A Study On The Application Of Improved Grey Model In Coal Surface Settlement Prediction

With the rapid development of our country,the increasing demand for energy,making the number and size of coal mine is also rising,At this time,how to ensure the safety of the open-pit mining is particularly important.Coal surface settlement is the main hidden trouble of pen-pit mining,which has complex mechanism,affected by many factors and nonlinear,so the top priority is how to accurately and effectively analyze,monitor and forecast the settlement.This paper made the two following research for the above problems:1.this paper improved the background value and the initial value based on the classical GM(1,1)prediction model,using the method of automatic optimum weights and least-squares method through the in-depth study of its modeling mechanism.2.on the basis of the research,we found that the classical dynamic GM(1,1)prediction theory has the disadvantages of empirical dimension selections and the origin of error,this paper adopted the theory of dimension optimization and introduced the monitoring value in real time,at the same time,this paper established an equivalent-dimensional real-time dynamic GM(1,1)prediction model.This paper study the application of the above revised models,which takes a coal surface settlement data as a data source.The results showed that we take 13#monitoring point for example,the residual error of fitted value is between 1mm and7 mm,but if we adopt the classical GM(1,1)model,the residual error of fitted value is about 10mm;For the predicted value,the residual error of the classical GM(1,1)model,the revised GM(1,1)model,the classical dynamic GM(1,1)model and the equivalent-dimensional real-time dynamic GM(1,1)model are 15-48 mm,11-39 mm,11-12 mm,1-3mm.The measurement method for the settlement of error is M=±2.2(mm).Results show that: 1.the accuracy of fitted value stand out and the prediction accuracy has certain enhancement;2.the equivalent-dimensional real-time dynamic GM(1,1)model has the best predictive effects in the coal surface settlement;3.in comparison to the non-dynamic GM(1,1)model,the dynamic GM(1,1)model has obvious advantage in long-term forecasting and it can better reflect the actual situation.