Accurate prediction of axis and spindle current signals is essential for model-based CNC process monitoring. Although representative datasets are essential, high product diversity, short life cycles, and varying process conditions lead to incrementally collected datasets with imbalanced or shifted distributions. Therefore, this paper proposes Physics-informed Dataset Optimization (PiDO), a data-centric PiML approach that enriches training datasets with physics-consistent samples before model retraining. PiDO uses locally parameterized physical equations (PEs) for axis- and spindle-current dynamics, derived from an energy-balance-based current model, to generate samples in a reduced discretized feature space. Physical consistency is constrained by local PE parameterization, feature-space bounds, NC-code and machine-kinematic information, and statistically observed ranges of unresolved variables. The approach is evaluated on 11 impeller geometries manufactured from POM-C on two three-axis CNC milling machines (DMC 60H and CMX 600V), resulting in 22 validation datasets, using real-data-only training as the baseline. Evaluation is based on RMSE, R2, significance tests, and dataset-balance measures. Compared with a global PE, the proposed local PE parameterization reduced RMSE by 43.0–74.8% and increased R2 from 0.072–0.448 to 0.821–0.947. In the downstream prediction task, PiDO improved dataset balance and achieved significant positive RMSE effects in several validation scenarios, particularly in sparse data regimes, without significant deterioration compared with the real-data-only baseline.
@article{IEEEAccess26_Stroebel-physics-informed,
title = {{Physics-Informed Dataset Optimization for CNC Axis-Current Prediction}},
author = {Robin Str\"obel and Kai Niklas Hofmann and Hafez Kader and Jan Baumg\"artner and Alexander Puchta and Benjamin Noack and J\"urgen Fleischer},
doi = {10.1109/ACCESS.2026.3718352},
journal = {IEEE Access},
month = jul,
year = {2026}
}
Robin Ströbel, Aaron Büttner, Hafez Kader, Alexander Puchta, Benjamin Noack, Jürgen Fleischer
Density-Based Dataset Balancing for Incremental Retraining in CNC Axis and Spindle Current Prediction IEEE Access, June, 2026.
Accurate prediction of the axis and main spindle current is essential for reliable model-based process monitoring in Computer Numerical Control (CNC) machining. However, modern manufacturing is characterized by a high level of product variety, short life cycles, and frequently changing process conditions. This results in unbalanced and constantly changing data distributions, which present a challenge to model training. Traditional data collection strategies generate large and redundant datasets that require substantial computational and storage resources. This work introduces a density-based approach to dataset balancing for incremental retraining that enables efficient and robust model training under evolving operating conditions. The proposed approach operates at the data level via a density-controlled replay memory, rather than modifying model parameters. A reduced, interpretable feature representation is derived to represent the underlying data structure, providing a basis for systematic discretization. Based on this representation, the adaptive memory mechanism prioritizes samples according to their spatial density, selectively removing redundant information, and ensuring adequate coverage of regions critical for model performance. The proposed method has been evaluated across multiple machines, machining parameters, and geometries. The results show that the balancing strategy yields significantly better results than static training while reducing the required data volume by over 40% compared to incremental data expansion. Consequently, it facilitates the development of data-efficient, scalable and robust training strategies for intelligent machining systems, while also supporting improved adaptability in the monitoring of flexible production environments.
@article{IEEEAccess26_Stroebel,
title = {{Density-Based Dataset Balancing for Incremental Retraining in CNC Axis and Spindle Current Prediction}},
author = {Robin Str\"obel and Aaron B\"uttner and Hafez Kader and Alexander Puchta and Benjamin Noack and J\"urgen Fleischer},
doi = {10.1109/ACCESS.2026.3708218},
journal = {IEEE Access},
month = jun,
year = {2026}
}
Robin Ströbel, Maximilian Noah Kuck, Florian Oexle, Hafez Kader, Alexander Puchta, Benjamin Noack, Jürgen Fleischer
A Multimodal Dataset for Process Monitoring and Anomaly Detection in Industrial CNC Milling Data in Brief, October, 2025.
