Contents I. Introduction
One of the most difficult problems of constructing the mathematical models of dynamics according to the observations results is the problem of structural identification [1]. In the process of this problem solving, you need to determinate a general view of the model [2]. Preferably, such models in discrete form are represented by difference equations. The Ukrainian researchers, under the guidance of academician O.G. Ivakhnenko, were among the first, who tried to solve this problem [3, 4]. For this purpose, the group method of data handling (GMDH) has been developed [3–5]. At the same time, during the structural identification of the discrete model, the observations results were considered to be “accurate”, which is not always true. In a number of papers, the cases when the observations results are inaccurate are considered, for example they are represented as numerical intervals, what is caused by measurement errors with known boundary values [6]. In this case, the methods of analysis of interval data [7, 8] are used. However, such a representation of the experimental datasets makes it impossible to use GMDH because of it’s high computational complexity. In this case, it is most advisable to use evolutionary methods and algorithms. To solve the optimization problems, for example – the problem of structural identification of interval discrete dynamic models (IDDM), a class of bee colony algorithms is often-used [9, 10]. One of such methods is described in the papers [1, 11]. However, the methods described in these works have a number of disadvantages, the main of them is the poor validity of formulas for the some operators. In particular, it is the operator to identify the profitable food sources. In the paper, two varieties of this operator realization are proposed. The essence of proposed operator realizations derives from the foraging behavior of the honey bee colony, that is, has a clearer background. The comparison of the efficiency of application of these operators was carried out on the example of construction of discrete models of dynamics of nitrogen dioxide.
