Prediction Of TD-SCDMA Base Station Electromagnetic Field Strength Based On Neural Network

With the development of mobile communication, people contact with each other more and more through the mobile phone now, in order to ensure call quality, all the major operators are large-scale construction of the base station. However, many people began to focus more and more attention to whether these communication base stations would be harm to the surrounding environment and human health while they enjoy their convenient of modern communication equipment. TD-SCDMA is one of the important standard currently3G mobile communication base station, and TD-SCDMA base station has more feedback wires, more power amplify factor and bigger antennas which means it has more serious electromagnetic radiation problem. Environmental protection departments of the state issued a series of measures to evaluate it. One of the key issues is how to measure and predict the electromagnetic signal accurately and effectively.This paper first introduces the propagation characteristics of electromagnetic waves and electromagnetic radiation evaluation criteria, through analyzing the working principle of TD-SCDMA base station, learned that TD-SCDMA is used in smart antenna, because of smart antenna is used in the working principle of beamforming, so the electromagnetic radiation base station electromagnetic radiation is different from the traditional ways; secondly study on the measurement method of the base station electromagnetic radiation, electromagnetic signal intensity on the administrative building of Zhengzhou University TD-SCDMA base station is measured. Finally the application of neural networks to the prediction model of electromagnetic field intensity aimed at the radiation characteristics of TD-SCDMA base station. Because neural network not only has the very good nonlinear approximation but also good fault tolerance.This paper presents models of predict the electromagnetic field strength based on neural network, the model can effectively overcome the traditional models of predic electromagnetic field intensity that cannot adapt to the complex construction environment in different community environment, and effects of other uncertain factors on the electromagnetic field strength. Electromagnetic signal intensity on the administrative building of Zhengzhou University TD-SCDMA base station is predict based on BP neural network and SVM. Compared with the traditional forecasting model show that The prediction error of the forecasting mode base on neural network which established by this paper is smaller, at the same time predictive effect of SVM is better than BP neural network.