Application Of Combined Model Based On Wavelet Denoising In Building Deformation Monitoring

With the rapid development of China's economy and the acceleration of urbanization,more and more high-rise buildings have risen from the ground.Due to the external influences of various factors such as man-made and nature,buildings occur during construction and operation.Different degrees of deformation are inevitable.If the deformation exceeds a certain range,it will cause immeasurable damage to the surrounding people and buildings.The observation of building settlement is mainly to regularly observe the buildings and pass the settlement of the buildings.The processing and analysis of the observation data can timely grasp the deformation state of the building,thus providing an effective guarantee for the safety of the building.In recent years,with the advancement of science and the need for data prediction,many model prediction methods have emerged,such as time series model,neural network model,gray model,wavelet analysis,etc.Different prediction models have their own advantages and disadvantages.In practical applications,if the advantages of different prediction models can be combined and a better combined model can be established,the accuracy of building deformation prediction can be greatly improved.Wavelet denoising can eliminate and interpolate outliers in the original data when processing the data,making the original data sequence smoother.Both the grey model and the BP neural network model are commonly used settlement prediction models.The advantage of the gray model is that it has a good prediction effect when the signal is incomplete and the sample data is small,but the data prediction is exponentially changing,often and gradually.The trend of the steady settlement observation data does not match,and the constant positive concave processing and the adjacent mean method for the original data can better compensate for this shortcoming and improve the accuracy of the model.The BP neural network model has better computing power and error correction capability.Therefore,combining the advantages of the three,a gray BP neural network model based on wavelet denoising can be established by appropriate combination,which can greatly improve the prediction accuracy of the model.This paper combines engineering examples,the research results of this paper are as follows:(1)Using the MATLAB program to establish the traditional GM(1,1)model and the optimized GM(1,1)model for 101 monitoring points and 202 monitoring points,the accuracy of the GM(1,1)model is obtained by comparing the accuracy evaluation standards..The BP neural network model was constructed and predicted by two MATLAB programs.The optimal GM(1,1)model and the BP neural network model are combined by the optimal weighted combination method.The MATLAB program is also used to construct the model.Through the comparison of the accuracy evaluation criteria,the prediction accuracy of the combined model is higher than that of the single model.(2)Wavelet denoising is performed on the original data of building settlement observation,and the optimal denoising parameters are set by comparing and analyzing different parameters of wavelet denoising.Based on the denoised data,the single model is realized again by MATLAB program.And the construction of the combined model,through the comparison of the accuracy evaluation standards MSE,SSE and MAPE,it is found that the single model after wavelet denoising has higher accuracy than the corresponding single model before wavelet denoising,and the combined model after wavelet denoising has higher precision.The combined model before wavelet denoising finally concludes that the optimized gray BP neural network combined model based on wavelet denoising can better meet the prediction of building deformation prediction.