Source-linked AI summary
A Cyber-Physical Machine Tool Framework with a Real-Time Machining Process Digital Twin
Khalil Chakal, Tero Kaarlela, Jose Outeiro, Carlos Andrade
TL;DR
Existing CPMT implementations provide limited synchronization between machine-tool and machining-process representations. This paper introduces a hierarchical framework integrating synchronized machine and process DTs with real-time workpiece and vibration data, then demonstrates real-time operation and persistent process management. The framework preserves monitoring and teleoperation while supporting replay, traceability, and future synthetic-data generation.
Problem
Existing machine tool and machining process DTs have been developed independently, limiting integrated synchronization within a unified CPMT framework.
Method
The paper extends a CPMT framework with synchronized machine and machining-process DTs, voxel workpiece updates, vibration measurements, and a persistent Part DT Repository.
Results
The framework maintains synchronized digital representations of the CNC machine and evolving workpiece while integrating machine states, geometry, and vibration measurements.
Takeaways & Limitations
Persistent synchronized machining information supports process replay and workpiece traceability while preserving machine-tool monitoring and teleoperation.
Takeaways & Limitations
MTConnect acquisition uses discrete 20 Hz HTTP sampling, so intermediate tool positions are not captured and higher rates trade against reconstruction and transport limits.
Abstract
from arXiv · showhide
Digital Twins (DTs) have emerged as a key technology for improving the monitoring, optimization, and automation of manufacturing systems. However, existing Cyber-Physical Machine Tool (CPMT) implementations primarily represent the machine tool, while the machining process remains only partially synchronized with its physical counterpart. This paper extends a previously presented CPMT framework by introducing a hierarchical DT framework that simultaneously maintains DTs of both the machine tool and the machining process. The proposed framework integrates real-time CNC operational data, a voxel-based workpiece representation, synchronized process vibration measurements, and a persistent part DT repository for process replay, traceability, and future synthetic data generation. Experimental evaluation demonstrated real-time operation at a 20 Hz machining-state update rate, interactive visualization exceeding 100 frames per second, and a mean depth reconstruction error of 0.16 mm. The implementation provides a foundation for AI-assisted machining applications while preserving the machine tool monitoring and teleoperation capabilities.
I. INTRODUCTION
Digital twins are being applied to improve monitoring, simulation, optimization, and control of CNC machining. However, existing work provides limited integration between synchronized machine-level representations and machining-process representations.
- Digital manufacturing tools supported by DTs and AI provide opportunities to improve machining quality, productivity, and resource efficiency.
- Recent DT approaches support monitoring, simulation, and optimization of CNC machining processes, including tool monitoring and closed-loop process control.
- Existing CPMT work enabled bidirectional communication between a physical CNC machine and its digital counterpart for real-time monitoring and control.
- Machining process DTs represent tool-workpiece interaction during material removal and use workpiece models to estimate evolving geometry and support adaptive machining and virtual verification.
D. Research Gap
Prior machine tool and machining process digital twins have largely developed as separate representations. This motivates a hierarchical CPMT framework integrating synchronized machine and process twins with persistent part data.
- Existing machine tool DTs primarily represent CNC operational state for monitoring, teleoperation, and operator training.
- Machining process DTs emphasize process simulation for toolpath verification and optimization.
- Existing machine tool and machining process DTs were developed independently rather than as a unified cyber-physical machining framework.
- The proposed work extends the CPMT framework into a hierarchical architecture that synchronizes DTs of both the machine tool and machining process.
- Real-time workpiece geometry, machining state, and vibration synchronization are combined with persistent storage for replay, traceability, and future synthetic datasets.
II. PROPOSED PROCESS DIGITAL TWIN CONCEPT
The proposed CPMT architecture combines physical and digital machine-workpiece twins with communication, synchronization, and persistent repository services. A cloud-hosted Unity WebGL client provides browser-based access to monitoring and control.
- The framework includes a physical CNC machine and workpiece, corresponding DTs, a communication and synchronization layer, and a Part DT Repository.