During the fourth industrial revolution, agile production methods have gained increasing importance to meet the growing demand for product individualization. Conventional process monitoring systems, which predominantly rely on static, statistically based approaches, are insufficient for the requirements of flexible manufacturing environments. Although numerous research initiatives have proposed machine learning (ML) based solutions for agile process monitoring, widespread adoption in industrial practice has not yet been achieved. This is partly due to a lack of system comparability and insufficient validation under realistic production conditions. To address this, a comprehensive dataset was recorded on a DMC 60 H three-axis milling machine by Deckel Maho. The dataset comprises multiple signals from a Siemens SINUMERIK 840D controller, recorded at 500 Hz via a Siemens SINUMERIK Edge. These were synchronized with force and acceleration data (sampled at 10 kHz) captured via a force measurement platform and acceleration sensors. A total of 32 experiments were conducted (15 with 8 distinct anomaly types), resulting in nearly 8 million data points per signal and six hours of process data. Key features include:•Realistic workpiece geometries (thermoforming molds, injection molds, pump impellers) representing diverse milling scenarios.•Multiple anomalies (e.g., tool wear, chatter, material defects) to enable targeted validation.•Full reproducibility through provided NC codes, CAD models, and raw/processed data formats (.json, .mat, .csv, .stp, .nc).The dataset is intended to serve as a benchmark for industrial stakeholders to evaluate monitoring systems, as well as providing researchers with a secondary resource for benchmarking and optimizing ML-based approaches. By fostering comparability, it aims to bridge the gap between theoretical frameworks and industrial application.
@article{DataInBrief25_Stroebel,
title = {{A Multimodal Dataset for Process Monitoring and Anomaly Detection in Industrial CNC Milling}},
author = {Robin Str\"obel and Maximilian Noah Kuck and Florian Oexle and Hafez Kader and Alexander Puchta and Benjamin Noack and J\"urgen Fleischer},
doi = {10.1016/j.dib.2025.112207},
journal = {Data in Brief},
month = oct,
year = {2025}
}
Robin Ströbel, Samuel Deucker, Hanlin Zhou, Hafez Kader, Alexander Puchta, Benjamin Noack, Jürgen Fleischer
Hybrid Machine Learning for CNC Process Monitoring IEEE Access, vol. 13, pp. 91875–91888, May, 2025.
The transition to highly customized, one-off production in modern manufacturing necessitates sophisticated process monitoring to reduce waste, minimise downtime, and alleviate operator burden. Computer Numerically Controlled (CNC) axes represent a fundamental component of automated manufacturing and offer a universal and accessible monitoring option through power supply data. By accurately predicting reference signals and comparing them with real-time measurements, deviations can be used for effective model-based process monitoring and anomaly detection. This study explores the efficacy of hybrid machine learning (ML) models in predicting reference signals for CNC axes using features derived from a physical model. Additionally, relevant but difficult-to-measure features such as process forces and the material removal rate (MMR) were made accessible through soft sensors. Various ML models were evaluated, including tree-based models (e.g. random forest (RF) and gradient boosting (GB)) and deep learning (DL) models (e.g. feed-forward neural networks (FNN), long short-term memory (LSTM), and transformers-based models (TF)). Feature importance analysis was performed, identifying velocity, acceleration, process forces, spindle torque, and MMR as crucial predictors that influence model performance. Key results indicate that tree-based models, specifically RF and GB, consistently delivered the highest accuracy, achieving R2 up to 0.98 for translatory axes and approximately 0.89 for the main spindle. These models demonstrated robustness, outperforming deep learning approaches, particularly when trained on smaller datasets. Although DL models improved with larger data volumes, their performance remained inferior compared to tree-based methods. The study underscores the potential of the integration of physical knowledge into hybrid ML models to enhance model-based process monitoring.
@article{IEEEAccess25_Stroebel,
title = {{Hybrid Machine Learning for CNC Process Monitoring}},
author = {Robin Str\"obel and Samuel Deucker and Hanlin Zhou and Hafez Kader and Alexander Puchta and Benjamin Noack and J\"urgen Fleischer},
doi = {10.1109/ACCESS.2025.3573400},
journal = {IEEE Access},
month = may,
pages = {91875–91888},
volume = {13},
year = {2025}
}
Robin Ströbel, Hafez Kader, Louisa Hutt, Hanlin Zhou, Marcus Mau, Alexander Puchta, Benjamin Noack, Jürgen Fleischer
Intelligente Prozessüberwachung für die flexible Produktion: Integration von Physics-Informed Machine Learning und Active Learning Zeitschrift für wirtschaftlichen Fabrikbetrieb, vol. 120, no. s1, pp. 224–231, Walter de Gruyter GmbH, March, 2025.
