A System with Minimal Redundancy for Intelligent Prediction Management

Abstract

This paper proposes a predictive redundancy method for management, which is used for intelligent prediction management, which can effectively improve the real-time online reliability of agents. This method changes the multi-redundant post-processing scheme of traditional system and adopts single-redundant real-time pre-switching. Through a large number of experiments, the feasibility of the system is verified, and the system can be built as a redundant fault-tolerant system. The redundant fault-tolerant system has good generalization and can be implemented in the scene. A redundant module that is about to fail is used, including: the proposed redundant module that continues to decline in health, a redundant module that rapidly decreases in health at a certain time, and a redundant module that is always in a sub-healthy state. After the health assessment, you can enter the early warning model and observe whether the early warning model can issue a health warning, which can verify the effectiveness of the early warning model.