基于神经网络和电子地图的电波传播预测研究

发布时间：2018-06-20 09:10

本文选题：电子地图 + Longley-Rice模型　；参考：《大连海事大学》2017年硕士论文

【摘要】：随着国家"一带一路"战略的提出,未来海上贸易将会得到更快的发展。因此需要进一步改善海上公共安全设施,提高海上无线通信的质量。海上无线电波传输损耗的研究,为更好的规划基站分布提供了良好的解决方案,是改善海上无线通信质量的主要方法之一。单纯对无线通信算法的研究,其实用价值相对较低。从国外引进的电磁仿真软件,虽然功能相对完善,但是可维护性差。因此需要开发适合国内需要的电磁场强预测的仿真软件。本文首先对海上无线通信电波传输损耗的经验模型(主要受载波频率、收/发天线高度、通信链路距离等参数的限制)进行研究,然后采用BP(BackPropagation)网络和GRNN(Generalized RegressionNeuralNetwork,广义回归神经网络)进行海上无线通信电波传输损耗预测的研究。研究结果表明,载波频率为100MHz时,使用GRNN的仿真误差平均值、标准差和均方根误差比BP网络的仿真结果分别低 0.98dB、0.16dB 和 0.86dB;600MHz 时,结果分别低 0.28dB、0.05dB 和 0.27dB;2000MHz时,结果分别低1.05dB、0.13dB和0.90dB。同时将使用BP网络与GRNN所建模型的仿真结果与Longley-Rice模型、Egli模型和Okumura-Hata模型分别进行比较。实验结果表明,该方法不仅可操作性强,而且精度高。其次利用GMAP.NET组件完成电子地图的开发,结合自由空间模型、Longley-Rice模型、Egli模型和Okumura-Hata模型设计并实现了简单的无线通信传输损耗预测和场强覆盖范围预测的仿真软件。利用参考文献中的实测数据,进行了电波传输损耗预测功能的验证。在北纬39°36'15.17"(即北纬39.604214),东经116°16'38.34"(即东经116.277316)处新建天线,进行了场强和电波传输损耗覆盖范围预测,并将预测结果与WRAP软件的结果进行比较,验证了该仿真软件的准确性和可靠性。本文仿真软件的开发方法对国内电磁仿真软件开发具有一定的指导意义和参考价值。
[Abstract]:With the development of Belt and Road strategy, maritime trade will develop more quickly in the future. Therefore, it is necessary to further improve maritime public safety facilities and improve the quality of maritime wireless communications. The study of radio wave transmission loss at sea provides a good solution for better planning of base station distribution and is one of the main methods to improve the quality of offshore wireless communication. The research on wireless communication algorithms is of relatively low practical value. The electromagnetic simulation software imported from abroad, although the function is relatively perfect, but the maintainability is poor. Therefore, it is necessary to develop a simulation software which is suitable for the strong prediction of electromagnetic field in China. In this paper, the empirical model of radio wave transmission loss (mainly limited by carrier frequency, receiver / transmitter antenna height, communication link distance, etc.) is studied in this paper. Then BP BackPropagation (BP) network and GRNN generalized Regeneration Neural Network (generalized regression neural network) are used to predict the radio wave transmission loss in marine wireless communication. The results show that when the carrier frequency is 100 MHz, the average value of simulation error, standard deviation and root mean square error of GRNN are 0.98 dB ~ 0.16 dB and 0.86 dB ~ (600 MHz) lower than those of BP network respectively, and the results are 0.28 dB ~ 0.05 dB and 0.27 dB ~ (2 000 MHz) lower than those of BP network, respectively, and the results are 1.05 dB ~ 0.13 dB and 0.90 dB, respectively. At the same time, the simulation results of BP network and GRNN model are compared with those of Longley-Rice model, Egli model and Okumura-Hata model, respectively. The experimental results show that this method is not only feasible but also accurate. Secondly, GMAP.NET component is used to complete the development of electronic map. Combined with the free space model, Longley-Rice model Egli model and Okumura-Hata model, a simple simulation software for wireless communication transmission loss prediction and field intensity coverage prediction is designed and implemented. The prediction function of radio wave transmission loss is verified by using the measured data in reference. A new antenna is built at latitude 39 掳36C 15.17 "N (39.604214N) and longitude 116 掳1638.34" (116.277316 E). The field strength and radio wave transmission loss coverage range are predicted. The results are compared with those of WRAP software, and the accuracy and reliability of the simulation software are verified. The development method of the simulation software in this paper has certain guiding significance and reference value for the development of electromagnetic simulation software in China.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN011

【参考文献】