Probability Analysis And Parameters Identification Of Cable Tested Tension By Frequency Method

Frequency method occupies an important place in the measurement of cable tension because of its economical, simple and efficient advantages. In recent decades, scholars at home and abroad fitted a dozen kinds of formulas calculating cable force with the frequency and other structural parameters, which gives calculation options to Frequency method. But these formulas are deterministic analysis not taking into account of the effect of frequency and structural parameters randomness on the cable tension, beside, how to choose the best one in a variety of formulas, we need an evaluation criteria. In the situation of structure parameter unknown or ambiguous, scholars analyzed and identificated on key parameters from the angle of inverse problem who always used ways similar to gradient method in a direction of reverse identification. Which easily makes a solution in extremal range for tension tests of such a nonlinear system. To avoid above shortages, the paper has done work in four aspects:(1) To meet the practical requirements of the premise in the target of high-precision, the paper designed an algorithm called neural networks and monte carlo purpose to calculating cable force To the extent of statistics, calculating failure probability of cable force measurement, evaluating the existing cable force formulas from the angle of accuracy and error propagation, and identifying unknown parameters on the measurement of cable tension from the positive direction. The algorithm is more precise in fitting formula and calculating the standard deviation of the force and failure probability than response surface method, than a direct finite element method to save time.(2) The paper used the algorithm to research the effect of multiple parameters on the result of long cables, middle cables and short cables. If coefficient of variation of bending stiffness, frequency, length of cable, boundary conditions and line quality is all0.1,the coefficient of variation of long cable is much than0.3,the coefficient of variation of short cable is much than0.5.Then the paper evaluated common formulas from the angle of accuracy and error propagation, Ren Weixin’s, Wang Jianfei’s, Hiroshi Zui’s, and Shao Xudong’s formulas are more excellet than others in the clay boundary condition.(3) The paper based on neural network and Monte Carlo method identified unknown parameters on the measurement of cable tension from the positive direction. Using multi-frequencies as target values, globally search for optimal solutions. Recognition results are stable and meet the application requirements.(4) On the background of NanPanjiang hanging arch cable force measurement, the paper applied the algorithm to engineering practice, selected the best formula for calculation of cable forces, determine the failure probability and possible range of cable tension, and to identify the unknown parameters, including identification of cable tension.