Tribological diagnostics is a non-destructive and non-disassembling diagnostic method, that uses lubricating oil as a source of information about processes and changes in mechanical systems, to which it is applied. The paper deals with the mathematical processing, monitoring and analysis of the oil field data obtained from the atomic emission spectrometry in frame of the tribodiagnostic oil tests. The mathematical methods based on a regression analysis and calculations are used in the paper for oil data analysis. The whole assessment procedure and their outcomes serve as identification of a suitable approximation trend. With the same result we are capable of determining in-service operation history as well as giving inputs to maintenance optimisation. Nowadays the system requirements are set up and evaluated in various manners. We would like to keep both preventive and corrective maintenance costs as low as possible. A system design and a maintenance system help to fulfil this task. We propose to use one of the approximations presented in order to track the system operation behaviour.