中文摘要：

针对传统实时子结构试验方法存在数值积分稳定性和伺服加载时滞等问题，提出了全局迭代的实时子结构试验方法。该方法的数值计算和试验加载分别在动力荷载地震作用时间段内独立进行，它们之间传递整个时间段的位移计算结果或试验恢复力，迭代进行；中间子时间步不存在数据交互，这就避免了对数值积分方法显式格式和稳定性及精度的苛刻要求，伺服加载的时滞效应也不存在。以高速铁路32 m常用跨径简支梁桥的简化模型为例，对所提试验方法进行了验证。结果表明随着迭代的进行，试验趋于收敛。经过5次迭代之后，不同工况的收敛指标值均小于2％，满足工程精度要求。进一步分析了数值模型初始误差对迭代次数的影响，结果表明，初始误差越小，迭代收敛的速度越快，达到误差限值所需的迭代次数越少。最后指出了本文方法的适用范围。本文所提全局迭代的实时子结构试验方法是可行和有效的，将为工程结构的试验提供新的有效途径。

英文摘要：

Problems such as numerical integration stability and servo-loading time delay exist in the traditional real-time substructure test method . In this paper, we propose a new real-time substructure test method to avoid these problems. The numerical calculation and load testing are performed independently in the whole time of dynamic load, with historic process of displacement in numerical calculation and resistance in loading test transferred, and with a progress of iteration existed. Without data exchange between the intermediate time steps, the requirements of explicit scheme, stability, and accuracy of numerical integration method can be avoided. Moreover, the time-delay of the actuator will also disappear. The proposed method is testing using a simplified model of a 32 m-span, simple-supported, high-speed railway bridge. The results show that the reaction tended to converge following the iteration implementation. The convergence index values of all cases are less than 2% after five iterations, meeting the engineering accuracy requirements. A further analysis is made to compare the effect of the initial estimation error of the numerical model parameters on the number of iterations. The smaller the initial parameter estimation error, the faster the convergence speed and the lesser the number of iterations needed to achieve the error limit. The proposed method is feasible and effective, and the iteration number is acceptable. Moreover, the method can provide an effective path for testing engineering structure.