Fuzzy Neural Network In The Deep Foundation Deformation Prediction

The deformation control design for retaining structure in deep excavation is an important subject in the field of underground engineering nowadays. And the control and designs for retaining structure is at first displacement prediction, in anther words it is to make a prediction and analyze for the displacement law and trend of retaining structure. For the complicated project geological factor and the mechanics mechanism of deep foundation pit deformation is also unable to be found out totally and stated accurately so far, so, it is difficult to attempt to carry on number value analysis through setting up totally correct mechanics models, Its accuracy is doubtful too. In fact, the deformation control goal of deep excavation not only includes retaining system, but also includes the adjoin environment. Its variety, fuzzy quality, mutability, randomness predicted degree of difficulty of the question after determining out of shape. In addition, retaining structure generally speaking made for the provisional project, but the provisional how long time is still a fuzzy concept of the quality for the provisional engineering, it is closely related to construction speed actually. The author thinks that the system of deformation control design for retaining structure in deep excavation is a fuzzy system actually.Seeing that there are a lot of uncertain and fuzzy factors in the deformation control system of deep foundation pit, it is a typical nonlinear kinetics system. Hence it is difficult to set up satisfied mathematics model utilizing traditional method relatively. In this paper, the author combined the fuzzy set theory with neural network technology, used the expertise and confirm to be under the jurisdiction of function, Adopted the ordinary non-linear structure as the fuzzy neural network of neural network structure formed of neuron directly, And it is under the jurisdiction of degree fuzzy number which inputs and export the information to import the corresponding network . Thus according to fuzzy neural network, displacement predictionmodel has been set up .Based on MATLAB programming environment, prediction procedure has bee made for this model. After working out the fuzzy neural network out of shape, Subside and predict that carries on the instance analysis in the location of the earth's surface of project of hole of the base deeply to some of Shanghai, The result indicates that the method of this text has better precision, adaptability and common ability. In addition, utilizing fuzzy neural network model of this paper, deformation prediction result can be revised according to accurate analyze between expert and actual displacement situation, it can prevent traditional neural model unable to change the structure of models in real time network model, and the shortcoming of the adaptability of the sudden change situation of future.