Nowadays, the priority role of innovations in modern industrial systems is emphasizing. At the same time such innovation processes can't be realized alone. For remote connection of several industrial and science organizations in clusters for innovative processes implementation, common software and analytical solution is implementation of VPN protocol in network LAN segment. The installation of this solution often allows solving the generalized problem of remote access to industrial enterprises and scientific organizations network. However, the lack of means to distinguish segmentation of VPN clients does not allow using this solution for effective and secure cooperation work. This paper was aimed to develop software for solve this problem. The proposed solution is based on an intelligent system for detecting topological cluster divisions for subsequent segmentation of VPN tunnel based on deep learning neural networks. Original features of authors' software are: ability to dynamically replace customer fingerprints and secure information interaction system for innovative process implementation in scientific and industrial cluster. Among other things, the use of the implemented primary encryption approach based on the Elliptic Curve signature and symmetric HMAC and FERNET encryption methods made it possible to provide dynamic generation of signatures to ensure the required level of security.