The Fuzzy Calculation Method For Damage Identification And Condition Evaluation Of Bridge Under Uncertainties

With the rapid development of modern economy, the traffic load that bridge in servicehas to bear increases gradually. Meanwhile, external harsh environment leads to materials’continuously aging problem, accompany with the easy occurrence of cumulative damage forbridge structure, so as to result in the insufficient carrying capacity. Therefore, it is of greatsocial and economic value to guarantee the safety operation of the structure, promote thebridge service life and reduce the occurrence rate of collapse accidents when taking thedamage identification and condition assessment of bridge structure into account, and timelygrasping the bridge damage location and degree as well as evaluating the bridge safety anddurability conditions.However, in the process of actual structural damage identification and conditionassessment, it is inevitably affected by some uncertainties, such as the external environment,the measure noise, subjective factors and so on. The existence of these uncertainties hindersthe practical application process of the existing damage identification and conditionassessment technologies so as to reduce the reliability of the identification and assessmentresults. Thus, how to consider and net out the effect of uncertainties in the calculationprocess has an important practical significance for promoting the engineering application ofthe damage identification and condition assessment technology.Aiming at the uncertainties, such as parameter error, information missing, temperature,information missing of the reference model, as well as subjective factors and so on, thispaper puts forward the damage identification and state assessment techniques with strongrobustness, based on fuzzy nearness, fuzzy neural network, fuzzy logic, fuzzy clustering andsome other fuzzy calculation methods. The specific research work is shown as follows:1. Aiming at the condition of parameter errors and information missing caused bynoise and incomplete test degree of freedom, this paper proposes fuzzy nearness-based method in damage identification of bridge structure by taking simply supported beam bridgeas the research object. This approach takes the improved vibrator ratio as damageidentification parameters with K-type parabolic and Bell-shaped function as membershipfunction of input and output parameters, respectively, so as to builds the damageidentification knowledge base by establishing a correspondence between structural damagestate and damage eigenvectors. The structural damage localization and damage stateidentification are realized by calculating the regular fuzzy approach degree between testsamples and knowledge base, according to approximate principle. By adding Gaussiandistribution random number into raw data to simulate noise and with missing node modedata simulating incomplete information condition, numerical simulation analysis ofmulti-girder simply supported beam bridge proves good effect in the damage identificationof single-location and multi-location under the condition of parameter errors andinformation missing. Slot beam with span1.5m is selected as the indoor test model toproves the practical application effect of FNBDI method.2. Aiming at the abnormal changes of modal parameters caused by temperature change,this paper proposes a fuzzy neural network approach that can eliminate the temperatureeffects in the damage identification of bridge structure. This method simulates thetemperature effect with the changes of structural elastic modulus and based on the modalfrequencies and vibration data of the structure, the corresponding structure uniform loadsurface and its curvature value are calculated. The uniform load curvature differences beforeand after the structural damage are taken as the damage identification parameters. Besides,the if thenrule base is automatically generated through adaptive training. This paperregards the normalized temperature and the uniform load surface curvature as the inputparameters of fuzzy neural network, and the numerical simulation analysis of simplysupported beam bridge verifies the effectiveness of the method.In order to better illustrate the effectiveness of the fuzzy neural network method, BPneural network is selected as the damage identification algorithm with frequency changesquare ratio and frequency change value regarded as the identification parameters of thedamage location and degree, respectively. By constructing vector similarity formula, this paper comparatively analyzes the identification accuracy between BP neural network andANFIS method. The result indicates that the similarity between the desired output and theactual output of testing samples with the ANFIS method is higher and the identificationresult is more accurate.3. Considering that the baseline model data of existing bridge is not available or thesimulation error is larger, it will lead to the failure of the damage identification method. Thischapter proposes a bridge structural damage identification method, based on modalcurvature, the Chebyshev polynomial and fuzzy reasoning theory without consideringbaseline model information.This method obtains the corresponding modal curvature valuesMSC dthroughsecond central differencing method on the basis of the node mode data of damage structure.According to this, we select theMSC dvalue of the feature node and get the modalcurvature valueMSC uwithout damage by Chebyshev polynomial fitting. By calculatingthe modal curvature difference MCD before and after the structural damage throughMSC dandMSC u, the structural damage location identification can be realized. Thedamage degree identification of bridge structure is realized by making the normalizedMCD as the input parameters of fuzzy reasoning system. The numerical simulationanalysis of simply supported beam bridge has verified the effectiveness of this method andthe results show that the calculated modal curvature difference of the single-location andmulti-location that is not dependent on the benchmark model data, is able to achieveaccurate damage positioning. Taking the normalized modal curvature difference of therelevant nodes as input parameters of fuzzy systems, we respectively construct fuzzyreasoning system of single-location and multi-location damage identification, so as toguarantee the damage identification results of the test samples are equipped with goodaccuracy. Selecting slot beam as indoor test model and taking the mid-span unit damageidentification for example, the effectiveness of the method proposed in this paper is verifiedby the measured first modal shape data.4. Since state assessment methods of the existing bridge are easily influenced by subjective factors, this paper proposes a unsupervised bridge superstructure state assessmentmethod based on fuzzy clustering according to bridge field measured data. Firstly, thismethod builds the durability and safety evaluation index system of bridges based on fieldmeasured parameters. And then, considering or not considering the index weight effectseparately, we select a certain number of bridge health monitoring data as clusteringsamples to obtain the fuzzy similarity matrix and fuzzy equivalent matrix of samples bycalculating. Finally, we select a different threshold to form a dynamic clustering map anddetermine the best classification based on statistic analysis. The clustering result is regardedas a sample base of bridge safety and durability state assessment. Taking the average of thecorresponding indicators of the same type bridges as the approximate center of this category,this method can analyze and evaluate the bridge state for assessment on the basis ofselecting the near principle by calculating the fuzzy nearness between the unknown bridgestate data and the center’s. This paper selects the Saide bridge and the border bridge inNanping (maoshan) in Jilin Province as the physical works of the bridge safety anddurability assessment to verify the effectiveness of fuzzy clustering method referred above.