Correction Method for Photoelectric Theodolite Measure Error Based on BP Neural Network

Abstract

The least square method is used widely to correct the system error in photoelectric theodolite, but it depends on the model tightly and neglects the residuals during modeling. A system error correction method based on BP neural network is put forward firstly, and then a class of stylebooks are chosen and fully enlarged. Finally, some stylebooks are chosen to train the BP neural networks and the remainders are used to validate the approach ability of BP neural network. The simulation results prove BP neural networks can attain ideal effect for error correction. © Springer-Verlag Berlin Heidelberg 2011.