Multi-information Fusion Fault Diagnosis Based on KNN and Improved Evidence Theory

Purpose This paper proposes a new multi-information fusion fault diagnosis method, which combines the K-Nearest Neighbor and the improved Dempster-Shafer (D-S) evidence theory to consider the uncertainty comprehensively. Method First, KNN is used to perform a local diagnosis of each fault to obtain the prior probability. Then the D-S evidence theory is improved by combining the Jousselme distance, and basic probability assignments of the evidence body are weighed and revised. Finally, the Dempster combination rule is used to fuse the information at the decision-making level. Results Using the test data of the rotor test bench in the laboratory, the effectiveness of the proposed KNN-DS method is proved and compared with the improved evidence theory method proposed by Murphy. Conclusion The results show that the proposed method obtains a better diagnosis result.

Automated summary

This paper introduces a new fault diagnosis method that combines K-Nearest Neighbor and improved Dempster-Shafer evidence theory. Through experimental testing on rotor test bench data, it is proven that the proposed KNN-DS method outperforms the improved evidence theory method proposed by Murphy in terms of diagnosis accuracy.