论文部分内容阅读
为提高航空装备事故预防的针对性、有效性和主动性,预防和减少事故的发生,降低事故造成的损失,提出一种时序的差分自回归滑动平均(ARIMA)模型。其建模过程先在时间序列基础上辨识一个试用模型,然后加以诊断,并作出必要调整,反复进行辨识、估计、诊断,直至获得较为满意的ARIMA预测模型。在实例验证中,所构建的用来预测美国空军飞行事故万时率的ARIMA模型,能够将预测的平均相对误差控制在7%以内,预测结果总体反映航空装备的实际安全状况。
In order to improve the pertinence, effectiveness and initiative of aviation equipment accident prevention, prevent and reduce accidents and reduce the losses caused by accidents, a time differential ARIMA model is proposed. The modeling process first identifies a trial model based on the time series, then diagnoses it and makes the necessary adjustments to repeatedly identify, estimate, and diagnose the ARIMA prediction model until it is more satisfactory. In the case validation, the ARIMA model, which is used to predict the annual rate of flight accidents of the USAF, can control the average relative error of the prediction to within 7%. The prediction results generally reflect the actual safety status of the aviation equipment.