This article details a new technique that uses intelligent methods in order to identify non-standard errors when controlling the technological process of maintenance petroleum equipment. First, a new formulation of the technological process control problem for the maintenance of petroleum equipment is presented. This is stated in terms of classifying errors introduced by measuring the parameters of the technological process. Intelligent methods have been proven as a tool to solve the classification problem. Various machine-learning methods have been considered: decision trees, artificial neural networks (ANN), and fuzzy logic. In this study, an effectiveness comparison of the proposed methods has been conducted using experimental data of petroleum equipment maintenance. Results indicate that ANN is the most efficient method to classify measurement errors. The proposed method will primarily improve repair quality of certain equipment components such as the pipeline system for transferring raw hydrocarbon materials. Moreover, it will improve the quality of maintenance work and durability of the pipeline system, which in turn can increase the hydrocarbon production efficiency.