With the introduction of energy management systems, an analysis of load profiles of manufacturing plants becomes increasingly important. Each manufacturing plant is characterized by a process and product specific power consumption. Often, electric drives are the main power consumers. In this paper methods for pattern recognition in load profiles of electric drives are presented on the example of a multiaxial lathe. A transfer of techniques used for speech recognition e.g. Hidden Markov Models, Fourier and Wavelet Transforms to manufacturing application is discussed. In combination with energy measurement systems, those techniques proved to be a good solution regarding energy efficiency calculations and derivation for key performance indicators. The investigated methods can also be applied to other process data with significant cost advantages, because a lot of process information can be extracted from a single sensor