The Comparison Of Modal Analysis Methods Under Ambient Vibration

In the context of civil engineering, ambient vibration testing is a very relevant topic, which can be carried out with simple devices and has no interruption on transportation. However, due to the small amplitude and strong randomness, whether it can fulfill the engineering precision aroused much discussion, especially under the assumption of stationary Gaussian white noise. Therefore, aiming at both the frequency domain method-Peak-Picking (PP), Frequency Domain Decomposition (FDD), PolyMAX and the time domain method-Stochastic Subspace Identification (SSI), careful research is conducted in the theoretical derivation, algorithm and computer realization. Then, based on the testing datas of a simple girder and the Hong Bridge under ambient vibration, the comparison on modal parameters, including frequency, damping ratio and mode shape, is made between the4operational modal analysis methods and the ANSYS calculations. Finally, analyze and summarize the characteristic, application scope and the identification accuracy of each method.The main work and conclusions are as follows:1. Program the algorithm implementations of4modal parameter identification methods, including Peak-Picking, Frequency Domain Decomposition, PolyMAX and the Data-driven Stochastic Subspace Identification method. Then testify their applicabilities through a simple girder testing.2. When constructing the3D FEM of Hong Bridge in Changde, Hunan Province, the author uses the shell element to build the main girder, the beam element for towers and combine-element to simulate support stuffiness. The comprehensive element-selection principle minimizes the time cost, while still keeps its high confidence. The analysis result shows a remarkable theoretical benchmark, which means a crucial role the refined FEM plays in choosing reference points, arranging test points and confirming real modal parameters.3. The modal parameter identification result of Hong Bridge under ambient vibration clearly illustrates that PP, FDD, PolyMAX and Data-driven SSI can be appled to complex strctures, each has its own advantages and disadvantages. Overall speaking, the eigenfrequency is the most stable modal parameter and fits the FEM result well. Besides, for methods using the stabilization diagram technique, bigger damping ratio covariance turns up where stable points are less. The low signal-to-noise ratio (SNR), modal energy and algorithm accuracy may degrade the shape precision greatly and make it hard to eliminate false modes. In this case, FEM always aids to tell the ture one from a mass of modes.4. Considering that each method can lose modes or generate a false one, the stabilization diagram, as a complementary technique, offers us a tremendous help. Essentially, compare with the classic measurements based on both the input and output data, the system identification method under ambient vibration has an inherit inferior with unknown inputs and less available information. Thus, in order to draw an authentic conclusion, engineers are recommended to adopt at least2separate methods.