For the urban pipeline is very complicated, by using the different between BP ANN which is good at dealing with many a data and studying by itself and Expert System which is good at word expression and logical reasoning, a comprehensive assessment model for pipeline's leakage has been established which can use to predict the urban pipeline's lifetime or pipeline's leakage based on BP ANN and Expert System. The qualitative and quantitative analysis has been gotten througth the assessment model.Firstly, five kinds of reasons for pipeline leakage has been classified after confirming the conception of leakage that is easy to misapprehend.By using Matlab training network, the best BP ANN model has been obtained also with the BP ANN's own characters. At the same time, the BP arithmetic was improved with the conjugate gradient method. The disadvantage of BP ANN, which is slowly to constringe and easily to concuss, has been inproved. The function of this model has been checked by fitting the output data of the network and contrasting with the pipeline leakage facts in Tai Yuan,Secondly, the reasons of leakage has been analysed by using fault tree. The rules, collections of facts and conclusions used in Expert System have been got. It is proved that the forward consequence ratiocination model was good at getting the direct reasons of leakage by communication with users.At last, the software of the model has been established in this paper. It has been programmed based objective original method, which is in VC++ of the Microsoft Visual Studio. The connection between the Access database which stores large number of data useful for BP ANN and Expert System has been carried out by ADO(ActiveX Data Object).
|