In-process softsensor for surface-conditioning during longitudinal turning of AISI 4140

• Applicants: Professorin Dr.-Ing. Gisela Lanza; Professor Dr.-Ing. Volker Schulze; Dr.-Ing. Bernd Wolter
• Subject Area: Metal-Cutting and Abrasive Manufacturing Engineering ; Measurement Systems ; Production Systems, Operations Management, Quality Management and Factory Planning
• Term: since 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 401821233

The objective of the entire research project is the development and qualification of an in-process capable, multisensory and modularly expandable control concept based on a soft sensor. This control concept is implemented for the application case of external longitudinal turning of quenched and tempered AISI4140 by means of suitable model approaches. The challenge in this machining process is to avoid the formation of thermally induced white layers near the surface and the associated tensile residual stress states, which are disadvantageous from a technical point of view. On the other hand, mechanically induced white layers associated with residual compressive stresses are to be permitted, so that the control concept simultaneously addresses several target variables of the surface layer state that are mutually dependent on one another.The goal of the second funding phase is validation of the process control by means of application scenarios, based on a multimodal soft sensor and the actuating variables cutting velocity and feed. The technological target values of the cutting process are to be identified by a multisensoric approach in order to transform target value deviations by a control module into actuating variable values. The process control requires a hardware and software update of the used CNC turning machine. A main task of the project phase is the realization and optimization of a multimodal soft sensor, which can be adapted efficiently to the target variable and can be fused by integrating, substituting and reducing sensor principles, taking into account a low combined uncertainty. This requires further development of the micromagnetic and acoustic nondestructive testing techniques with regard to efficiency and robustness. Micromagnetic simulations are used to support the experimental investigations in the project. Furthermore the process strategy of longitudinal turning is adapted by the use of emulsion flood cooling, in order to reach higher tool lives and hence to guarantee the industrial relevance of the research project. Experimental and numerical cutting analyses will be conducted, to adapt the sensor and process models to the new environmental conditions.The control relies on the database of longitudinal turning, which enables the derivation models equations for the mapping of actuating and target variables. For the control, classical approaches are pursued, as well as a model-predictive and a reinforcement learning approach are pursued. In the latter case, an agent models the technical characteristics of a classical control module. The control approach which proved suitable in the implementation phase, thereafter should be verified and validated with different application scenarios.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2086: Surface Conditioning in Machining Processes
• Co-Investigators: Dr.-Ing. Michael Gerstenmeyer; Dr.-Ing. Benjamin Straß