Research On Fault Diagnosis Of Pumping Unit Based On Support Vector Machine Optimized By Firefly Algorithm

In the domestic and foreign oil field production process,the mechanical oil production still occupies a large part.In the entire mechanical oil extraction and other oil extraction equipment for oil production is very important.Pumping equipment,such as pumping unit,is very important to the production of oil field in the whole mechanical oil recovery.As the pumping unit operates mostly in the field,running in the high temperature long-term,high load and other harsh environment,and the pumping unit working condition down the mines is extremely complex,so it has the breakdown frequently.While the economic loss caused by pumping failure is enormous,so it's necessary for pumping to diagnose the faults of pumping unit.Machine learning method based on artificial intelligence can be used to construct multi-class classification learning model and fault diagnosis,and has become a focus of current research.Therefore,this paper presents a method of fault diagnosis of pumping unit based on the support vector machine optimized by firefly algorithm,and the main contents are as follows:First of all,according to the problems encountered in the production of oil field,the overall structure of pumping unit working condition diagnosis system is designed.Then the formation of the theoretical dynamometer and the characteristics of some typical fault dynamometer card are studied.And then the preconditioning has been done in the means of image gray,filtering,and edge detection.The dynamometer card is transformed into the image with the largest boundary region filling.Secondly,in view of the incomplete information contained in the Hu moment eigenvalue,The wavelet transform has strong anti-interference ability and better ability to reflect local information,so Wavelet moment invariants is adopted to extract the local and global feature of pumping unit indicator diagram and then set up sample library of typical indicator diagram.On the basis of this,the in-depth analysis of the vector machine algorithm is carried out.The selection of SVM parameters and the combination of parameters(penalty factor c and kernel function parameter ?)will affect the classification accuracy,so the firefly algorithm is introduced to optimize the parameters c and ?.In order to avoid the local optimal solution of the traditional firefly algorithm,the improved algorithm is applied to the parameter selection of SVM.Compared with the SVM optimized by improved particle swarm and traditional firefly algorithm,the SVM optimized by improved firefly algorithm has better classification accuracy.Therefore,the SVM optimized by improved firefly algorithm is applied to the fault diagnosis of pumping unit and establish the SVM fault diagnosis model.The test results show that the method has high accuracy and outstanding anti-jamming ability.On the basis of above research,the fault diagnosis system of pumping unit is designed and tested by the means of C# programming language and ORACLE database.The test results show that the system is stable and reliable.Therefore,the SVM method based on the optimization of firefly algorithm can be applied to the study of pumping unit fault diagnosis.