Identification algorithm with transition in the parameters domain

The identification algorithm of static object with restrictions is considered. The case when the error of output measurement of object $y$ leads to exceeding the bounds of area of admissible estimations of parameters $H$ with some probability $p$ for all $n$-dimensional blocks and the case of greater errors when the probability is strictly equal to zero are offered. The connection of error of measurement and probable distribution of parameters estimation error is analysed with use of Kramer formula.