最小二乘AR模型的惯性器件故障预测

Fault forecasting of inertial component based on the least square AR model

以随机信号时间序列AR模型为基础,将激励噪声信号构造为准则函数,采用最小二乘法对AR模型参数进行辨识,得到惯性器件漂移误差系数最小二乘AR预测模型,泛化用于故障预测。仿真结果表明:与功率谱估计的参数化AR、ARMA模型相比,最小二乘AR预测模型在高、低阶次时辨识精度较高、泛化能力较强,这一特点在低阶时尤为明显,其预测结果可为发现故障隐患、辅助决策提供依据。

The forecasting model of drift error parameter of inertial component is introduced s the forecasting model is the least square AR model. To build the model, the criterion function is constructed with driving noise in the AR model, the least square method is used to distinguish parameters of the AR model. The achieved least square AR model is applied to fault forecasting, the result of simulation shows: (1)It has better forecasting accuracy and generalizing capacity than AR and ARMA model whose parameters are estimated by power spectrum; (2) It is superior to others when step of the model is lower; (3) It provide basis for prediction of latent fault and aiding decision.