Font Size: a A A

Research On False-Alarm-Rate Decreasing At Sensor-Level Of Built-in Test Systems In Mechatronic Products

Posted on:2004-08-26Degree:DoctorType:Dissertation
Country:ChinaCandidate:G YangFull Text:PDF
GTID:1118360152457206Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
Reliability, maintainability and testability are required to be considered at the design stage of mechatronic products because complexity of their structure and function has increased recently. Built-in test (BIT) is one of efficient technologies to achieve these requirements. Research and application of BIT are of importance for improving combat readiness and task success, reducing maintenance manpower and support cost of modern weapon systems. But high false alarm rate (FAR) is a dominating problem obstructing the development of BIT. How to reduce FAR of BIT has been an important content in the research of BIT. In this paper, theory for reasons of false alarm (FA) and technologies for reducing FAR at sensor level of BIT systems are studied. The main contents are as follows:1) Analysis of reasons and mechanism of FA is the precondition of research of technologies for decreasing FAR. But it presents unsystematic from current research. Aiming at finding main causes leading to FA, a strategy for FA mechanism analysis is proposed, which is based on information process model of BIT. Factors in the process are summarized and analyzed. Then approaches to decreasing influence of these factors on FAR are suggested. Factors at sensor level are emphasized and analyzed through mechanism analysis and mathematical model analysis. It is proved that bias or abnormal of sensor in BIT systems is an important cause of FA.2) Because bias or abnormal of sensor in BIT systems is an important cause of FA. Analysis of the causes of these abnormalities is the precondition for taking measures. Aiming at research of uncertainty of sensor data leading to FA in BIT systems, influencing factors on sensor data are summarized and uncertainty analysis is studied for evaluating data quality. Rules are given for identification of abnormal sensor data. Then the influencing factors are summarized and abstracted through qualitative analysis based on MSA (Measurement System Analysis). Measures for decreasing FAR are preliminarily proposed. In order to quantitative analysis of sensor data uncertainty, static uncertainty analysis is studied to evaluate influence of factors on sensor data. Then dynamic uncertainty analysis method is proposed to evaluate the evolvement of data uncertainty follow time. Suggestions for optimal design and improvement of sensor system can be induced from the results of uncertainty analysis. These methods are applied to a new type eddy current sensor used for Maglev Vehicle.3) In order to decreasing and avoiding FA, theory analysis for FA in BIT systems is made through mathematical model. A probability model of FAR is given, which is consistent with engineering computation model. Then FAR model of ordinary BIT system is set up. Based on analysis of this model and preliminary measures for decreasing FAR, a new BIT model is proposed, which is named Sensor Level Feedback BIT and its FAR model is set up. The efficiency of improved model is validated through analysis ofdecreasing uncertainty of sensor data and FAR, increasing FDR of BIT systems. Fault probability of sensor (s) and missed rate of detection of abnormal sensor data (s)are abstracted from the improved model and considered as key parameters influencing FAR. Then strategies and technologies are argued which can control s and s . Theresult of theoretical analysis can be instructive to research of technologies for decreasing FAR.4) Aiming at eliminating FA caused by the faults of hardware of sensor systems, optimal design of sensor system is studied. The common process and criterions for optimal sensor selection are given. Reliability of sensor system is analyzed in terms of M-out-of-N principle. Reliability of system will be maximum because of the influence of number of sensors and fault rate of an sensor. Then a mixed integer nonlinear programming (MINLP) model is proposed for optimal placement of sensors, which is based on directed bipartite graph about fault mode of equipment and sensors of BIT system. The objects of optimization are minimization of fault rate and cost of...
Keywords/Search Tags:Built-in Test (BIT), False Alarm (FA), False Alarm Rate (FAR), sensor system, uncertainty analysis, optimal sensor placement, sensor validation
PDF Full Text Request
Related items