Study On Numerical Optimization Of Safety Monitoring Model And Deformation Early-warning Index For Concrete Dam

Data processing of dam observation is the important research content of safetymonitoring model. Aiming at fitting residual, component extraction and indexdetermination, taking concrete dam as the research object, combining with the damdeformation prototype observation data, and comprehensive using of statisticalmethod, genetic algorithm, artificial neural network, chaos theory, least squaressupport vector machine (LS-SVM) algorithm and finite element method, thenumerical optimization methods for safety monitoring model are studied, the inversemethods for dam main physical and mechanical parameters are explored, and theprotocoling methods for dam deformation early-warning index are provided in thisdissertation based on the establishment of concrete dam safety monitoring model. Themain research contents are as follows:(1) The methods for establishing concrete dam safety monitoring model arestudied. On this basis, the dam deformation fitting residuals are forecasted with themethod of neural network optimized by genetic algorithm, chaos theory and phasespace reconstruction technology. Taking the prediction of fitting residuals as thechaos factor of deformation safety monitoring model, a concrete dam chaos hybridmodel which considering residuals chaos factor is established, and its availability isverified.(2) The inverse methods for concrete dam comprehensive elastic modulus areexplored. On this basis, a LS-SVM inverse algorithm which can reflect the nonlinearmapping relation of dam deformation and elastic modulus is presented. And theprocedure of this inverse algorithm is developed based on MATLAB.(3) The deformation process and the variation feature of concrete gravity damand arch dam are analyzed. And the methods of protocoling concrete damdeformation warning index are studied based on the positive and negative analysis ofdam monitoring data.(4) Taking one concrete gravity dam which is in service as research object, thedam deformation statistic model, hybrid model and chaos hybrid model whichconsidering residuals chaos factor are established based on the analysis of horizontal deformation change rule and above-mentioned methods. And the dam deformationearly-warning index is protocoled with the positive and negative analysis of damdeformation monitoring data, so as to provide theoretical basis for judging the state ofthe dam safety.