D-S理论和Markov链组合的桥梁性能退化预测研究OA北大核心CSTPCD

To accurately predict bridge performance degradation,the inherent data randomness and the subtle perturbations leading to state transitions were considered.A combined prediction method for the bridge per-formance degradation based on the D-S theory and the Markov chain,and the performance degradation rate concept,were proposed.In this model,the exponential smoothing(ES)methodology was employed as the ba-sis for generating new sequences of predictive data.It was continuously optimized through the utilization of the Markov chains and the Dempster-Shafer(D-S)evidence theory.The combined prediction of bridge performance degradation was achieved.The application results from practical engineering show that,the performance degra-dation rate serves as an intuitive indicator of the speed at which the bridge performance degrades.Subsequent-ly,the combined model demonstrates an average relative error of 1.54%,improves by 1.11%,0.88%,and 2.8%in accuracy,respectively in comparison with other models of the regression,the grey system,and the fuzzy weighted Markov chain.Additionally,the calculated posterior difference ratio is 0.242,well below the established threshold of 0.35.In terms of stability,the standard deviation of the model is 9.021,reduces by 3.978,3.405 and 7.500,respectively compared with those of the other 3 models.The coefficient of variation is 0.109,indicating a significant reduction in comparison to those of the other models.The combined prediction model,with verified accuracy and stability,establishes a theoretical foundation for prediction and maintenance of in-service bridg-es'structural performance degradation.