桥梁监测数据的动态线性建模与可靠性预测

Dynamic Linear Modeling of Bridge Monitored Data and Reliability Prediction

摘要:

引入贝叶斯动态线性模型(BDLM)，建立桥梁监测应力的状态方程和观测方程，并利用贝叶斯因子监控监测应力.通过监测应力和应力状态的先验信息，对监测应力的状态参数进行贝叶斯后验概率推断，并不断进行“概率预测修正”递推运算，获得最优监测应力的状态概率估计来预测桥梁的应力.基于贝叶斯动态修正的应力概率模型，建立桥梁结构可靠性的预测公式.最后，通过实例验证了本文所建模型的合理性和适用性.

关键词:桥梁监测应力; 贝叶斯方法; 动态线性模型; 折扣因子

Abstract:

A Bayesian dynamic linear model (BDLM) is introduced, which includes state equation and observation equation of bridge monitoring stress, and the stress is monitored with Bayesian factors. Combining parameters prior information with the early stress monitored data containing noise, the monitored stress state parameters are deduced with Bayesian posterior probability. Optimal stress state estimation uses continuous probability forecastfixed recursion operator to predict the bridge stress. The prediction formula of bridge reliability is given based on BDLM of bridge stress. Finally, an actual example is provided to demonstrate the applicability and feasibility of the proposed model.

Key words:bridge monitored stress; Bayesian method; dynamic linear model; discount factor