Study On Reliability Assessment System Of Existing Buildings And Software Development

It's of great economical and social significance in assessing the large amount of existing buildings reliability of our country and timely maintaining and reinforcing. To develop a special structure assessment software, this paper combines the practical assessment methods and neural networks based on some related identification standards. Main work shows as below:Firstly, determining the reliability evaluation index of components, sub-units and identification unit of existing building, and giving a quantization standards to those non-quantitative reliability evaluation index, in order to establishing a complete reliability assessment framework of existing buildings.Secondly, the preparation of BP neural network evaluation and test the subroutine subroutine. It encodes the BP neural network program, based on the theory of structure and back propagation algorithm. The neural network evaluation model is established and the weights and thresholds of network structure are given by using identification units' samples,collected. Through the analysis of network training, the error convergence rate, error precision and accuracy assessments etc, the applicability of self-encoded BP neural network program in reliability evaluation has been verified.Finally, by calling mcc compiler, associated user interface files and callback function files are compiled. Package related data packet and the background image, remove the black dos window, to generate the reliability assessment software(BP-EBRAS) of existing buildings, which can be run independently. The evaluation software has a friendly style, and operates easily, and it basically meets the functional requirements of reliability assessment of existing buildings.