Paper deep dive
Federated Learning for Distributed CNC Tool Wear Prediction
Afsana Khan, Morris Stallmann, Marcin Pietrasik, Charis Kouzinopoulos, Anna Wilbik
Intelligence
Status: succeeded | Model: Gemma-4-26B-A4B | Prompt: intel-v1 | Confidence: 91%
Last extracted: 8/16/2026, 3:23:28 AM
Summary
This paper investigates the application of Federated Learning (FL) for distributed CNC tool wear prediction using the MATWI dataset. It compares FL performance against centralized and local client baselines using sensor and image modalities. Results indicate that FL achieves performance close to centralized learning and significantly outperforms local-only training, demonstrating its viability for privacy-preserving collaborative manufacturing environments.
Entities (10)
Relation Signals (7)
Federated Learning → usedfor → CNC Tool Wear Prediction
confidence 95% · This paper investigates federated learning for CNC tool wear prediction.
Federated Learning → outperforms → Local Client Models
confidence 93% · Results show that federated learning ... improves significantly over local client models.
MATWI Dataset → usedin → Federated Learning
confidence 92% · we propose to use the MATWI dataset [3] in combination with two novel FL systems to study this question.
Federated Learning → approachesperformanceof → Centralized Learning
confidence 90% · Results show that federated learning achieves performance close to centralized learning...
FedAvg → usedby → Federated Learning
confidence 90% · Federated learning is implemented with Flower 1.30.0 using FedAvg.
1D-CNN → usedfor → Sensor Modality
confidence 88% · The sensor model is a 1D-CNN.
ResNet50 → usedfor → Image Modality
confidence 88% · The image pipeline uses a ResNet50 regression model trained on cutting edge images.
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:Tool wear prediction is an important task in CNC machining, where accurate monitoring of tool condition supports product quality and process reliability. Machine learning methods have shown potential for this task, but their use in industrial environments is limited by the distributed nature of machining data and by restrictions on data sharing between machines, sites, or organizations. Federated learning offers a suitable framework for this setting by enabling collaborative model training without transferring raw operational data. This paper investigates federated learning for CNC tool wear prediction. Tool trajectories are distributed across simulated clients to represent a federated learning scenario. The federated models are compared against centralized references and local client baselines. Results show that federated learning achieves performance close to centralized learning and improves significantly over local client models. These findings indicate that federated learning can support collaborative tool wear prediction in distributed CNC manufacturing environments.
Tags
Links
- Source: https://arxiv.org/abs/2608.11281v1
- Canonical: https://arxiv.org/abs/2608.11281v1
Trouble viewing inline? Open PDF directly →
Full Text
39,462 characters extracted from source content.
Expand or collapse full text
Federated Learning for Distributed CNC Tool Wear Prediction Afsana Khan ⋆ , Morris Stallmann, Marcin Pietrasik, Charis Kouzinopoulos, and Anna Wilbik Maastricht University, Maastricht, Netherlands a.khan@maastrichtuniversity.nl Abstract. Tool wear prediction is an important task in CNC machining, where accurate monitoring of tool condition supports product quality and process reliability. Machine learning methods have shown potential for this task, but their use in industrial environments is limited by the distributed nature of machining data and by restrictions on data shar- ing between machines, sites, or organizations. Federated learning offers a suitable framework for this setting by enabling collaborative model training without transferring raw operational data. This paper investi- gates federated learning for CNC tool wear prediction. Tool trajectories are distributed across simulated clients to represent a federated learning scenario. The federated models are compared against centralized refer- ences and local client baselines. Results show that federated learning achieves performance close to centralized learning and improves signif- icantly over local client models. These findings indicate that federated learning can support collaborative tool wear prediction in distributed CNC manufacturing environments. Keywords: Federated learning· Tool wear prediction· CNC milling· Distributed systems 1 Introduction Cutting tools are crucial in machining processes such as milling, drilling, or sawing to guarantee product quality. Even though