Analysis of experimental statistics of intervals. Verification of interval distribution exponentially

Known methods of looking for indications of self-sustained chain reaction in a nuclear fuel are based on the assumption, that appearance of neutrons from fission chains distorts a distribution function of the base neutron flux. This assertion for the case of great subcriticality has been checked for the three sources: α-current of a radiator (²³⁴U + ²³⁵U) of fission chamber KHT-31-1; laboratory PuBe neutron source; spent fuel assemblies (SFA) in the spent fuel storage of Chornobyl NPP. The first two sources with guarantee do not have the distorting pulses from fission chains neutrons, whereas in SFA such effect is possible. In the experiment the time moments of registration of every pulse by the detecting systems were fixed. The difference between such moments gives intervals between acts of registration, and at measurements during enough long time the long continuous set of intervals is received. The statistics of the intervals distributions between acts of neutrons registrations, including exponentially of the distribution, has been investigated. Checking of exponentially was fulfilled as checking of linearity of probability density function at drawing in a logarithmic scale. The ratio of variance to mean for analyzing set of intervals, and also scatter of experimental data relatively of approximated line in a logarithmic scale were checked. It was shown the dependence of final result of experimental data processing from time span width, used for the histogram of probability density function (the histogram is necessary to check exponentially). It was shown, that there is an optimal value for this time intervals, which can be expressed in parts (0.2?0.5) of a mean value of the sample of intervals. Use of such ratio parameter under histogram building enables comparing of statistic properties of diverse origin intervals, that is comparison of histograms from different sources of ionizing radiation. The distinctions between investigated statistical parameters were not found. This mean, that there are no signs of probability function distortion at the level of subcriticality (k_ef approx 03 -:0.5) of a separated SFA.