Within the framework of a computer statistical experiment, the test problem of identifying the parameters of a field electron emission signal using a regression model based on the Fowler - Nordheim law is considered. Two approaches to determining the parameter estimates are used in the work: ordinary least squares and random search with training. It is shown that the random search error can be neglected if, for the considered ratios of the noise level to the signal power, the number of statistical tests is approximately 10³. The result allows us to expand the class of functionals used to identify the response without changing the method. The article notes the advantages of the random search method for the problem under consideration and the prospects of its applicability to tasks in a more general setting.