Study on temperature drift modeling and compensation of FOG based on AFSA optimizing LS-SVM

Fiber optic gyroscope(FOG) is sensitive to the ambient temperature variation which affects seriously the precision and performance of inertial navigation system in deep space. In this paper, the least square-support vector machine (LS-SVM) as a novel learning machine based on statistical learning theory (SLT) is discussed and used to the FOG temperature drift modeling and compensation to reduce the influence of temperature. To validate effectiveness of the proposed method, a set of temperature experiments of FOG bias were done. Besides, wavelet transform (WT) is applied to eliminate any constant and temperature noises as a preprocess procedure of the output of FOG. Then, while the parameters of LS-SVM is tuned by artificial fish swarm algorithm (AFSA) which is an optimization algorithm based on the simulation of fish swarm behavior, and the optimized FOG temperature drift model is established. Moreover, comparison between the traditional back-propagation (BP) neural network approach and the proposed algorithm is given. The modeling and compensation results indicate the AFSA optimized LS-SVM is effective in temperature drift modeling of the FOG.