Research On Optimization Algorithm Of Key Design Parameters Based On Built Continuous Rigid Frame Bridge

With the development of continuous rigid frame bridge,it has become one of the main bridge types in the field of bridge by virtue of its good mechanical characteristics,large span capacity and low project cost.But at present,the design of continuous rigid frame bridge is mainly based on the previous experience,supplemented by software calculation,so a lot of design tests need to be carried out in the early stage of design,so it is particularly important to improve the design efficiency.In this paper,based on the Fujiang bridge project,combined with the existing bridge design experience,combined with the self-learning ability of intelligent optimization algorithm,the key design parameters of the bridge are optimized.At the same time,the finite element model of the bridge before and after optimization is compared and evaluated based on the entropy method（1）Based on the project of Fujiang super large bridge,this paper analyzes the sensitivity of key design parameters of continuous rigid frame bridge,and determines five design parameters as optimization parameters:power of beam bottom,ratio of side to middle span,ratio of height to span of fulcrum,ratio of height to span of middle span and spacing between piers and legs.（2）Based on the statistics of more than 200 continuous rigid frame bridges at home and abroad,the empirical values of the above design parameters are obtained from the statistical analysis of My SQL database system:the power of beam bottom is 1.5～2.0parabola,the ratio of side to mid span is 0.5～0.7,the height of fulcrum beam is 4～15m,the height of mid span beam is 2～5m,and the net spacing between pier legs is 3～7m.Through the orthogonal design table and uniform design table,the optimal test results are as follows:the power of beam bottom is 1.6,the ratio of side to middle span is 0.53,the ratio of beam height is 3,and the net distance between pier legs is 7m,in which the height of fulcrum beam is 12.5m,and the height of middle span beam is 4.17m.（3）On the basis of orthogonal table experiment,₉（3⁴）design sample,₁₆（4⁵）design sample,₂₅（5⁶）design samples and 100 groups of samples obtained by parameter traversal design are combined to form 4 groups of test samples respectively,which are used as the training samples of neural network system.The training results show that the neural network achieves good fitting effect,meets the convergence conditions,and the training effect fully meets the engineering requirements.（4）Combining the trained neural network with genetic algorithm,taking the Fujiang bridge as the engineering background,the combination of design parameters is predicted and optimized.Through the combined intelligent optimization algorithm,the average value is calculated for 10 times,and the variance of the results is analyzed.Under 100groups of sample data,the optimal parameter combination is obtained as follows:the power of beam bottom is 1.6,the ratio of side to middle span is 0.53,the ratio of beam height is 3.298,the net distance between pier legs is 7m,the height of fulcrum beam is12.5m,and the height of middle span beam is 3.79m.（5）Based on the principle of entropy method,the bridge performance evaluation is divided into three first level evaluation indexes and nine second level evaluation indexes.Three first level evaluation indexes are girder deflection,girder stress and structural weight;nine second level evaluation indexes are midspan deflection,midspan deflection,midspan stress,midspan stress and structural weight.Through the weight of each index,the final evaluation score of each model before and after optimization is obtained.The results show that the final evaluation scores of the optimized scheme are higher than the actual situation of Fujiang special project,indicating that the optimization effect is obvious,verify the optimization results.