The Prediction And Research Of Poor Information Deformation Monitoring Based On Grey Model

The prediction and analysis of deformation monitoring has important significance in the engineering construction. Poor information deformation monitoring system has been analyzed in this paper.Several methods of data processing has been introduced which about missing data and gross error,the errors can be eliminated through matlab program. Grey model and grey-time series combination model was used to obtain the predicted results of deformation monitoring.Main research and results in this paper are as follows:1.For those engineering observation data with some certainty development trend,grey model and linear regression model were established to analyze this trend.Posterior variance examination was used to assess the precision of grey model,Engineering practice analysis proved that the precision of grey model is at first level;fitting degree examination was used to assess the precision of linear regression model,engineering practice analysis proved that the fitting rate of regression equation was high.2.The error of traditional GM(1,1) model mainly caused by the background value generated and the initial value chose. The traditional GM(1,1) model has been improved in this paper,either the background value was constructed when the weight of the average relative error is smallest or the variance of the model is minimum,both of them can improve the precision of the grey model.3.Traditional grey model,new information model and metabolism model were established in this paper, through the comparing and analyzing the experiment,we draw some conclusions: the precision of the metabolism model is highest, new information model took the second place, the traditional model was the lowest.4.Through the experiment we can found that:the length of the data have influence on the prediction accuracy. Grey model group was established when tectonic domain family was constructed based on the latest data and the oldest data.Neither the high dimension is the best nor the low, suitable forecasting dimension should be chose according to different data.So, suitable forecasting dimension should be chose according to the data characteristics in different project.After the tectonic domain family was constructed,the prediction dimension can be chose with corresponding standard.5.Two data sets were chose in this paper.For data set 1,grey model and traditional grey-time series combination model were used to predict the last five stage data,the experiment improved that:traditional grey-time series combination model is better than grey model.For data set 2,Traditional grey-time series combination model and improved grey-time series combination model were used to predict the last five stage data,the experiment improved that:the prediction accuracy of traditional grey-time series combination model is high enough but the improved Grey-time series combination model is better than it.