A Fault Diagnosis Method for Rod Pumped Oil Wells

Abstract

The invention relates to a fault diagnosis method for a rod pumped oil well, which obtains known and to-be-diagnosed rod pumped oil well ground dynamometer diagrams, and converts the rod pumped oil well surface dynamometer diagram into a rod pumped oil well pump dynamometer In Fig. 16, the 16-direction chain code is used to redraw the boundary chain code of the normalized pump dynamometer diagram, and the pump dynamometer diagram after the boundary chain code redrawing is divided into regions, and the feature vectors of each region are extracted. The spectral clustering algorithm based on particle swarm optimization clusters the quantized pump dynamometer map feature vectors to complete the fault diagnosis of rod pump oil wells. Quantification can effectively and meticulously describe the graphic features. The fault diagnosis of rod pumped oil wells uses the spectral clustering algorithm based on particle swarm optimization to cluster the quantized pump dynamometer feature vectors. It does not rely on training samples, which can improve diagnosis. performance.

Description

technical field [0001] The invention belongs to the field of fault diagnosis of pumping wells, and in particular relates to a fault diagnosis method for rod pump pumping wells. Background technique [0002] Rod pump pump dynamometer diagram plays a very important role in oil field production, and it is the main basis for judging the downhole working status of the pumping system. The pump dynamometer diagram can be used for fault diagnosis of downhole production conditions, and can also be used to calculate the fluid output of the wellhead, and can also be used to calculate the position of the downhole dynamic fluid level. In traditional production, technicians generally rely on manual analysis of the collected pump dynamometer diagrams to judge the actual working status of the pumping system. This manual method has a long working cycle and is greatly affected by subjective factors. It is increasingly unable to meet the needs of enterprises to continuously improve production...

AI Technical Summary

Problems solved by technology

This supervised learning method mainly has the following two problems: first, the accuracy of diagnosis largely depends on whether the sample feature vector of the training sample set can accurately reflect the typical characteristics of the sample; second, this supervised method It mainly relies on artificially dividing the training sample set, which is easily affected by subjective factors.

Images