Research on Test Data Expansion Method Based on UAV Typical Index Evaluation

In order to scientifically and effectively evaluate the performance indexes of aircraft, the common Bayes multi-source data fusion method can comprehensively utilize a variety of information types under the condition that the decision risk is as small as possible, but there are risks such as insufficient data utilization and small sample data being “submerged” by large sample data in the fusion. This paper proposes a test data expansion method based on the combination of probability features and BP neural network model, and is verified by typical application to ensure that the sample size of multi-source data is comparable. Supplementary data not only retains the probability features of the original data, but also the sample generation has physical connotation.