传感器性能实时评估方法  

Real Time Evaluation Method for Sensor Performance

作　　者：刘志成[1] 陈祥光[1] 李兵[1] 徐远清[1] 

机构地区：[1]北京理工大学化工与环境学院

出　　处：《数据采集与处理》2007年第2期161-165,共5页Journal of Data Acquisition and Processing

摘　　要：为了实时监视传感器的工作状态,基于传感器的性能指标,确定了以传感器输出数据有效度作为评估基准的评估方法。该方法采用小波实时滤波技术对传感器的输出时间序列进行滤波处理,运用人工神经网络预测传感器输出值,并与实测值进行比较求出残差,由此来确定传感器的有效度指数。实验结果表明,本文提出评估传感器性能的方法是实时有效的,并能够在线检测过程参数的突变情况。In order to real time monitor the working status of a sensor, the effective index of the sensor output data is taken as an evaluation benchmark based on the performance indexes of the sensor. The time series from sensor output to be processed by the wavelet transform is proposed, and the real time method of the wavelet filter is discussed. The predicting value for the sensor output using artificial neural network is compared with the measuring value in the real field, and then the residual can be worked out. Therefore, the sensor effective index is figured out. Experimental results show that the evaluation method for sensor performance is real-time and effective, and the abrupt changing of process parameters can be online detected.

关 键 词：传感器 性能评估 小波滤波 神经网络 

分 类 号：TP273.1[自动化与计算机技术—检测技术与自动化装置]