Temperature error compensation for open-loop fiber optical gyro using back-propagation neural networks with optimal structure

Open-Loop Fiber Optic Gyroscopes (FOG) is widely used, which is easily affected by the temperature around it. Its temperature model has a very complicated nonlinear characteristic. A BP neural network model with advantage of approximating the nonlinear function was developed to simulate outputs of an open-loop FOG and then compensate the FOG's temperature error in full temperature range (-50°C~ +70 °C). With experimental data, the networks with one-hidden-layer structure adopted the temperature and the temperature change rate as network inputs, and the outputs of FOG as network targets. The results showed that the number of hidden-layer neurons plays an important role in simulation performance, and the network with 11 hidden-layer neurons offered better precision and generalization. Meanwhile, the comparison of 4 different training algorithms demonstrated that the Levenberg-Marquardt algorithm resulted in a better convergence during training processes. With the chosen structure and training algorithm, the BP neural network model was used to compensate the temperature error of the FOG. It was found that the compensated outputs of the FOG became more accurate and more robust. In addition, the neural network model further proved its superiority of precision and robustness by comparison with a multiple linear regression model and a quadratic curve fitting model.