Studies On Engineering Structural Modal Parameter Identification Under Ambient Excitation

Engineering structures are an important part of the national economic constructionand people’s lives, their safety, durability and maintenance while normal used bound tobe the people’s focus. For the engineering structural health monitoring and damagedetection, it is necessary to fully understand the dynamic characteristics of engineeringstructures. The modal parameters of engineering structures are the main parameters ofthe dynamic characteristics,the identification method is generally divided intotraditional modal parameter identification method and modal parameter identificationunder ambient excitation. It is difficult to adapt to practical engineering, that traditionalmodal parameter identification method need artificially incentives and the effectiveexcitation signal. Modal parameter identification under ambient excitation have beenwidely used in the engineering field and becomes a research focus without applyingartificial incentives, affecting the normal working, measuring the excitation signal, withlow-cost and more in line with the actual situation, etc.This article researched on engineering structural modal parameter identificationunder ambient excitation, proposed de-noising, given order and related improvementsmethods for several commonly used time domain identification method. To sum up, themain contents and conclusions:①Briefly introduced the importance of engineering structures, the significance ofstructural health monitoring and the damage identification, then discusses in detail theresearch background and significance of modal parameter identification under ambientexcitation. Given a literature review on the main modal parameter identification methodunder ambient excitation, and finally the main work were described.②Described the theoretical basis of the modal parameter identification in detail,including: transfer function and frequency response function for single degree freedomand multi-degree-of-freedom system; impulse response function and its relationshipwith the frequency response function; mechanical characteristics of the white noiseexcitation signal.③In-depth studies on two methods of signal pre-processing (RDT and NExTmethod) and three modal parameter identification method (STD method, the complexexponential method, ARMA method). Simulation Charpy example shows: NExTmethod in accuracy, noise resistance is better than RDT method in pre-processing method；three identification methods in noise resistance and frequency identificationhave good performance,but damping ratio identification error is comparatively large.Introduced singular entropy theory in detail, proposed de-noising method based onsingular entropy, numerical simulation and the processing of the experimental data showthat: the method can remove the noise to improve the SNR;at the same time,it canaccurately determine the system modal order, solved the determination of model orderproblem in recognition process.④Introduced the basic theory of stochastic subspace method in detail, the resultsof modal parameter identification on a12-story reinforced concrete model shaking tabletest data show that: it is feasible and accurate that structure modal parameteridentification under ground motion by stochastic subspace method.We simulates a steeltruss arch bridge by MIDAS, the identification results show: the stochastic subspace canaccurately identify the modal parameters of the bridge structure under vehicle load.Finally, we proposed the determination method of model order based on singularentropy, experimental data processing indicated: the method is feasible, can accuratelydetermine the model order,having small calculation, can automatically get the modelorder without human impact.⑤We systematically studied on the theory and identification process of HHT,numerical simulation results indicate: HHT can accurately identify modal parameters.Asmode mixing problem of EMD,used adding mask signal method, and systematic studiedthe frequency value of mask signal, proposed improved value method, the experimentaldata processing show: the method can effectively eliminate mode mixing phenomenon.Proposed three parameter identification methods based on improved EMD with addingmask signal, and used these three methods to process the experimental data, the resultsprove the correctness and feasibility of three new identification methods.