In order to accurately and timely judge the fault location of pumping unit and change the inefficient manual fault diagnosis mode

a fault identification method of pumping unit based on Mahalanobis Taguchi system was proposed . Firstly

the geometric and shape characteristics of the indicator diagram were analyzed. Secondly

feature extraction of indicator diagram was carried out by combining global and local features. Then

Mahalanobis distance was used as the measurement index

and the sensitive feature variables of each fault were screened by constructing and modifying feature subset. Finally

the multi-classification Mahalanobis Taguchi system was established to realize fault identification and compared with the traditional manual identification method

and the feasibility and effectiveness of the proposed method was verified by taking the oil well data in the Shixi operation area of Karamay as an example.