioning not play the best role.On the other hand,all kinds of status data in the base station environment have been monitored and collected all the time.The accumulation of a large amount of data provides a new way to improve energy-saving optimization of air conditioning.Under this background,this paper uses related technologies such as data mining and deep learning to analyze the streaming data collected by the base station,constructs the segmented control model of base station air conditioning and the classified prediction model of base station energy-saving control mode,realizes the real-time control and adjustment of the set temperature of the base station air conditioning,and greatly improves the energy-saving rate of base station air conditioning.The main content and innovative achievements of this research are divided into the following parts:(1)The energy consumption of air conditioning in base station is modeled,and the optimal temperature setting method under single temperature control of air conditioning is studied.In view of the unreasonable energy consumption of air conditioning caused by frequent start-up and shut-down of air conditioning under single temperature setting,a dual temperature control model of air conditioning is proposed,and the optimal control strategy of air conditioning under dual temperature control is given.On this basis,in order to deal with the influence of uncertain factors such as outdoor temperature and energy consumption of communication equipment,an air conditioning temperature control algorithm based on X-means clustering is proposed.By analyzing the clustering results,the optimal control strategies of air conditioning dual temperature are given for different power modes.The experimental results show that the air conditioning energy saving rate of the tested base station is more than 20%,and the energy saving effect is obvious.(2)Aiming at the problem that the accuracy of clustering model is not high in dividing samples by distance,a classification model of LSTM base station electricity pattern with attention mechanism is proposed.At the same time,considering the concept drift of base station flow data,the single classification model can not adapt to all control requirements in time,so a concept drift detection algorithm based on dynamic sliding window is added to the model.Once the concept drift is detected,the learning adjustment of the model is started,which ensures the validity of the model.The experimental results show that this model can accurately detect the concept drift of stream data,and the accuracy is 11.1%,5.1% and 3.4% higher than BP neural network,RNN and LSTM network classification models,respectively.(3)Finally,aiming at the problem of untimely air conditioning control caused by sliding window in concept drift detection algorithm,a prediction model of energy saving control mode based on multi-channel time series fusion was proposed.The model uses the historical state information to predict the power consumption pattern of base station in a certain period of time in the future,and adjusts the temperature control strategy of air conditioner in advance,so as to achieve the purpose of real-time and accurate temperature control.At the same time,the time series is divided into adjacent time series and periodic time series,which are taken as the input of the prediction model,thus effectively improving the prediction accuracy of the model.Through experimental comparison and verification,the prediction accuracy of energy-saving control mode based on multi-channel time series fusion proposed in this paper reaches 95.1%,which is 8.6%and 3.5% higher than the original RNN and ATTENtion-LSTM models based on adjacent time input sequence,and further improves the real-time performance and accuracy of base station air conditioning temperature control.