significant effort is put into engineering high-quality tools [11], even the best ones are affected by wear due to abrasion, adhesion, and diffusion during the machining process and eventually break or reach a state where the produced parts become scrap [14]. Therefore, it is important to recognize the wear state to plan a timely replacement of the tool. Several recent studies have shown that machine learning (ML) methods achieve accurate wear state predictions [18], [17]. While training reliable and generalizable ML models often depend on the availability of high-quality and sufficiently diverse data, collecting such a dataset may be challenging for a single ⋆ Corresponding author arXiv:2608.11281v1 [cs.LG] 11 Aug 2026 2A. Khan et al. manufacturer or production cell, but it may become feasible if multiple parties collaborate and combine their data. In a realistic CNC manufacturing environ- ment, machining data may be distributed across machines, production cells, factories, or organizations. Centralizing such data can be costly because sensor recordings and images are large, continuously generated, and tied to operational processes. It can also be undesirable because tool wear data can reveal informa- tion about production conditions, machining strategies, material properties, or process quality. Federated Learning (FL) is a machine learning paradigm that allows collaborators to jointly train a machine learning model without having to share raw data instances and while keeping local data private. As such, it is a natural fit for environments with distributed data and data sharing constraints such as CNC manufacturing, as several recent works have noticed (Section 2). Although these works show promising results, they leave open whether FL actually provides benefits over siloed local client training under the assumption of realistic data processing challenges in distributed systems (Section 2). In this work, we propose to use the MATWI dataset [3] in combination with two novel FL systems to study this question. Each system is composed of modules for data preprocessing to harmonize disparate local datasets, learning algorithms to identify local patterns, and an averaging mechanism to derive the global model (Section 3), and is benchmarked against local client training. The MATWI dataset contains images and measurements of accelerometer and acoustic sensors along with the wear labels of 17 sets of cutting tools (Section 3.1). The data is unprocessed, sufficiently large to simulate a federated setting, contains multiple modalities, and is naturally distributed across different machines. In Section 2, we introduce the federated learning framework and related works that apply it to learn tool wear prediction models. Section 3 formulates the distributed learning problem of this work and describes how to solve it using FL. In Section 4, Section 5 Section 6, the evaluation on the MATWI dataset is described. Limitations of this work and potential future research directions are discussed in Section 7. Lastly, the work is concluded in Section 8. 2 Related Work 2.1 Federated Learning In FL, multiple clients collaboratively train a ML model to utilize distributed data while preserving privacy. Each client holds a local dataset that must remain private, that is, it cannot be shared with any other entity. In centralized FL, the training protocol can be described in terms of five steps [5]: 1. The orchestrator (central server) selects the clients that participate in the next round of training. 2. The central server shares the global model with the selected clients. 3. The clients optimize a local learning objective on their local data and share the local updates with the central server. Federated Learning for Distributed CNC Tool Wear Prediction3 4. The central server aggregates all local updates to a global update and applies it to derive the new global model. 5. The updated global model is shared with the clients. If a convergence crite- rion is met, the training ends. If not, the training process continues from the first step. FL settings differ in how data is partitioned (vertically or horizontally) and how the learning process is orchestrated (centralized or decentralized). Horizon- tally split data is partitioned by sample, and vertically split data by feature. In centralized FL, the model training is orchestrated by a central server, whereas clients communicate directly with each other in the decentralized case. We con- sider centralized, horizontal Federated Learning in this work. Common challenges in Federated Learning stem from the distributed nature of the data and include communication overhead, system heterogeneity, and data heterogeneity [16]. Data heterogeneity, or non-IIDness, in FL occurs when the clients’ local data follow different distributions, which can impact training con- vergence and model performance. Approaches addressing the key challenge of data heterogeneity often apply variants of the standard aggregation algorithm FedAvg (see Section 3), group compatible clients for focused model updates, or apply data augmentation techniques [20]. Despite its challenges, the academic literature demonstrates the applicability of FL to various application domains, including health care [13], financial service security [7], and predictive mainte- nance [15]. 