Method On Deflection Enhancement Coefficient Calculation Of Steel-concrete Composite Beam Based On BP Neural Network

Steel-concrete composite structure is a new structure comprehensively utilizingthe advantages of both reinforced concrete and steel beams, due to its significanttechnical and economic benefits. Currently, this new structure has been widely used inmany practical engineering, including city construction, roads and bridges and superhigh-rise buildings.Steel-concrete composite beam deflection in the normal state is an importantcontrol parameter in the engineering design of roads and bridges. Concrete bridgedeck and steel beam are connected into an overall flexural member through shearconnectors, because the strains of the steel beam and concrete at the flange interfaceare different and shear connectors will produce elastic and plastic deformation aftershearing itself, resulting in a relative slip effect, and an increase of deflection.Previous studies have found that, even composite beams with full shear connection, innormal use, the Horizontal displacement effects at the materials interface will alsomake the beam cross section to produce about15percent additional deflection.Therefore, correctly calculating the deflection of composite beams in normal use stateis particularly important.Many scholars at home and abroad have studied on the calculation theory ofcomposite beam. The deflection calculated by transformed section method thatignores the slip effect has a large inaccuracy, causing an unsafe calculation result.Some mathematical analytical solutions are derived based on various theories, whichare not conductive to bridge design due to their too cumbersome theory applicationand formula calculation. Reduced rigidity method adopted by the current specificationtakes the slip effect into consideration, based on simplified formula after theoreticalanalysis of the beam micro-segment. But rigidity reduction factor formula is morecomplicated to calculate a large workload. Studied have found that when calculating the partially connected composite beams, the inaccuracy is relatively larger. And somerelevant studies show that when stud rigidity is small, with the increase of this studrigidity, the rigidity reduction factor of composite beams decreases contrarily, whichruns counter to the actual situation. Thus, it is very necessary and of great practicalvalue to propose a simple and accurate method.View of this, this paper proposed a deflection calculation method, which isdeflection increasing coefficient method for composite beams. Take simply supportedI-shaped composite beam bearing uniformly distributed load and mid-spanconcentrated load for example, when considering the slip effect influence ofcomposite beam. Based on the composite beam ignoring slip effect, the concept ofdeflection increasing coefficient is introduced. After taking BP Neural Network intothe calculation of mid-span deflection increasing coefficient for composite beams, theclosed form solution of deflection increasing coefficient for composite beams can beachieved. The neural network training and testing samples come from the ANSYSmodel after checking the test data. This method provides a fast and accurate deflectioncalculation for the design and testing of the composite beam bridge, due to its highaccuracy, and good versatility.