Предложен подход к оценке качества моделей в полимодельных комплексах на основе топологической типи-зации, формирующий векторное пространство статистического образа мультимоделирования и числовые меры верификации с точки зрения устойчивости статистик. Приведена схема каркаса системы поддержки принятия решений по динамической типизации топологии полимодельных комплексов, позволяющая повы-сить эффективность принимаемых решений по оценке качества моделирования сложных технических систем. The paper considers aspects of the functioning of complex technical systems (СTS) of critical application, for which the most important task is to ensure the quality of their models, which are polymodel complexes (PMCs) with a complex, in most cases, variable topology. The operation of such СTS is implemented in conditions of uncertainty of states, lack of possibility of field tests, high cost of obtaining data, complexity of monitoring states due to special operating conditions, high cost of resources. In existing studies of estimates of the results of simulation modeling, the criteria of Student, Fisher, Mann-Whitney and Wilcoxon, and others, are used. However, the issues of model stability metrics and their complexes remain insufficiently developed. Stability is a fundamental property of dynamical systems, which is usually studied in two planes: both the reaction of the system to external perturbations of the dynamic character, and the change in parameters in response to these perturbations. The proposed prospective studies related to the formation of estimates and metrics of the statistical stability of models, statistical volatility of models, and the statistical image of the PMС allow us to assess the degree of instability of the statistics depending on the volume of repeated runs of the model and are a logical continuation of the authors' research in this area. The study also considers the statistical stability of the parameter studied by the model. In contrast to the statistical stability of the model itself, the statistical stability of the model parameter allows you to fine-tune the model in the PMС. For models of large dimensions with a variable number of runs, the authors propose to type the topology of the PMС, which will allow us to evaluate the models most accurately. If you have defined a vector that reflects the characteristics of x0 models Mx, which belongs to the vector space Vn: x0  Vn, then the set of such vectors reflecting the set of properties Mx, x  Vn, satisfying the inequality maxρ(x, x0, t) < h(t), it is called a ball of PMC properties with a radius of h(t) is at a point in time t, '0ttt (the power of this property for the model Mx) centered at a point x0. A set of vectors that reflect the properties of a set of models Mx in PMC: x  Rn, satisfying the inequality ρ(x, x0, t) ≤ h(t), ' 0t tt  form the polar coordinate system of the model properties in the PMC relative to the property x0. The polar representation of the properties of the PMC is its statistical image. The article proposes the principles of dynamic typing of PMC topologies, on the basis of which the framework of the system for supporting decision-making on the quality assessment of PMC is presented. In further research, the authors propose to expand the set of statistical metrics of PMC, as well as to clarify the structure of the decision support system for problems of qualimetric analysis of applied CTS.