The authors analysed the noise-induced problems with diagnostics of the sucker rod pumping units (SRPUs) related to the peculiarities of their field operation. The authors demonstrated that the existing diagnostic methods based on interpretation of the dynamometer cards built on the signals received from the force and stroke sensors did not allow to solve the diagnostic and control problems in real time. Therefore they proposed a technology for determination of the robust normalized correlation functions, which were used to form combinations of informative attributes corresponding to the possible emergency states of SR-PU. Identification of those states is duplicated by determination and formation of combinations, which correspond to the noise characteristics of the force signal and improve the reliability of the results. When the proposed technology is introduced in the oilfields with a large number of wells, the combinations of the relevant reference coefficients will be determined for each SPRU during their operation, as various kinds of faults occur. Thus, after a certain period of operation, the combinations of the reference coefficients for the corresponding fault types will be formed and saved in the identification units of the SRPU robust control systems at all wells. When the reference coefficients for all the possible SRPU fault types are formed and saved in the identification units, the system will be switched to the automated mode of identification and control. Simplicity of realization of the processing algorithms makes it possible to solve the problem of the force signal identification by means of inexpensive controllers in real-time mode. Application of the technology in more than 100 real objects demonstrated that the profitability of the oil wells increased significantly due to the energy savings and an increase of the overhaul period.