A Method for Analyzing Data in Measuring Devices Using a Specialized Computing System for Synthesizing and Evaluating Statistical Hypotheses

The article examines the application of a method for processing experimental data in measurement systems by generating hypotheses about the presence of a given set of parameters and estimating their probability. The difference between the method is the use of the error probability distribution function with an independent parameter "distribution scale" and the search for a combination of process model parameters at which the posterior probability function takes the maximum value. The method is characterized by high resistance to impulse noise, as a result of which it can be used in measuring systems. A special feature of the method is the large number of calculations required to generate hypotheses and test their probability. To solve this problem, a parallel computing system architecture has been proposed that generates hypotheses and evaluates them using independent processing elements.