基于WLS-SVM的飞机状态监控与预测方法

介绍了加权最小二乘支持向量机(WLS-SVM)在时间序列预测中应用的基本方法,给出了一维时间序列建模预测的一般框架。提出采用BIC准则选取嵌入维数,并给出了基于统计量的模型性能评价方法。针对飞机发动机的典型状态参数,分别进行基于加权最小二乘支持向量机和AR模型的建模与预测,给出了详细的比较结果。试验表明,由于加权最小二乘支持向量机采用了新型的结构风险最小化准则,因而表现出优秀的推广能力,可预测区间较长且具有较高的准确度。

A Monitoring and Forecasting Method of Airplane Status Based on WLS - SVM

The essential method of application of weighted least squares support vector machines(WLS-SVM) to time series forecasting is introduced in detail in this paper,and the general framework for one dimensional time series modeling forecasting is proposed.BIC rule is adopted to select the embedded dimension,and a model performance evaluation method based on statistic is presented.The WLS-SVM model and AR model are set up and used to forecast the status of airplane based on the representative parameters,also the comparison result between the two models is given.The result shows that the WLS-SVM has excellent extended capability because the new type of structural risk minimization principle is adopted,and simultaneously it is of high accuracy and has long forecasting intervals.