2.2 Tool Wear Prediction with Federated Learning FL has attracted increasing research attention for tool wear prediction in indus- trial environments. Using sensory information, including accelerometers, vibra- tion, or acoustic emission, the authors of [6] formulate tool wear as a forecasting problem. They propose a combination of feature calculation techniques and a federated (Bi)LSTM architecture to estimate the target value and demonstrate promising results on the PHM dataset [8]. However, it remains open whether the federated approach is beneficial compared to local-only training since the work focused on proposing a novel federated method. In [4], a new federated aggrega- tion algorithm addressing client data heterogeneity is introduced and evaluated on the tool wear dataset introduced in [9]. Since the focus of the work is on the aggregation algorithm, no emphasis is put on data processing challenges in distributed environments. In fact, [9] reports that data normalization is applied to raw data, a step that may be hard to replicate in a distributed, privacy- preserving setting. A tool wear prediction approach using images is introduced in [2]. The authors observe that FL models can produce better results than per-client training, thereby motivating its application in tool wear prediction. However, they also acknowledge the limited generality of their results given the small scale of their experimental data. In summary, several recent works demonstrate the applicability of Feder- ated Learning to tool wear prediction in the machining industry. However, these 4A. Khan et al. works either neglect data preprocessing challenges inherent to distributed data settings or do not prove the benefits of training a FL model justify the increased complexity of such an implementation. 3 Methodology This paper formulates tool wear prediction as a distributed learning problem in which industrial data are generated and retained by separate data owners. The proposed setting therefore treats each client as an autonomous industrial site that keeps its raw data local while participating in collaborative model training. The considered setting corresponds to horizontal federated learning. In hori- zontal federated learning, clients share the same feature space and learning task but hold different samples. This matches the tool wear scenario because each client trains the same type of prediction model for a given modality, while the samples are distributed across different tools or sites. Federated learning provides the distributed coordination mechanism. Instead of moving raw data to a central location, the model is sent to the data. A central server coordinates training rounds, while clients perform local computation using their own data. The server never observes raw sensor recordings or images. It only receives model parameters from the clients and aggregates them into a new global model. This design reduces the need for data centralization and supports data locality by construction. The approach is privacy-preserving by design, in that raw operational data remain at the client sites. Let K denote the number of clients. Each client k owns a local dataset D k containing n k samples, and let n = P K k=1 n k be the total number of samples across all clients. For a sample (x,y) ∈ D k , where x is the model input and y is the corresponding tool wear label, let ℓ(θ;x,y) denote the loss for model parameters θ. The local objective for client k is L k (θ) = 1 n k X (x,y)∈D k ℓ(θ;x,y). The global objective is to learn a single model that minimizes the sample- weighted average of the local client objectives: L(θ) = K X k=1 n k n L k (θ), and the target model is θ ∗ = arg min θ L(θ). Training follows a client-server protocol based on synchronous communica- tion rounds. At round t, the server broadcasts the current global parameters θ t to all participating clients. Each client initializes its local model with θ t , performs Federated Learning for Distributed CNC Tool Wear Prediction5 Fig. 1: Federated Learning Architecture for Tool Wear Prediction local optimization