Research On Predicting Flecural Capacity Of RC Beam Strengthened By Bonding Based On Neural Network Algorithms

The author have a try to apply the techniques of artificial network, which has been applied in fields of many kinds, in the field of strengthened structural engineering in this paper. Firstly, the author analysis and studies the theoretical results in beams strengthened with extended bonded steel plates and CFRP. Then an developed method based on BP network is put forward, which connects coefficient of momentum and adaptive learning ratio, to solve the complex nonlinear problem of multi-factor reasons influencing the efficient of strengthened structure. Furthermore the author believe that it is benefit for us to research and establish the intelligent system about strengthened structural engineering.At the same time, The author collects the actual data in the experiment and process them into BP nerve network training module, establish two concrete model of nerve network to predict the bearing capacity of RC beams strengthened with steel plates and CFRP . In the end, applying with the BP intellectual analysis model, develops into a predictable applicable software. The results prove that it is feasible to solve the predicting problems with this software. If the technical approach and the applicable software is used, the author believe that it will be applicable and contribute to developing of the techniques of strengthening structures.