基于ARMA模型判别气压传感器故障  

A Fault Detection Method of ARMA-Based Pressure Sensor

作　　者：康立宁 卢会国[2] 裴翀[3] 胡学英[3] 刘银锋[3] 

机构地区：[1]河北省灵寿县气象局,灵寿050500 [2]成都信息工程学院,成都610225 [3]中国气象局气象探测中心,北京100081

出　　处：《气象科技》2014年第6期986-991,共6页Meteorological Science and Technology

基　　金：公益性行业(气象)科研专项"地面观测网自动化运行监控技术研究"资助

摘　　要：我国现有的自动气象站设备缺乏能够反映其自身运行状态的参数信息,导致对设备运行监控能力不足,无法快速定位设备故障。为了判别气压传感器的工作状态,从观测数据出发,提出了基于ARMA模型的气压预测方法,利用已测气压值,对当前时刻气压值进行合理的预测,并给出该预测值的置信区间。利用气压传感器的实测值与此预测值进行比较,如果实测值偏离了该预测值的置信区间,可以认为气压传感器或相关数据采集单元出现故障。通过实例计算和比较,模型ARMA(2,1)对于自动气象站气压值的一步预测误差为0,具有较好的预测效果,可以有效判别气压传感器的工作状态。The AWS equipment in China is running daily, lacking of information reflecting its running status. In order to test the running status of the pressure sensor, a pressure forecasting model based on the time series analysis is proposed. Through this model, the next value can be estimated reasonably by using the latest values together with a confidence interval, and then compared with the actual data. If the actual data deviate from the confidence interval, then the pressure sensor is considered failed. The validity of the model ARMA is verified by the results of one forecasting example, and the best model is ARMA （2, 1） with the forecast error being 0.

关 键 词：气压传感器 时间序列 ARMA 气压预测 

分 类 号：P415.12[天文地球—大气科学及气象学]