BP神经网络在城市有轨电车GPS/RFID组合定位中的应用研究

在城市有轨电车定位系统中,单一的GPS定位方式已很难满足电车连续精确定位的要求。采用GPS和RFID组合定位的方法,可实现在弱信号环境下的连续精确定位。针对GPS/RFID组合定位时,因加入RFID观测值带来的较高计算复杂度而引起定位时间延长,以及对系统定位误差影响不确定性等问题,建立基于BP神经网络的城市有轨电车GPS/RFID组合定位模型。仿真结果表明,采用BP神经网络进行分析时,将GPS和RFID观测值归一化后输入到训练好的网络中,可以在较短的时间内得到可靠的网络输出。经训练后的网络输出较未经训练的输出更接近于期望值,且更为稳定,证明在GPS信号受遮挡条件下城市有轨电车定位系统的定位精度和定位时长得到了有效改善。

Application of BP Neural Network to the Analysis of Positioning Deviation on City Trams

It is difficult to realize the continuous and precise positioning in the positioning system of city trams only by GPS,while it can be performed with the integration of GPS and RFID in the environments with weak signals. A model of GPS / RFID integrated positioning of city trams with the application of BP neural network is established to solve the problems of prolonged positioning caused by high computation complexity and the uncertainties of the impact on the system positioning errors with the introduction of RFID observations in GPS / RFID integrated positioning. The analysis indicates that the reliable network output values are to be obtained in a short period of time after the input of the normalized GPS and RFID observations into the trained network in positioning analysis with the application of BP neural network.The output values of the trained network,which are more stable and closer to the expectations than the ones of the untrained network,demonstrate the improvement of positioning accuracy and the shortening of positioning time in the positioning system of city trams under the condition of blocked GPS signals.