Fault Diagnosis Of Indicator Diagram Based On Feature Fusion

In recent years, pumping unit is the most widely way of mechanical oil recovery. Due to most domestic oilfield has entered the middle and later periods of mining, underground working condition is more and more complex, the failure rate is very high which affect the normal production of oil field seriously. Therefore, it is necessary to diagnose the existing fault of pumping unit accurately in order to take timely measures. The common diagnostic method is analyze of underground working condition according to the shape of pumping unit indicator diagram,but traditional indicator diagram diagnosis method cannot meet the needs of digital oilfield.In order to solve this problem,with the pumping unit indicator diagram image as the research object, according to the theory of image recognition,this paper proposes a indicator diagram recognition method based on feature fusion, which diagnose fault quickly and efficiently.First of all, on the basis of a large number of domestic and foreign related literature review, this paper introduces the development course of pumping unit fault diagnosis technology, analyzes the research status of multiple features fusion, puts forward applying the multiple features fusion method to diagnosis indicator diagram. Secondly, this paper introduces the drawing process of theory indicator diagram, compares the graphical features of several typical fault indicator diagram,provides theory basis for the subsequent indicator diagram diagnosis,in view of the indicator diagram image contour feature and regional feature, puts forward the feature extraction algorithm of circular degree and invariant wavelet moment.Again, this paper analyzes the principle and advantages of feature fusion in the image fusion, comparing the two kinds of feature fusion method, choose linear discriminant analysis algorithm for feature fusion.Finally, introduces the basic principle of fuzzy kernel clustering, constructs the fuzzy kernel clustering classifier based on the indicator diagram, takes examples of oilfield measured indicator diagram data,diagnoses common pumping unit faults intelligently which obtained the good effect.