基于FIG-SVM的MEMS陀螺随机漂移预测

Forecasting of MEMS Gyroscope's Random Drift Based on FIG-SVM

针对传统预测方法不能对MEMS陀螺仪随机漂移进行精确预测的缺点,提出了一种基于模糊信息粒化的支持向量机模型的区间预测方法。该模型首先利用模糊信息粒化算法对原始数据进行预处理,将样本空间划分为多个粒(子空间),降低样本规模,减小时间复杂度;然后将模糊粒化后的数据进行相空间重构和归一化,利用SVM进行回归分析,同时利用交叉验证选出最优的调节参数,避免出现过学习和欠学习;最后利用训练得到的模型进行随机漂移预测。实验结果表明,该方法能够有效预测随机漂移变化趋势和变化区间,具有良好的工程应用前景。

Considering that traditional methods can＇t make accurate predictions to MEMS gyroscope＇s random drift, we put forward an interval prediction method by using the Support Vector Machine （SVM） model based on fuzzy information granulation. First, the original data is preprocessed with fuzzy information granulation algorithm to divide the sample space into multiple snbspaces for reducing the sample size and decreasing the time complexity. Then, the scalar gyroscope random drift time series is embedded to an assistant phase space by the technology of phase construction and the data is normalized. SVM is used to conduct regression analysis, and the optimal regulation parameters of the model are obtained by using cross validation algorithm, thus to avoid over fitting and under fitting phenomenon. At last, we predict the random drifts with the trained model. The results show that the model can effectively predict variation trend and interval. Therefore, the model has a good prospect in engineering.