Modeling Of Time Series And It's Application In Forecast Of Coal Mine Gas

With the Internet and information time coming, it is unavoidable that a great amount of data appears. People are almost submerged in sea-like data while knowledge is so limited. Because inefficient traditional data statistic method can not fit for the above situations, and then, data mining technical compensates this disadvantage point and develops greatly. Till now, data mining technical was widely used in the business activities and military areas, etc.The analysis of the time series is a statistic theory which is based on dynamic system structure and rule. The basic thought is based on finite records of system(observed data),then establishes the dynamic dependence mathematics model accurately ,at last forecasts the future of system.This thesis keens on the DM utilizing research in coal mine gas forecasting. It is explained about domestic and outboard researching present conditions of time series data mining, time series analysis, formalized expression and model establishment of time series. Through analysising the features of coal mine gas observing data and several species of model. This thesis declares a suitable data and model ARMA. At the same time, this thesis creates a model of specific data and also analyses test and verify it's conclusion. The concrete constant concludes the followings:Firstly, on the base of a lot of document arrange this thesis summarized the time series data mining and use it in coal mine gas checking. Also, the thesis analyzes the feature, the related problems and the technology of time series data.Secondly, in it, the time series statistic feature, stability, statistic model and data formalization are explained in details. After analyzing the AR(Auto regression), combined with the feature of coal mine gas data, the thesis defined that using AR model as coal mine is suitable.At last, according to the ARMA(Auto Regression Moving Average) creation method , the thesis established a model and tested and verified the definited figures. Time series model creating is away that concerned with the background knowledge and stability of time series considered the changeable pattern of variable and use fringe inference system to desirable the variance of time series. This thesis has the following steps to do with the whole mining process of time series: pre-dealing process of dynamic data, the choosing and recognizing of the model's form, the initial estimating of the model's parameter, the subtle estimating of the parameter, the judging of the result and the testing and mending of the model.There are a lot of time series data in the world and it is important to apply the research of time series into practice, so the theory should be used into the real world after testifying its' theory mainly used on the subject of model creation, prediction, test and improvement of coal mine gas density data. The fact shows that it is an effective method which has a leader function to practice.