A.N. Vasilyev - Dr.Sc. (Eng.), Professor, St.-Petersburg State Polytechnical University. E-mail: a.n.vasilyev@gmail.com
V.N. Denisov - Dr.Sc. (Eng.), Professor, St.-Petersburg Mining University. E-mail: 3565451@mail.ru
D.A. Thakhov - Dr.Sc. (Eng.), Professor, St.-Petersburg State Polytechnical University. E-mail: dtarkhov@gmail.com
V.N. Fedotov - Associate Professor, St.-Petersburg Mining University. E-mail: nik2k@mail.ru

Ecology worsening of megalopolises leads to the need for the solution to the problem of constructing the hierarchical intellectual system of monitoring and predicting the influence of motor transport and objects of transport infrastructure on the environment, and progress in the field of information-computing technologies makes the solution of this problem possible. In the publication scientific bases  methods and algorithms  are proposed for the construction of such hierarchical intellectual systems on the basis of the neural networks models. These models are constantly adjusted to updatable data and sent from one level of hierarchy to another one. Isolation of the control parameters in the three-level hierarchical information system and their optimum selection will make it possible to reduce ecological backlash from the transport infrastructure on the environment. It is important to create a set of the neural network models of the standard urban elements (templates «Main street», «Canyon», «Cross-road» and other), which allow refinement in the process of construction and functioning on the basis of real measurements. The basis of project is created by the authors universal approach to the construction of the hierarchy of neural networks models for each of the cases: «equations» or «data sets»,  and in the mixed heterogeneous situation: «equations + data». Using this approach it is possible to build resistant to the errors and capable of learning new pieces of information neural network models.