As products and their production become more personalized and variant rich, traditional approaches to process monitoring are reaching their limits. These are typically based on static data sets or recurring process patterns, which can lead to inaccurate predictions and increased false alarms in agile production environments. This article presents a concept for flexibilising the process monitoring of discrete production processes based on the combination of Physicsinformed Machine Learning (PIML) and Active Learning (AL). In agile production environments, this allows not only the detection of anomalies, but also the automatic update of the monitoring model in case of false alarms. As a result, the monitoring system remains accurate under variable production conditions, reducing the number of false alarms and contributing to improved Overall Equipment Effectiveness (OEE).
@article{ZWF25_Stroebel,
title = {{Intelligente Prozess\"uberwachung f\"ur die flexible Produktion: Integration von Physics-Informed Machine Learning und Active Learning}},
author = {Robin Str\"obel and Hafez Kader and Louisa Hutt and Hanlin Zhou and Marcus Mau and Alexander Puchta and Benjamin Noack and J\"urgen Fleischer},
doi = {10.1515/zwf-2024-0154},
issn = {0947-0085},
journal = {Zeitschrift f\"ur wirtschaftlichen Fabrikbetrieb},
month = mar,
number = {s1},
pages = {224–231},
publisher = {Walter de Gruyter GmbH},
volume = {120},
year = {2025}
}
Conference Proceedings
Hafez Kader, Benjamin Noack
Genetic Algorithm-Based Feature Selection for CNC Energy Consumption Time Series Prediction (accepted) Proceedings of the 23rd IFAC World Congress (IFAC 2026), Busan, South Korea, August, 2026.
BibTeX
@inproceedings{IFAC26_Kader,
title = {{Genetic Algorithm-Based Feature Selection for CNC Energy Consumption Time Series Prediction (accepted)}},
author = {Hafez Kader and Benjamin Noack},
booktitle = {Proceedings of the 23rd IFAC World Congress (IFAC 2026)},
address = {Busan, South Korea},
month = aug,
year = {2026}
}
Florian Bariszlovich, Hafez Kader, Benjamin Noack
Label Efficiency for Signal Quality Classification of Wearable Respiratory Inductance Plethysmography (accepted) Proceedings of the 39th IEEE International Symposium on Computer Based Medical Systems (CBMS2026), Limassol, Cyprus, June, 2026.
BibTeX
@inproceedings{CMBS26_Bariszlovich,
title = {{Label Efficiency for Signal Quality Classification of Wearable Respiratory Inductance Plethysmography (accepted)}},
author = {Florian Bariszlovich and Hafez Kader and Benjamin Noack},
booktitle = {Proceedings of the 39th IEEE International Symposium on Computer Based Medical Systems (CBMS2026)},
address = {Limassol, Cyprus},
month = jun,
year = {2026}
}
Hafez Kader, Steven C. Marcrum, Milena Engelke, Niklas K. Edvall, Berthold Langguth, Birgit Mazurek, Jose Antonio Lopez Escamez, Dimitros Kikidis, Rilana Cima, Patrick Neff, Winfried Schlee, Christopher R. Cederroth, Benjamin Noack, Myra Spiliopoulou, Stefan Schoisswohl
Classifying Residual Inhibition in the Context of Tinnitus: An Interpretable Machine Learning Approach Proceedings of the 38th IEEE International Symposium on Computer Based Medical Systems (CBMS2025), pp. 605–611, Madrid, Spain, June, 2025.
Residual inhibition (RI) is a phenomenon
observed in many tinnitus patients, where tinnitus
remains temporarily suppressed for a short duration—
typically less than a minute—after the cessation
of an appropriate masking stimulus. Despite decades
of clinical interest in RI, machine learning (ML)-based,
feature-driven classification approaches remain scarce.
In this study, we investigate the potential of ML
models to classify RI by developing a dedicated data
analysis pipeline. Given the heterogeneous nature of
the features—including numerical, binary, and ordinal
variables—we apply feature importance techniques tailored
for mixed-type data to ensure a comprehensive
evaluation and improve interpretability. Our results
demonstrate a clear separation between RI classes,
highlighting the relevance of specific clinical and audiological
factors in distinguishing them.
Building on this, we assess the predictive power of
RI classifications with high confidence within a supervised
learning framework to determine their relevance
for treatment outcome prediction. While our findings
confirm that RI can be effectively classified, they also
suggest that RI alone is not sufficient to serve as a
reliable predictor for treatment outcomes.
