改进支持向量机的光纤陀螺温度漂移补偿方法

温度漂移是影响光纤陀螺精度的主要因素之一，温度漂移建模和补偿是消除和减小温度漂移的有效方法。首先分析了影响光纤陀螺温度漂移的关键因素，同时进行了光纤陀螺温度漂移测试实验。然后采用泛化能力较神经网络更好的支持向量机对光纤陀螺温度漂移进行回归、建模，其中支持向量机的核函数采用了具有更好数据集适应性的径向基核函数。为了提高支持向量机的建模精度，引入人工鱼群算法对支持向量机的核心参数C（惩罚系数）和核函数的参数进行寻优。最后，使用实际的光纤陀螺温度漂移数据对提出的补偿方法进行实验验证，结果表明采用该方法补偿后的剩余光纤陀螺误差较采用线性回归方法减小了四五个数量级。

Compensation method of FOG temperature drift with improved support vector machine

Temperature drift is one of the main factors that affect the accuracy of fiber optic gyroscope (FOG), and its modeling and compensation are effective methods to eliminate and reduce the drift. The key factors that affect the temperature drift of FOG were analyzed. Meanwhile, the test experiment of FOG temperature drift was carried out. Then, the support vector machine which had better generalization ability than the neural network was used to regress and model the temperature drift of FOG, and the radial basis kernel function was adopted as the kernel function of support vector machine which had better data set adaptability. In order to improve the modeling accuracy of support vector machine, the artificial fish swarm algorithm was used to optimize the penalty factor C of support vector machine and the factor of kernel function. Finally, the proposed compensation method was verified by the actual temperature drift data of FOG, which showed that the remaining error of FOG compensated by the proposed method was reduced by 4-5 orders of magnitude than that compensated by the linear regression method.