Weighed ranking of aprioristic and experimental data in control system functioning efficiency estimation problem with Pascal-distributed number of tests

Abstract

Introduction: In order to steadily estimate the efficiency of control systems for new objects, a great number of prototypes should be tested, which is not always possible in practice. The estimation quality can be improved by joint processing of the a priori information you have before the tests by analyzing certain indicators, and the data obtained from the tests. To estimate the efficiency a posteriori, taking into account both the a priori knowledge and the test results, you have to find their functional dependence on each of them, and specify the parameters of this dependence. Purpose: Integrated processing of the results from both aprioristic and experimental research of a control system, and obtaining posterior estimations of the efficiency indices. Results: A control system efficiency estimation method is proposed, which integrates the aprioristic and experimental estimations of the efficiency indices obtained a priori and during a limited number of tests of system prototypes. It can be used when the results of aprioristic research and the tests are presented by point estimations of the efficiency indices, and the most common methods are difficult to apply. We present analytical expressions for posterior estimation of the probability that the system will perform its task, along with the indicators which are used to study the influence of the aprioristic information on the estimation accuracy and number of tests. The working capacity of the method is illustrated by a real-life example. This approach, unlike others, takes into account how close the aprioristic estimations are to the experimental ones. Practical relevance: The proposed approach is universal enough, as it allows you to integrate the information obtained at various stages of studying the system, and essentially improve the efficiency estimation accuracy, specifying the gain in the number of tests in all the cases when the aprioristic research results are in consonance with the experimental data.