Abstract: Grey relational analysis (GRA) is usually used to validate the simulation models because of its advantage in the field of time series relational analysis. The ordinary GRA stresses the similarity but ignores the proximity between two test time series while the proximity is part of model error. The grey relational degree given by a set of flight test sequences and a set of simulation test sequences is inconsistent with the practical problems of firing range. The ordinary GRA cannot be used in the overall performance evaluation of simulation model, which leads to a high risk of decision-making. An improved grey relational analysis is proposed to reduce the risk, in which the similarity and the proximity between the simulation test time series and the flight test time series are considered. At the same time it can analyze the time series which has not only one dimension but also more than one dimension by calculating probability sum, and so it fuses random factor of random process and makes full use of simulation test data with small sample flying test. Besides the new grey relational analysis improves the performance of GRA and has higher precision and reliability for the validation of simulation models. The proposed model can be used to solve the relational analysis problem with many characteristics of behavior, and the operational steps are listed in detail. Data test and case study are given to show the reasonability and validity of the improved method.

Key words: ordnance science and technology, validation of simulation model, grey relational analysis, credibility, probability sum