Abstract. We propose the fundamentals of the software for the analysis of states of complex systems by probabilistic inference methods based on belief fuzzy network models and their extensions. We introduce the concepts of fuzzy potentials and operations over them, develop mathematical tools of two-phase exact probabilistic inference and the procedure for evaluating and predicting the states of the system under study, and describe the architectural aspects of the computer implementation of the relevant information technology. Figs: 3. Tabl.:1. Refs: 16 titles.

Keywords: fuzzy probability, fuzzy potential, belief network models, fuzzy Bayesian networks, fuzzy influence diagram, network transformation, probabilistic inference, nodal tree, information technologies.