Driven by the opportunities that are created by digitalization, data-based maintenance strategies such as predictive maintenance are increasingly getting into focus. However, successful implementations of predictive maintenance are rare in practice. Reasons for this include lack of clarity in the process of acquiring process data, problems in the compatibility of IT systems or difficulties in selecting relevant process parameters. While the phases of data preparation and data modeling are frequently discussed in the literature, little attention is paid to the technical concept and the process of acquiring process data in an industrial environment. In this thesis, a process model for the acquisition of process data from industrial plants is developed. The process model extends the CRISP-DM reference model for data analysis projects with the phase of data acquisition. Based on the findings of a structured literature review on process models in the field of predictive maintenance, a process model was deductively derived. According to the design science research approach the process model was developed and evaluated in practice based on the the data acquisition process of three pilot plants. Particularly, the procedure of selecting relevant parameters from machine controls is described. Furthermore, central technical components are described and a recommendation for action for the phase of data acquisition is given. In this recommendation both the process itself and responsibilities are defined. With the realization of the data acquisition according to the presented process model, the basis for the further procedure towards predictive maintenance is laid.