Vehicle Integration
Working-student role in vehicle integration, on virtual full-vehicle assemblies. Scope and method are on the project page.
Open the vehicle-integration project pageMihir Raj Rathore
Virtual CNC/CAM planning in Siemens NX and a vibration condition-monitoring prototype for a CNC machine.
rathore.mihirraj@gmail.comSelect a block to jump to its stage. On touch screens, tap a block.
Near-real-time acquisition cycle. The block highlighting order is illustrative, not per-stage timing. Timing is not stated.
I am a mechanical engineer based in Rosenheim, Bavaria, with an M.Eng. in Engineering Sciences from Technische Hochschule Rosenheim (exam passed 20 November 2025).
As a working student in vehicle integration at BMW AG (04/2024–04/2025) I ran virtual contact-point and clash analyses on full-vehicle assemblies in CATIA V5. In my Master's work I planned a virtual CNC machining process in Siemens NX CAM and built a vibration condition-monitoring prototype for a CNC machine.
I am looking for engineering roles in Germany and Austria and am willing to relocate.
Full scope statements (what I did, what was virtual, what was not done) sit on each project page.
Working-student role in vehicle integration, on virtual full-vehicle assemblies. Scope and method are on the project page.
Open the vehicle-integration project pageI developed the virtual CNC machining process and simulation in Siemens NX CAM. Physical machining trials were not conducted.
Open the CNC & CAM project pageA sensor-to-dashboard prototype that acquires raw vibration data from a CNC machine, processes it in Python and applies rule-based alert logic against baselines.
Open the thesis project pageAn independent Python prototype for screening geometric relationships in synthetic geometry and generating structured inspection records.
Explore the prototypeProfessional and academic work are kept apart. Each entry states its scope.
BMW AG · Munich
Virtual contact-point and clash analyses on full-vehicle assemblies in CATIA V5.
Vehicle-integration project pageTH Rosenheim · Spinner VC850
A sensor-to-dashboard prototype that acquires raw vibration data from a CNC machine, processes it in Python and applies rule-based alert logic against baselines.
Thesis project pageTH Rosenheim · Siemens NX CAM
Developed the virtual CNC machining process and simulation in Siemens NX CAM. Physical machining trials were not conducted.
CNC & CAM project pageTH Rosenheim · industry partner TGW
Coordinated a five-person student team; developed concepts for several assemblies; CAD in CATIA V5.
Vellore Institute of Technology
“Nonlinear Transient Response of Viscoelastic Sandwich Panels with Composite Laminate Face Sheets.” Contributed to model formulation, debugged the MATLAB code, evaluated and interpreted results.
ELEATION · DANISH Private Ltd. · JK Lakshmi Cement & Power Plant
ELEATION (CAE intern, 03–05/2021): structured ANSYS Mechanical FEA training with a static-structural analysis of a rack-and-pinion steering assembly. DANISH Private Ltd. (mechanical design intern, 10–12/2020): supported mechanical assemblies and technical/fabrication drawings. JK Lakshmi Cement & Power Plant (intern, 01–03/2021): exposure to plant operations, maintenance and troubleshooting.
Specialist field: Mechanical Engineering and Plastics Engineering. Exam passed 20 November 2025.
The machine under test is a Spinner VC850 vertical machining centre with a Heidenhain TNC 640 control, in the university lab. The signal of interest is vibration at the spindle housing.
The prototype was built around this one machine. No claim is made about other machines or machine types.
An IFM VSA004 piezo accelerometer is stud-mounted on the spindle housing. It is one sensor at one measuring point.
The IFM VSE100 diagnostic module processes the sensor signal at the edge and records raw data when commanded.
An IFM VOS050 OPC UA server, with a VOD001 device licence, connects the VSE100 to the Python client. A passive subscription did not deliver the raw data. A command-and-retrieve routine did:
UaExpert was used to explore the server before the client was written.
A Python 3.9 client runs the command sequence and passes records through a thread-safe queue (producer–consumer), so acquisition and processing run in separate threads.
Recording, transfer and analysis repeat as a near-real-time acquisition cycle.
Each record is detrended, band-pass filtered with a 4th-order Butterworth filter and normalised with a Z-score.
Time-domain features are RMS, peak, crest factor and kurtosis. The FFT gives the spectrum.
Baselines were recorded with the machine running without load, thermally stabilised, at discrete spindle speeds.
Alert and alarm thresholds were defined in the thesis from baseline statistics using μ+2σ and μ+3σ. Rule-based diagnostic logic is applied to the features, and values are trended in persistent CSV files.
| Test | Result |
|---|---|
| Artificial impact applied to the sensor | In a healthy-machine test, an artificial impact applied to the sensor produced a kurtosis excursion from about −0.1 to above 230, and the alert logic responded. |
| 2000 RPM, aluminium 6061: no load versus light facing cut | Raw RMS 0.03 g in both. Normalised kurtosis 26.29 versus 26.13. Kurtosis values were similar between the two states and are not interpreted here; one machine, one light cut, one spindle speed. |
A Streamlit dashboard shows current feature values, alert state and trend history.
The full statement of what I did and did not do for this project is on its project page.
Emailrathore.mihirraj@gmail.com