The Research Of Gas Emission Quantity Prediction Based On Chaotic Time Series

Various factors with a complex relationship between them and much random disturbancecomposition affect the amount of mine gas emission, and there exists great difficulties inaccurately predicting the mine gas emission, for the methane outflow law presents a dynamicnonlinear variation process. To simplify complexity, many scholars view it as a linearrelationship between affecting factors, such as gas emission and coal seam’s depth, thethickness and gas pressure, and make prediction research on gas emission with the sciencestatistics theory, analogy method and linear regression method. At present, the existingprediction methods have their own applicable conditions, but lack consideration on nonlinearvariation process of the amount of gas emission. In addition, because of numerous affectingfactors, it is impossible to consider all the influence factors while choosing predictors, andwith certain subjectivity in the choice of prediction model. The development of Chaos theorymakes it possible to select prediction models without establishing subjective model inadvance and without consideration of the relationship between various affecting factors, butaccording to the objective law calculated from the sequence of the actual observation data,which can avoid subjectivity and increase the prediction precision and reliability.This paper attempts to forecast the gas emission using chaos time series forecastingmethod, and reconstructs phase spaces to the original time series of the amount of mine gasemission by choosing proper time delay and embedding dimension m. Identify thechaotic property of gas emission system with the calculation of the maximal Lyapunovexponent. At the same time, in allusion to the problem in C-C algorithms that rounding errorsof subsequence’s correlation integrals are strengthened and expanded in calculation process,improve the calculated mode of subsequence’s related integrals. To the question that theblindness of choosing field radius in G-P algorithm will cause the calculation amount toolarge, optimize its calculation process. Introducing the concept of correlation degree in GreyTheory into the first order weighted forecasting model makes the forecast mechanism morescientific and reasonable. Analyze and predict the time series of gas emission in three testpoints of coal mine using the improved method. The results show that the gas emissionsystem has the chaotic property, using chaotic time series forecasting method to forecast the amount of gas emission is feasible, and the prediction accuracy is improved after improvingsome parameters algorithms in the chaotic time series prediction algorithm.