- A cloud server hosts the DT runtime and provides location-independent access to CNC control and monitoring.
- The Unity WebGL application supports desktop and immersive modes through WebXR-compatible devices.
- The cloud server provides communication-broker and part-repository services for location- and device-independent CPMT monitoring and control.
- WebRTC displays live video from the physical CNC machine as a texture overlaid on the workpiece surface in the Unity WebGL environment.
B. Physical Twin
The physical twin is a 3-axis Haas Mini Mill with ball-end and flat-end tools, while the process twin persistently updates a voxel workpiece representation from material-removal data. The architecture combines this representation with sensor and communication components for visualization and robust updates.
- B. Physical Twin: The physical twin is a 3-axis Haas Mini Mill equipped with a 12 mm ball-end mill and a 10 mm flat-end mill.
- C. CNC Machine Digital Twin: The CNC machine DT represents the active tool geometry through tool envelopes used to register material removal.
- D. Machining Process Digital Twin: The machining process DT persistently updates a voxel-based workpiece state estimate driven by real-time machine data.
- D. Machining Process Digital Twin: The stock is represented as a uniform voxel field over its axis-aligned bounding box, with chunks composed of voxels.
- D. Machining Process Digital Twin: The architecture includes a CNC with MTConnect, an acceleration sensor, a web camera, an edge PC, a cloud server, and a user device.
- D. Machining Process Digital Twin: Continuous voxel densities support marching-cubes surface extraction with sub-voxel accuracy from a coarse grid.
- D. Machining Process Digital Twin: Material removal uses the minimum accumulated signed distance to the swept tool envelope, reducing visible seams across overlapping cuts.
- D. Machining Process Digital Twin: Voxel representation is preferred over dexel schemes for generality and robust, accurate updates during rotational cuts, despite higher memory costs.
E. Communication and Synchronization
The framework uses MQTT-based communication to synchronize CNC operational and vibration data with the digital twin while supporting both machine monitoring and teleoperation.
- Capabilities: The system therefore supports both monitoring and teleoperation of the CNC machine.Operational data flows to the DT, while command messages flow back through a separate channel.
- Data communication: MTConnect supplies CNC operational data to the digital twin through an MQTT-based communication layer.A Python edge application collects, parses, time-synchronizes, and publishes the data.
- Data synchronization: Vibration measurements are collected in the background and packaged with timestamps alongside machining-state data.The process DT message includes dominant frequency and vibration amplitude values.
- Bidirectional control: The forward machine-to-DT path uses MTConnect, while the return DT-to-machine path uses TCP commands through a Python command bridge.The Unity client publishes JSON commands over MQTT, which the bridge translates into CNC actions.
F. Part Digital Twin Repository
The Part DT Repository persistently stores synchronized machining, geometry, tool, parameter, vibration, and timing information for each manufactured workpiece.
- Repository architecture: The Part DT Repository uses cloud-hosted MariaDB to centrally manage part-specific digital-twin information.Authorized users can access the stored information remotely.
- Stored information: Each machining operation records CNC data, workpiece geometry, active tool information, machining parameters, vibration measurements, and timestamps.These records are associated with the corresponding part identifier and represent the workpiece’s manufacturing history.
G. Cybersecurity
The deployed framework restricts network access, encrypts communications, and uses automated vulnerability scans to protect CNC and cloud infrastructure.
- Network controls: The edge PC permits only secure MQTT access on port 8883, preventing external users from issuing low-level CNC commands through other ports.The firewall is configured for encrypted and authenticated MQTT communications.
- Encrypted services: The cloud server limits access to MQTT, HTTPS, and SSH services, while MQTT and HTTPS connections require CA-certificate-backed encryption.Both Apache and Mosquitto accept exclusively encrypted client connections.
- Ongoing maintenance: Periodic vulnerability scans identify outdated software and firmware on the cloud server and laboratory edge PC.Mitigation includes upgrading, reverting to a secure version, or disabling affected components.
III. RESULTS AND DISCUSSION
The results demonstrate a synchronized machine-and-workpiece digital twin that updates the evolving stock during machining while retaining interactive performance, geometric accuracy, and process-history capabilities.
