The Reserch On The Control Of IDC Room Temperature

Due to the wide coverage of the rapid development of information technology and the globalization of business, the Internet industry is thriving,causing major areas of demand for IDC room continues to grow. But over time, the use of IDC room are also emerging in the process of urgent problems, the most obvious is the enormous power the engine room and the room temperature seriously overweight problems. Worse still, in some places appeared IDC room cooling capacity due to lack of air conditioning, the room temperature is too high threat to the safe operation of the equipment room and other issues, besides IDC business also has high energy consumption characteristics. So if you want to build a successful IDC room you need to solve a reasonable run CRAC systems, this process has become now IDC room building and the use of part of the inevitable process.Based on the analysis of current computer room air conditioning system energy use, based on the trend of future changes in working conditions by predicting server load, combined with real-time acquisition of data center temperature data, the establishment of BP neural networks with multiple data sources. Get the data center to the next time period predicted temperature by the neural network model, and then complete the establishment of IDC room neural network control system, to achieve energy saving control, reduce energy consumption costs. Principle of the method is its real-time collection room equipment(such as servers) CPU utilization, CPU utilization to establish the use of the original time series, time series analysis method using the original time series seasonal decomposition, to give time series data characteristics and data fluctuation law, and the use of Holt. Winters’ exponential smoothing method for predicting the sequence distribution of the next time period. Will get rid of random fluctuations CPU utilization data and subsequent prediction condition data interval division, established discrete devices table load conditions, namely through the discrete data processing, reduce complexity and volatility of the follow-up control. Layout analysis engine room equipment, the interior space zoning. Combined room equipment load condition changes, the use of BP neural network forecasting model IDC room temperature hot zone. For within the CRAC temperature control system of the work environment, the use of neural network applications in the control system, the establishment of IDC room neural network control system, improve the temperature-controlled room air conditioning system adaptability and reliability. Meanwhile, this paper demonstrates the feasibility study simulation test methods.