on D k , and returns updated parameters θ k t to the server. The server then applies FedAvg aggregation [12]: θ t+1 = K X k=1 n k n θ k t . This aggregation provides clients with more local samples with proportion- ally greater influence while maintaining a single shared global model. From a distributed systems perspective, each round consists of global synchronization, parallel local computation, communication of model updates, and server-side aggregation. The cost of collaboration is therefore expressed through communi- cation rounds and exchanged model parameters rather than through raw data transfer. The distributed training protocol is independent of modality. The same client- server coordination, local training, communication, and aggregation procedure can be applied to different data modalities, provided that the clients use a com- patible model architecture for the selected modality. In this paper, the protocol is instantiated for sensor data and image data. The model architecture differs by modality, but the system-level protocol remains unchanged. 6A. Khan et al. 3.1 The MATWI Dataset The Multimodal Automatic Tool Wear Inspection (MATWI) dataset was intro- duced by De Pauw et al. [3] to support the development of automated tool-wear monitoring methods for CNC milling. It contains data from 17 cutting-tool sets, with each set representing the wear progression of one physical milling insert over its operational lifetime. For each machining cycle, MATWI provides labeled wear observations together with synchronized process-sensor recordings and an image of the cutting edge. The sensor data include cutting-force measurements along three axes, accelerometer signals, and acoustic measurements, while the visual modality consists of images captured after machining to document the condition of the insert. Each sample is associated with quantitative wear labels and wear-type an- notations. The wear categories include flank wear, adhesive wear, and combined flank-and-adhesive wear. In [3], sets 1–13 were used for the main baseline image- based experiments. Sets 14–17 were excluded because they were collected while machining a different workpiece material, which resulted in increased adhesive wear and a visibly different tool appearance. This created a domain shift relative to the earlier sets. Following the same protocol, Sets 1–13 are used for the bench- marking experiments. Each sample is associated with quantitative wear labels and wear-type annotations. The wear categories include flank wear, adhesive wear, and combined flank-and-adhesive wear. The structure of MATWI is suitable for federated learning because each set represents the complete wear trajectory of a single physical tool. In practical manufacturing environments, comparable data may be generated and retained locally by different machines, production cells, or sites rather than centrally pooled. To emulate this setting, tool sets are allocated across federated clients, while all sensor recordings and images associated with a particular tool remain at the same client. Clients collaboratively train a shared model through model updates without exchanging raw machining data. This partitioning also prevents tool-level data leakage and preserves natural heterogeneity across tools, including differences in wear progression and wear type. 3.2 Preprocessing The sensor recordings contain both air-cut intervals and active-cutting intervals. Air cuts occur when the tool follows the machining path without engaging the workpiece and therefore does not remove material. These portions contain lim- ited information about tool condition, whereas active cutting intervals capture the interaction between the tool and the workpiece and contain patterns relevant to wear prediction. To identify the active cutting interval, a level four discrete wavelet decom- position [10] with the Daubechies 4 wavelet [1] is applied to the selected sensor channels. For a sensor signal x[n], the decomposition is represented as DWT(x[n]) = A 4 ,D 4 ,D 3 ,D 2 ,D 1 , Federated Learning for Distributed CNC Tool Wear Prediction7 where A 4 denotes the approximation coefficients and D j denotes the detail co- efficients at level j. The magnitude of the finest scale detail coefficients is used as an activity measure: E[k] =|D 1 [k]|. A baseline is estimated from the first 1000 values of E[k]. Let μ E and σ E denote the mean and standard deviation of this baseline. The normalized activity score is calculated as: z[k] = E[k]− μ E σ E + 10 −9 . After the scores are aligned with the original signal length, samples satisfying z[n] > 10 are considered active. The first and last active samples define the preliminary cutting boundaries. The