Research On Dynamic Performance Of Full Prestressed Concrete Beam And Identification Of Virtual Prestressing Force

Prestressed concrete structure is an internal stress concrete structure that relys on stretching tendons. The effectivity of prestressing tendon is directly related to the reliability, applicability and viability. Engineering practice has proved that, the concrete structure damage brought by the breakage of prestressing tendon is can not rescure. So, it is important to correctly assess the loss of prestressing force. Detecting the present prestressing force by routine method is impossible, unless puts the intelligent monitoring instruments into the prestressed construction when it is constructing.Vibration test methods and neural network identification are used to detecte the existing prestressing force of prestressed concrete beams. New experiments are supplement based on the previously test data collection. The intrinsic relation beteewen the change of frequency and the change of prestressing force is found by theoretical analysis. Modify the dynamical model of prestressed concrete beam, and then write the Ansys Parameter Design Language and User Interface Design Language finite element program with the modified dyanimaical model. Validate the finite element program with test data, and then generate more simulation beams data by validated finite element program. It can make up for lack of experimental data and large prestressed concrete beam data that can not be realized in indoor test. Create neural networks predicted the present prestressing foce of beam with test and simulation data as input data. A set of data are left as simulation input in order to test the generalization ability of neural networks. So following works are done in this paper:1. Dynamic experiments are done by 8 prestressed concrete beams. The regularity is found by summing up past data and collecting this test data: the natural frequencies of prestressed concrete beams are increasing with the increased of prestressing force, no matter what degree of concrete and no matter what site of prestreesing cable. The rangeability of frequencies increasing is not same when the ratio of slenderness and the linear eccentricity are different in the prestressed concrete beams.2. The deformation modulus is tested by single loading test program. Test results show that concrete deformation modulus is increasing with the pressure rised when the pressure is relatively small (the pressure is less-than 50% design value of compression strength of concrete). 3. In order to analyze the real cause that the natural frequencies of prestressed concrete beams are increasing with the increase of prestressing force, orthotropic materials dynamic model is using in the first. Its frequency is increasing with the pre-stressed increased, when the prestressed concrete beams are treated as orthotropic material. But this analysis method requires the solution of high-order simultaneous equations, and the calculation is cumbersome, the result is not very precise.4. Finite element theory is used to analyzied the prestressed concrete beams, and use the energy method to derive the dynamic equations of characteristic root and its eigenvector. Analysis the item involved in the equations, and find that the prestressing force does not directly appear in the equations. The deformation modulus which changes with the pressure is appeared in the stiffness matrix and damping matrix. It is proved the test result that deformation modulus of concrete increases with the pressure increaseing, and causing the natural frequencies increasing of prestressed beams.5. Treat the concrete beam as the beam structure supported compressive stress at both ends. When the prestressing tendor is tensed it will have flexural rigidity. Add the flexural rigidity of prestessing tendor to that of concrete in prestressed concrete beams, and treat prestressed concrete beams as distributed parameter system. Self-vibration frequency calculation formula of simply supported prestressed concrete beam is derived. In this formula, the prestressing force item is eliminated. It means that if the prestressed concrete beam is treated as normal linear-elastic material, then the natural frequencies of the beam have nothing to do with the prestressing force. The reason of frequency changing can only be attributed to the changing of deformation modulus with the pressure.6. Compile the static and dynamic parameterization programs with Ansys which is commercial finite element analysis software. Construct the beam model with the input parameters. According to previous study result, regard the elastic modulus as a variable of prestressing force as calculating the dynamic performance of prestressed concrete beam. By mapping transformation analysis, find decision combination factors to the changing of stiffness and frequency. Then find the function of this mapping by regression analysis. This function would be used in the dynamic model of prestressed concrete beams. The results indicate that the modified Ansys model calculating values agreement with experimental results well. To compensate for lack of experimental data, a number of prestressed concrete beams are calculated with different cross-section sizes, different lengths, and different distributed steel, and different concrete strength by modified model. These data are as the neural network inputs.7. In order to predict the natural frequencies of prestressed concrete beams, when the prestressing forces are known, the Back Propagation (BP) neural network and Radial Basis Function (RBF) neural network are constructed. At the same time these neural networks are constructed to identify the existing prestressing force when the natural frequencies are known. Input the experimental data and Ansys generated data as training samples to neural networks builded. Two beams data are lefted as simulation data test network generalization. The results indicate that RBF Neural Network in predicting the frequencies and identification of the existing prestressing force is superior to BP neural network. The trained network can be used to the works in the frequency prediction and prestress loss prediction of concrete beam structure.