利用人工神经网络仿真GPS误差信号

GPS（Global positioning system）是一种在军事和民用方面广泛应用的定位系统。如何降低成本，提高精度是一个重要的课题。本文应用人工神经网络能够实现高度非线性的特点，在对地理位置已知点进行大量GSP实际测量的基础上，设计出一种BP网络，作为GPS误差信号模拟器。在给出时间和天气情况的条件下，该模型器能够输出GPS的实时误差，为应用系统中对GPS误差进行补偿提供依据。通过与实际测试数据相比较，证明这种方法具有较好的模拟效果。

Using ANN as GPS Measurement Error Simulator

GPS is widely used in civil as well as military. This method for improving its performance and to reduce its cost is an important one. The method to improve its performance commonly used is diffe rential GPS(DGPS),which could cause a cost increase because of the difference station. In the view of the shortcoming of DGPS, a new way to study GPS error is needed. Utilizing the high nonlinear property of the ANN and based on many actual measurements of a point, where precise position is known, this paper designs a BP network as a simulator of GPS error signal. When the time and the weather condition are given, the simulator can output the GPS error used in actual system to compensate the error. The method is proved to be effective in practice.