神经网络和滞后变量回归的车载终端误差修正

Enhancing Vehicle Terminal Positioning Accuracy Using Neural Network and Lagged Variable Regression

摘要: 针对现有车载定位终端存在定位误差大和更新速度慢的问题，深入分析了车载终端定位误差的影响因素，并提出了基于BP神经网络和滞后变量回归的车载终端定位误差修正方法。对比车载终端三次测量数据修正前后的定位误差，最大定位误差分别减小了88.2%、85.4%和85.8%。通过实测数据对比了车载终端修正前后的定位误差，证明了使用BP神经网络和滞后变量回归建立的车载终端定位误差模型是有效的，定位误差修正效果较好。

Abstract: The existing onboard positioning terminals face challenges of significant positioning errors and slow update speeds. This paper analyzes the factors influencing these errors and proposes a method for correcting vehicle terminal positioning errors using a BP neural network and lagged variable regression. Comparative analysis of three measurement data sets before and after correction shows maximum positioning error reductions of 88.2%, 85.4%, and 85.8%, respectively. Additionally, the comparison of pre- and post-correction positioning errors with measured data validates the effectiveness of the developed model utilizing BP neural networks and lagged variable regression for positioning error correction.