Research On Working Condition Trend Analysis Method Of Oil Wells Based On Data Stream

With the rapid development of information technology and digital communications, thecontinuous integration of a variety of technologies enrich the data type of the data elements.The development of Oil industry led to the development of various industries and technology,such as Data Stream technology,but there are still many problems to be solved. Real-timetrend analysis of Oil Wells using the dynamometer conditions which can real-time find keyinfluence factors of Oil Wells’ working conditions in fault diagnosis, fault analysis, faultwarning and hidden faults, for the protection of normal oil production and improve fieldmanagement level have important significance.Firstly,Centralized Data Mining System Architecture and Data Stream MiningArchitecture are referenced to build general framework of working condition trend analysismethod based on Data Stream,it proposed a new solution for real-time forecasting,trendanalysis and pattern anomaly. Next,use a large number of working condtions data collected tosimulate Data Stream,it studies Data Stream Mining methods,and it proposes a PiecewiseAggregate Approximation Algorithm of Data Stream based on Variable SlidingWindow(VSW-PAA) which completed pattern representation of Data Stream. Finally,it usestrend analysis techniques based on Data Stream to real-time analysis and forecasting onworking conditions data of Oil Wells,combining with SW algorithm and OSD algorithm,itproposes a trend analysis method of least squares regression based on split point, thealgorithm of trends feature extraction using multiple linear regression method, the split pointdetected by the use of the F tests and confidence intervals are two methods, while achieving asplit points and outliers test.In this thesis, an oilfield dynamometer data is referenced to form analog Data Stream, inaccordance with the general framework to build the structure, forming working conditiontrend analysis experimental system of Oil Wells based on Data Stream, the algorithm isconstantly revised, verified by experiments with a certain general framework versatility,VSW-PAA algorithm and trend analysis methods have good usability, and it detects threekinds of pattern dynamometer using anomaly detection method, detection effect is good.