detected interval is expanded by 5 The procedure is performed independently for the two selected sensor chan- nels. Given detected intervals [s 1 ,e 1 ] and [s 2 ,e 2 ], the final retained interval is defined as: s = max(s 1 ,s 2 ), e = min(e 1 ,e 2 ). When fewer than 1000 active samples are detected for a channel, the complete recording is retained. Only the resulting active cutting segments are used in the experiments (Figure 2). To evaluate the effect of air-cut removal, an ablation experiment was con- ducted for the model trained on sensor data. The model was compared under two conditions: using the complete sensor signals and using only the detected active cutting intervals. As shown in Figure 3, air-cut removal reduces the total MAE from 26.24 μm to 16.40 μm. The improvement is mainly driven by flank wear, where the MAE decreases from 26.02 μm to 11.36 μm. Since flank wear is the dominant wear type in the test set, this improvement has a strong effect on the total result. Adhesion and combined flank wear & adhesion do not show the same improvement, which may be due to their smaller number of samples present in the dataset and the fact that adhesion is a surface material sticking phenomenon that is less consistently reflected in sensor signals. Overall, the ab- lation supports the use of air-cut removal because it focuses the sensor model on the active cutting region and improves the total sensor modality performance. 4 Experimental Setup A centralized baseline is used as the reference setting. In this setting, all training data are pooled at one location, and a single model is trained. Following the MATWI dataset paper protocol [3], Sets 1, 2, 5, 7, 8, 10, and 11 are used for training, Sets 3, 6, and 12 for validation, and Sets 4, 9, and 13 for testing. The split is performed at the tool set level, so no cuts from the same physical tool appear in more than one split. 8A. Khan et al. Fig. 2: Sensor recordings for machine cycle on Set 1 before and after air cut removal. The left column shows the complete machining cycle, with the retained active cutting interval highlighted. The right column shows the corresponding retained signals for the force, accelerometer, and acoustic channels. 4.1 Prediction Task The task is supervised regression, where the model predicts the continuous tool wear value for each machining cycle. Two pipelines are evaluated: a sensor-based pipeline using process signals recorded during machining, and an image-based pipeline using cutting edge images captured after machining. Both pipelines are evaluated under centralized, local, and federated training. All models are trained with Huber loss, which is used to reduce sensitivity to occasional large prediction errors while retaining smooth regression behaviour for small errors: L δ (r) = ( 1 2 r 2 ,|r|≤ δ, δ |r|− 1 2 δ , |r| > δ. where r = y− ˆy, and δ = 1.0. 4.2 Sensor Pipeline The full active cutting signal is not used as a single model input because it is long and contains time-varying patterns across the machining cycle. Yang Federated Learning for Distributed CNC Tool Wear Prediction9 Fig. 3: Effect of air-cut removal on performance of the model trained on sensor data. The model architecture is a 1D-CNN, and the reported metric is MAE in μm for per wear type and in total. et al. [19] used local segmentation of tool wear sensor signals to capture local signal characteristics before modeling the overall wear condition. Following the same motivation, each active cutting signal is divided into sliding windows. The window size is 2048 samples, and the stride is 1024 samples. Each window inherits the cut-level wear label. During inference, window predictions from the same cut are averaged to obtain one cut-level prediction. The sensor model is a 1D-CNN. Training settings are Adam optimizer, learn- ing rate 10 −4 , batch size 32, and Huber loss with δ = 1.0. In addition to the windowed sensor input, the model receives a context vector formed by concate- nating the previous wear value with a one-hot encoding of the wear type. Since the wear type has three categories, flank wear, adhesive wear, and combined flank and adhesive wear, the resulting context vector has dimension 4. Early stopping with patience 10 is used for centralized, local, and federated training. In federated sensor training, the maximum number of communication rounds is 100, with 5 local epochs per round. 