基于协整的多传感器测试仿真模型的研究  

Simulation Model for Multi-sensors Measuring Based on Cointegration

作　　者：黎亮[1] 师伟[2] 秦开宇[1] 

机构地区：[1]电子科技大学,成都610054 [2]中国燃气涡轮研究院,江油621703

出　　处：《系统仿真学报》2014年第7期1559-1565,共7页Journal of System Simulation

基　　金：航空科学基金(20101024006)

摘　　要：在航空发动机的测试中通常使用多传感器组进行测量,刻画它们之间的关系并利用其发现和修正错误是本研究的目的。以协整和向量误差修正理论为基础,提出了一种用于描述同质传感器测试序列相互关系的模型,对该模型用测试数据序列进行了检验,并进一步设计出了传感器之间相互仿真和故障纠正的算法。通过某新型发动机的实际测试数据,表明上述模型能够用多个传感器的数据拟合仿真出某一传感器的输出,再由仿真值和实际值的比较可以判断出此传感器可能的故障。本方法能够以较少的输入数据,拟合出较好的模型参数,并进而达到较好的预测和发现异常的效果。本模型方法可以用于在发动机模拟试验平台中,实时监测并纠正错误。Multi-sensors measurement is often used in the complex environment such as aero engine testing. Our purpose is to find out the relations among these sensors and use them for error correction. Using Cointegration Theory and Vector Error Correction Model, a fitting model for the homogeneous sensors measuring and an algorithm to implement the model were proposed. The model could simulate the output of one sensor from the outputs of the others. When comparing the simulated value with the actual output, the methods could detect the possible faults. By a small amount of data, the algorithm could find the right forecasting parameters for a specified sensor. When using the data from real aero engine to test the model, our method works well. It not only can monitor the abnormalities but also can correct the possible errors.

关 键 词：多传感器 协整 误差修正 实时 数据融合 航空发动机 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]