A modelling process of an unknown multi-dimensional system is mostly performed with methods which describe the system by a multi-dimensional surface (e.g. neural networks (NNs)). Some systems, however, does not have a surface nature. On the contrary – their behavior resembles multi-dimensional chains. Obviously, as it was proven in numerous applications, always better results can be obtained when the modelling method corresponds to the system nature. Therefore, when a data distribution of an unknown system has a chain characteristic, the system should be also modelled with a chain, not a surface, method. The aim of this article is to present the alternative approach to the modelling process, in which the multi-dimensional model of an unknown system is built on the basis of a set of two-dimensional NNs instead of one multi-dimensional NN. The proposed approach results in a chain multi-dimensional model of an analyzed system.