Elicitation Of Expert Information In Reliability Evaluation

The high-tech weaponry have some characteristics such as complicated, costly for test. So Bayesian small sample reliability test and evaluation method received more and more attentions. Expert information is an important source of prior information in Bayesian method. The application of expert information in reliability test and evaluation is studied in this thesis, and the method how to get the prior distribution from the individual expert is given .The contributions of this thesis are as follows.The process for eliciting expert information is divided into four steps, i.e. description, collection, transformation and fusion. The detailed process of these steps are given, and the application of expert information in reality is offered.Through enumerating the current description forms of expert information, the description forms are classified as 3 classes (statistical, fuzzy and other type) by different description methods, and every description form is divided into more detailed classifications. This is the foundation of the follow-up application of the expert's information.The consistency of expert's information is the guarantee of expert's information validity. The consistency definition and check method is given under two different conditions: first, one description form (the quantiles as example) of individual expert information; second, several kinds of description forms (the quantiles and the moments as example) of individual expert's information.The existing methods which change the expert information into prior distribution are synthesized and classified. The connection between them is also given. The method for changing every description form into prior distribution is offered.Associated with the reliability test of a equipment, the process for eliciting expert information in this thesis is applied, and the example of the application of expert information in reliability test and evaluation is given.