- Synchronized process DT: The hierarchical framework synchronizes CNC machine state, evolving workpiece geometry, and vibration information throughout the machining cycle.The voxel-based stock updates from measured tool position and active tool geometry received from the CNC.
- Interactive performance: 20 Hz machining-state updates and 110.4 frames/s average visualization performance were achieved at the baseline configuration.The baseline process DT used a 160 × 224 × 64 voxel grid, with a median frame time of 8 ms and a 1% low frame rate of 49.1 frames/s.
- Interactive performance: Chunk size strongly affected interactive performance: increasing chunks from 4 to 16 voxels per axis raised average frame rate from 39.0 to 110.4 frames/s and reduced frame hitches from 811 to 6.A further increase to 32 voxels per axis reduced performance to 81.4 frames/s.
- Interactive performance: The distribution of frame times revealed 29 times more hitches for 32-voxel chunks than for 16-voxel chunks despite identical median frame times.Surface reconstruction costs concentrate in frames where the tool crosses a chunk boundary.
- Interactive performance: Voxel resolution also bounded performance, with the coarse configuration reaching 120.0 frames/s and the high configuration falling to 2.3 frames/s during cutting.The high-resolution configuration had a median cutting frame time of 422 ms, indicating surface reconstruction as the main cost.
- Geometric fidelity: The reconstructed workpiece achieved a mean absolute depth error of 0.16 mm across ten drilled features, with a maximum error of 0.67 mm.Nine of ten features were within the 0.39 mm bound associated with the baseline resolution’s ±1/2-voxel discretization limit.
- Limitations: Geometric fidelity remains dependent on voxel resolution and CNC data-update frequency.The framework’s synchronization fidelity is also limited because periodic MTConnect requests do not capture intermediate tool positions.
- Process history: Persistent process histories support replay, workpiece traceability, and synthetic digital-twin dataset generation for AI training and validation.The repository stores synchronized CNC, geometry, tool, machining, vibration, and timestamp data for retrieval.
B. Current Limitations
The framework is constrained by voxel discretization, discrete MTConnect sampling, and missing workpiece dimensions. These limitations bound geometric fidelity, trajectory representation, and calibration automation.
- A single scalar per voxel can miss cuts thinner than one voxel, although such features do not arise at the current scale.A two-parameter frame-crossing representation could recover these features at additional memory cost.
- The reconstruction error is dominated by voxel size rather than the synchronization pipeline.
- 20 Hz HTTP sampling omits intermediate tool positions, limiting trajectory fidelity for operations shorter than the sampling interval.Higher rates must balance voxel reconstruction time against transport capacity and end-to-end latency.
- The simulation omits workpiece dimensions, so calibration currently requires manual adjustment.
C. Future Work
Future work will improve trajectory reconstruction, expand AI-assisted process monitoring, and use persistent machining data to support synthetic-data development.
- Integrating the NC program with AI-based trajectory estimation will reconstruct intermediate tool positions missing from MTConnect measurements.The goal is more continuous synchronization of process measurements, workpiece evolution, and machine-tool states.
- AI applications will target chatter detection, machining-parameter optimization, and tool-wear estimation.
- The Part DT Repository can generate synthetic machining datasets for training and validating models for monitoring, anomaly detection, quality prediction, tool-wear estimation, and adaptive optimization.
IV. CONCLUSIONS
The framework unifies machine-tool and machining-process digital twins with synchronized operational, geometric, vibration, and persistent part data. It supports replay and traceability while establishing a basis for industrial validation and AI-driven optimization.
- The framework integrates real-time CNC data, voxel-based workpiece evolution, synchronized vibration measurements, and a cloud-based Part DT Repository.
- Synchronized visualization and persistent process information support machining-process replay and workpiece traceability.
- Future evaluation will validate industrial machining scenarios and add cutting-force, surface-quality, and chip-formation sensing.
- Future work will investigate AI-driven process optimization using recorded and synthetically generated datasets from the Part DT Repository.