In the Dempster-Shafer evidence theory,the traditional D-S evidence combination rule fails to identify the actual conditions with highly conflicting evidence,which may result in contradicting conclusions.In order to solve this problem,this paper proposes an improved algorithm for the fusion of conflicting evidence,which detects and eliminates the conflicting evidence by calculating the deviation from all evidence to the average support level of the proposition,and amends the results to combine the evidence from different recognition frameworks.Numerical results show that the proposed method can effectively deal with the conflicts and achieve accurate identification,and that it is prior to other typical algorithms in terms of convergence and reliability.