@inproceedings{CMBS25_Kader,
title = {{Classifying Residual Inhibition in the Context of Tinnitus: An Interpretable Machine Learning Approach}},
author = {Hafez Kader and Steven C. Marcrum and Milena Engelke and Niklas K. Edvall and Berthold Langguth and Birgit Mazurek and Jose Antonio Lopez Escamez and Dimitros Kikidis and Rilana Cima and Patrick Neff and Winfried Schlee and Christopher R. Cederroth and Benjamin Noack and Myra Spiliopoulou and Stefan Schoisswohl},
booktitle = {Proceedings of the 38th IEEE International Symposium on Computer Based Medical Systems (CBMS2025)},
address = {Madrid, Spain},
doi = {10.1109/CBMS65348.2025.00127},
month = jun,
pages = {605–611},
year = {2025}
}
Hafez Kader, Robin Ströbel, Alexander Puchta, Jürgen Fleischer, Benjamin Noack, Myra Spiliopoulou
Finding Predictive Features for Energy Consumption of
CNC Machines Proceedings of the Berlin Workshop on Artificial Intelligence for Engineering Applications (AI4EA24), Berlin, Germany, November, 2024.
abstract
BibTeX
With rising energy costs and a growing emphasis on sustainable and efficient production, predicting the energy consumption of CNC machines
has become increasingly important. Accurate predictions can lead to significant energy savings, better planning, more informed decision-making, and alignment with smart manufacturing and Industry 4.0 initiatives.
Extensive research has been conducted in this area, utilizing both physical and analytical models, as well as expert knowledge from experiments. More recently, machine learning models have also been employed using a wide range of input features. In this paper, we examine the energy consumption of CNC machines by analyzing various features explored in different studies. We propose a method that ranks these features based on their predictive power, then groups the rankings to highlight a small subset of the most predictive features. Furthermore, we assess the stability of the predictive power of these features over time, allowing us to not only rank them by their predictive strength but also evaluate their long-term stability. Our findings indicate that only a few features are highly predictive, and their predictive
power remains consistent over time.
@inproceedings{AI4EA24_Kader,
title = {{Finding Predictive Features for Energy Consumption of
CNC Machines}},
author = {Hafez Kader and Robin Str\"obel and Alexander Puchta and J\"urgen Fleischer and Benjamin Noack and Myra Spiliopoulou},
booktitle = {Proceedings of the Berlin Workshop on Artificial Intelligence for Engineering Applications (AI4EA24)},
address = {Berlin, Germany},
isbn = {978-3-942709-34-7},
month = nov,
year = {2024}
}
Hafez Kader, Robin Ströbel, Alexander Puchta, Jürgen Fleischer, Benjamin Noack, Myra Spiliopoulou
Feature Ranking for the Prediction of Energy Consumption on CNC Machining Processes Proceedings of the 2024 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI 2024), Pilsen, Czech Republic, September, 2024.
Energy consumption is a critical factor that negatively impacts the environment. Sustainable production is essential for addressing the climate crisis, as low-emission manufacturing can both reduce costs and minimize environmental impact. Energy-efficient CNC machine tools significantly contribute to achieving ambitious environmental objectives. In recent years, numerous studies have focused on low-energy consumption production, analyzing factors that contribute to sustainable manufacturing. When using the analytical or empirical model, factors and corrections might be omitted. With advancements in machine learning and the increasing availability of large datasets, models are being developed to predict energy consumption with high accuracy. However, these models often overlook the importance of features that contribute to a transparent prediction process and their influence on the results. In our paper, an LSTM model is initially utilized to predict the energy consumption of CNC machines. Following this, a method is devised to rank the features based on their predictive power, considering temporal variations. We show that some of the features ranked in the top positions agree with independent literature findings, while others are new and demand further investigation.
@inproceedings{MFI24_Kader,
title = {{Feature Ranking for the Prediction of Energy Consumption on CNC Machining Processes}},
author = {Hafez Kader and Robin Str\"obel and Alexander Puchta and J\"urgen Fleischer and Benjamin Noack and Myra Spiliopoulou},
booktitle = {Proceedings of the 2024 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI 2024)},
address = {Pilsen, Czech Republic},
doi = {10.1109/MFI62651.2024.10705783},
month = sep,
year = {2024}
}
Last Modification: 14.05.2025 - Contact Person: Webmaster
