虽然基于大型渡槽实测资料建立数学监控模型可以有效地分析渡槽工作性态,但大型渡槽的工作状态复杂,而数学监控模型易受经验性和多重共线性的影响,导致钢筋计实测应力的可靠性难以判断。结合某大型渡槽实测资料,对该渡槽1号、2号钢筋计建立实测钢筋应力统计模型和混合模型,然后分别采用逐步回归法和复合形优化算法对上述模型进行参数估计,进而对比分析不同模型分离出的水压分量。分析表明,虽然该渡槽2号钢筋计实测应力具有一定的规律,但不同分析方法分离出的水压分量变化规律均不符合一般力学规律,由此认为该钢筋计测值异常,不具有可靠性,故该钢筋计测值不可用于分析渡槽的实际工作性态。

The working condition of aqueduct can be analyzed effectively by mathematical monitoring model based on measured data. The reliability of measured steel bar stress, however, is difficult to be confirmed as the mathematical model is prone to be affected by the experience and multicolinearity due to the complex working conditions of aqueduct. In this research, the statistical model and mixed model of measured steel bar stress of a large aqueduct were established based on measured data; subsequently, the parameters of the established models were estimated using stepwise regression and compound model method to compare the water pressure components separated by different models. The analysis result showed that although the measured stress of steel bar No. 2 in the aqueduct is somewhat regular, the variation law of the water pressure component separated by different analysis methods does not conform to the general laws of mechanics. Therefore, the measured value of the steel bar is considered abnormal and unreliable, hence cannot be employed to analyze the actual status of aqueduct.