4.3 Image Pipeline The image pipeline uses a ResNet50 regression model trained on cutting edge images. Training settings follow the original MATWI baseline: Adam optimizer, learning rate 3× 10 −4 , batch size 16, and Huber loss with δ = 1.0. Images are cropped using the per set crop coordinates from the dataset metadata, resized to 224× 224 pixels using Lanczos resampling, and normalized with ImageNet statistics: mean [0.485, 0.456, 0.406], standard deviation [0.229, 0.224, 0.225]. Early stopping with patience 10 is used for centralized, local, and federated image training. In federated image training, the maximum number of communi- cation rounds is 30, with 5 local epochs per round. 10A. Khan et al. 4.4 Federated Training and Evaluation To simulate a distributed industrial environment, the training sets are parti- tioned across three clients, each representing an independent manufacturing site. Client A receives Sets 1, 5, and 7; Client B receives Sets 2 and 10; and Client C receives Sets 8 and 11. The validation and test sets remain unchanged across all experiments. In addition to centralized training, a local client baseline is evalu- ated, where each client trains a separate model using only its own assigned data. The federated setting uses the same client partitions, but the clients collabora- tively train a shared model by exchanging model updates. The comparison is designed to assess whether federated training improves over the local baseline while approaching the centralized baseline. Federated learning is implemented with Flower 1.30.0 using FedAvg. In each round, all three clients participate. The server sends the global model to the clients, clients train locally, and the server aggregates the returned parameters using a sample- weighted average. Final evaluation is performed on test Sets 4, 9, and 13. Results are reported using MAE and the standard deviation of absolute errors per wear category, following the original MATWI evaluation format. All models are implemented in PyTorch and trained on the Snellius HPC cluster using a single NVIDIA H100 GPU. A fixed seed of 777 is used for Python, NumPy, PyTorch, CUDA, and DataLoader shuffling. 5 Results This section presents the performance of the centralized, federated, and local client models on the held-out test sets, Sets 4, 9, and 13. Performance is reported using mean absolute error (MAE) measured in micrometers. The centralized model represents the pooled-data reference, the local client models represent isolated training without collaboration, and the federated model represents col- laborative training without raw data sharing. 5.1 Sensor Modality Figure 4 shows the performance of the models trained on the sensor modality. For flank wear, which is the most represented and practically important wear type in this evaluation, the two models show comparable performance. For adhesion, both models show larger deviations from the reference trend, with the federated model having a higher error than the centralized model. For combined flank wear and adhesion, the federated model performs slightly better than the centralized model according to the MAE reported in Table 1. The local client models provide the isolated training baselines for the sensor modality. Compared with these baselines (Table 1), the federated model achieves lower total error than Clients B and C and remains close to Client A. This shows that federated training provides a competitive distributed model while allowing the clients to keep their raw operational data local. Federated Learning for Distributed CNC Tool Wear Prediction11 (a) Centralized model(b) Federated model Fig. 4: Performance of centralised and federated models (Sensor Modality) 5.2 Image Modality (a) Centralized model(b) Federated model Fig. 5: Performance of centralised and federated models (Image Modality) Figure 5 shows the performance of the models trained on the image modality. The federated model has performance close to the centralized model as well, indicating that the distributed training protocol produces results close to the pooled-data reference for this modality. From Table 2 we observe that the federated model remains close to the cen- tralized reference in total MAE. The flank wear type shows similar performance between the two models. The federated model achieves lower error for adhesion, while the centralized model performs better for combined flank wear and adhe- sion. The local client models provide the isolated training baselines for the image modality (Table 2). The federated model achieves lower total error than all three 12A. Khan et al. local client models. This shows that collaborative learning across clients is more effective than training each client independently. This improvement is achieved without exchanging raw operational data between clients or with the server. 