INTERVAL ESTIMATION OF RELIABILITY OF ONE-OFF SPACECRAFT

Abstract

Aim. Many space technology products fall into the category of one-off (unique) or are manufactured in small batches of 3 to 5. In accordance with the regulatory documentation, the design and development activities in the space industry must involve quality assurance of products with interval estimation of dependability indicators. However, for one-off unique spacecraft that account for a fair share of the overall space industry product output, acquiring such estimates is associated with the problem of availability of original statistical data. That is due to the high cost of both the spacecraft itself and its testing, which does not allow testing large numbers of samples in the process of spacecraft development. In the context of restricted funding of the space industry, a practice has arisen that involves conduction each planned type of test on a single sample. The test samples have different configurations and versions of components (dimension and mass models, thermal analogues, etc.). In this case, it is impossible to acquire homogeneous statistical data in order to substantiate the compliance with the dependability requirements.Results. The article proposes a method of interval estimation of the probability of no-failure of a one-off spacecraft based on the results of flight tests using a priori information acquired at the stage of pre-delivery and acceptance testing. The authors compare the feasibility of using computational, experimental or computational and experimental methods of spacecraft dependability indicators evaluation. As initial data, electric and radio engineering and thermal vacuum tests results of spacecraft flight model are used. The fact that only the electric and radio engineering tests results are taken into consideration is due to the dependability of spacecraft being primarily defined by the dependability of the electronic equipment. The scope of tests (normally, about 50 for each spacecraft) allows obtaining highly reliable and informative estimates. This method can also be used at the stage of operation in order to evaluate and supervise dependability, e.g. after a year of operation. The correctness of aggregation of the a priori information and the information obtained at the said stage is verified with Fisher’s Z-value.Conclusions. The proposed method allows estimating pointwise values of probability of no-failure of one-off spacecraft, lower confidence bounds and mean-square deviation of the probability of no-failure at the stages of pre-delivery and acceptance testing, flight testing and operation using a priori information. An example is given of interval estimation of probability of no-failure of one-off spacecraft based on the results of flight operations using a priori information obtained at the stages of pre-delivery and acceptance testing.