Intelligent industrial protocol conversion system and method based on flow prediction

Abstract

The invention relates to an intelligent industrial protocol conversion system and method based on flow prediction, and belongs to the technical field of industrial networks. The system comprises an industrial heterogeneous network protocol converter, an SDN controller and an industrial network node. The SDN controller issues a traffic query frame of the industrial network node to the protocol converter, and obtains a returned industrial network protocol current traffic query result; through a prediction method based on deep reinforcement learning, predicting the network traffic of the next time period according to the queried current traffic and the recorded historical traffic characteristics, and issuing a prediction result to a protocol converter; the industrial heterogeneous network protocol converter reserves and allocates network resources required by protocol conversion such as storage and calculation according to the flow prediction result of the industrial protocol, and the conversion speed of the industrial protocol is improved.

Description

technical field [0001] The invention belongs to the technical field of industrial networks, and relates to an intelligent industrial protocol conversion system and method based on flow prediction. Background technique [0002] The factory site network is usually composed of many different types of networks. Due to the differences in the transmission media and protocols of different types of networks, it is difficult to communicate directly between heterogeneous industrial networks. Therefore, protocol conversion is the key to realize the interconnection of heterogeneous industrial networks. technology. During the operation of industrial networks, the conversion efficiency of different heterogeneous protocols is closely related to the dynamically changing network traffic. How to accurately predict the traffic volume of various protocols and timely adapt network resources such as storage and computing in advance? Configuration reservation is an important way to realize effici...

AI Technical Summary

Problems solved by technology

However, in industrial networks, the combination of software-defined network (Software Defined Network, SDN) technology and deep reinforcement learning has not been fully used to effectively predict industrial network traffic.

In addition, the existing flow prediction methods usually need to collect a large amount of flow data for offline training, and collecting and counting the flow data of all network nodes in an industrial network is time-consuming and complicated. The network traffic prediction method supports the reservation and allocation of network resources, and realizes the intelligent and rapid conversion of heterogeneous protocols

Images