Research On The Method Of Tertiary Oil Recovery Early Warning Based On Data Mining

In the production process of oil field, the proportion of the tertiary oil to total oil production is increasing, the scale of industrialization of the tertiary oil has expanded year by year. Beacuse of the factors that affect the tertiary oil production is increasing day by day,thus the expert experience and knowledge system for the analysis of the tertiary oil recovery is still to be improved, the difficulty of abnormal early warning is higher and the processing efficiency of production is low.However, the current early warning system for the tertiary oil is the lack of a more systematic approach. At present, the main basis to complete the early warning is according to the work experience using the method of characteristic factor analysis and other traditional methods.These lead to not only low efficiency, large resource consumption, but also poor accuracy and timeliness.Aiming at this problem, this paper puts forward tertiary oil early warning model based on data mining. Firstly, the original data is analyzed according to the target of the early warning, introducing time series to do the data processing with the best granularity.Through the rapidly determination function to screening data in order to obtain the effective influence factor set of high precision. Then, comparing many data mining algorithms and studys for the sake of selecting and improving the nonlinear regression method to build the warning model combining with the preliminary data processing results. For the mining optimal model, the significant validation of the model is necessary. Finally, the model is applied to the actual production by real-time monitoring the change of data and fitting the production data to the optimal model. This will be able to timely warning tertiary oil production anomalies, the efficiency of abnormal handling will be improved and the loss of the actual production will be reduced.The early warning system of the tertiary oil production was developed based on the early warning model of data mining. Taking historical data of an oil production plant as an example,the rationality and the implementation of the model are verified. Production exception handling efficiency is improved, and the system based on the model also has good application value.