Research Of Monitoring Data Analysis And Prediction On The Settlement Of Goaf Surface

Mining resources is important foundation of the whole society development.Usually, large-scale goaf will be formated in the process of mineral resourcesexploitation.The goaf instability will lead to serious hazards such as surface subsidence,and people’s lives and property will be threatened, the normal production activities willbe destructed, and ultimately the economic development of the region will behindered.In order to avoid such accidents, it has great significance to research thecontrol of surfacesubsidence of mining area problem. Therefore, it is very important tomonitor the goaf surface settlement real-time.The important task and direction ofsurface subsidence monitoring study is to find abnormal data timely and accuratly froma large number of surface subsidence monitoring data, warn the goaf instability situationand provide adopted law，reliable predictive value，and warning evaluation criteria.In this paper, based on the Xi Haozhuang iron ore goaf of Hebei Province, in theprocess of goafsettlement data analysis and prediction, it uses frequency characteristicsof the wavelet transform theory, the advantages of dealing with nonlinear problems byneural networks, fuzzy identifiable theory and relevance theory of neural networks.Eventualy, it establishes the wavelet denoising model, settlement prediction model andgob destabilization recognition model, and studies the instability warning of the goafcombination of the Xi Haozhuang iron ore VI-4orebody goaf. The main contents are asfollows：First, based on the study of denoising parameters including different wavelet，decomposition level, denoising approaches and methodologies, it selects waveletdenoising to analysis and remove the interference information on goaf surface subsidence monitoring data,and research the law of the monitoring regional grounddisplacement and deformation.Secondly, based on the improved BP neural network, it uses Wavelet NeuralNetwork with BP on the basis of the three-tier structure to predict the the goafsettlement etc.The equivalent indicators of the prediction model includs7variablescharacterization of the changes of monitoring regional such as sink, tilt, curvatureetc,using the scientific computing software MATLAB to achieve forecasts. It shows thatthe forecast result reflects the trend of ground displacement and deformation throughcomparison the predicted results with real-time monitoring data and it is a moresatisfactory prediction results.Finally, using the wavelet neural network fuzzy identification theory and identifingmodel of existing training data, it comes to the conclusion that monitoring regional isstability by identifing the stability of the monitoring region. On the basis of identifingthe stability of the monitoring regional, through relatively analysising the monitoringdata, refer to the Ministry of Construction promulgated《the building foundation designspecifications》and the Coal Industry Bureau developed《the buildings, water, rail andmain roadway coal Pillar Design and coal mining regulations》, it brings up theinstability criteria of the Xi Haozhuang Iron Ore mined Area. According to test resultsof the follow-up monitoring data, it shows that the standard has stronger practicality.