基于方差补偿自适应Kalman滤波的ARMA与PSO-SVM模型变形预测

The Deformation Prediction of ARMA and PSO-SVM Model Based on Variance Compensation Adaptive Kalman Filter

根据变形监测数据非线性、波动性特征及实时动态数据处理的要求,在选取方差补偿自适应Kalman滤波进行随机扰动剔除及模型误差削弱分析的基础上,采取自回归移动平均模型(ARMA)构建趋势预测值,最后通过粒子群(PSO)优化参数的支持向量机(SVM)获得误差补偿修正的ARMA模型。应用该方法对变形监测工程实例进行沉降预测,预测结果验证了该方法能较好地描述复杂环境因素下的工程实际变形量,对工程预警预测有一定的参考价值。

According to the non-linearity,volatility characteristics and real-time dynamic data processing of deformation monitoring data,the auto-regressive and moving average model（ARMA）is used to construct the trend,based on the selection of variance compensation adaptive Kalman filter for stochastic disturbance rejection and model error weakening analysis.The error compensation and correction ARMA model is obtained by using particle swarm optimization（PSO）parameter optimization support vector machine（SVM）.The method is used to predict the deformation monitoring engineering.The prediction results show that the method can describe the actual deformation of engineering under complex environmental factors and play a certain reference value in forecasting the project.