6 Discussion The results show that federated learning provides a practical compromise be- tween centralized and local client models for CNC tool wear prediction. The cen- tralized model represents the pooled-data reference, while the federated model allows clients to contribute to a shared model through parameter updates. The comparison with local client models shows whether collaboration through feder- ated learning is more effective than keeping each client model isolated. For the sensor modality, the federated model remains close to the centralized model. This is mainly driven by the flank wear results, where the two models have very similar errors. Since flank wear is the largest wear type in the test set, similar performance on this wear type has a strong effect on the overall result. For combined flank wear and adhesion, the federated model slightly improves over the centralized model, while for adhesion the federated model has a higher error. This difference is expected because adhesion can be more difficult to infer from process signals alone. Sensor signals capture indirect effects of wear during cutting, whereas adhesion refers to material sticking to the tool surface and may not always produce a clear or consistent signal pattern. For the image modality, the federated model also remains close to the cen- tralized model. The difference between the two models is small, showing that the federated protocol also works effectively when the input modality is visual. The federated model performs better for adhesion, while the centralized model per- forms better for flank wear and combined flank wear and adhesion. The stronger adhesion result in the image modality is plausible because adhesion is a visual surface phenomenon, as it occurs when workpiece material sticks to the cut- ting edge, making it easier to observe directly in images than through sensor measurements. The comparison with local client models shows the benefit of federated col- laboration. A local client model is built only from the data available at one client, so it reflects a narrower subset of the available tool trajectories. In the image modality, the federated model achieves lower total error than all three local client models. In the sensor modality, the federated model achieves lower total error than Clients B and C and remains close to Client A. These results show that federated learning provides a stronger alternative to isolated local models by allowing information from multiple clients to be combined through model aggregation. This improvement is achieved without exchanging raw sensor recordings or images between clients or with the server. The remaining gap between federated and centralized models reflects the con- straints of federated optimization. The centralized model has access to pooled data and can optimize directly over the combined training set. The federated model, in contrast, depends on client-side updates followed by server-side aggre- Federated Learning for Distributed CNC Tool Wear Prediction13 Table 1: Tool wear prediction performance of centralized, federated, and local client models using sensor modality. Wear Type Centralized Model MAE ( μ m) Federated Model MAE ( μ m) Client A Model textbfMAE ( μ m) Client B Model textbfMAE ( μ m) Client C Model textbfMAE ( μ m) Adhesion 58 . 36 ± 55 . 74 70 . 75 ± 52 . 05 56 . 26 ± 63 . 40 57 . 62 ± 40 . 14 62 . 33 ± 55 . 78 Flank Wear 11 . 36 ± 12 . 04 12 . 50 ± 12 . 58 13 . 67 ± 13 . 39 27 . 56 ± 27 . 84 16 . 63 ± 20 . 84 Flank Wear & Adhesion 23 . 66 ± 18 . 37 22 . 96 ± 9 . 70 30 . 27 ± 19 . 03 49 . 45 ± 25 . 64 62 . 52 ± 26 . 32 Total 16 . 40 ± 22 . 24 17 . 90 ± 22 . 32 19 . 21 ± 23 . 95 33 . 28 ± 30 . 21 27 . 58 ± 32 . 02 Table 2: Tool wear prediction performance of centralized, federated, and local client models using image modality. Wear Type Centralized Model MAE ( μ m) Federated Model MAE ( μ m) Client A Model textbfMAE ( μ m) Client B Model textbfMAE ( μ m) Client C Model textbfMAE ( μ m) Adhesion 35 . 19 ± 40 . 43 26 . 73 ± 44 . 21 39 . 17 ± 52 . 70 36 . 39 ± 40 . 06 39 . 60 ± 36 . 82 Flank Wear 13 . 64 ± 11 . 20 14 . 05 ± 13 . 98 28 . 15 ± 21 . 73 31 . 43 ± 23 . 79 40 . 65 ± 20 . 41 Flank Wear & Adhesion 31 . 08 ± 21 . 42 36 . 00 ± 31 . 38 34 . 58 ± 31 . 82 53 . 61 ± 30 . 31 108 . 62 ± 169 . 62 Total 18 . 06 ± 18 . 42 18 . 73 ± 22 . 73 29 . 97 ± 26 . 83 35 . 68 ± 27 . 62 52 . 69 ± 78 . 76 14A. Khan et al. gation. Differences in client data size, wear progression, and wear type composi- tion can influence the updates received by the server. As a result, the aggregated model may not exactly match the centralized reference, even when all clients par- ticipate in each communication round. The wear type results also show that performance is not equally stable across all labels. Flank wear forms the largest part of the train and test set and gives the most consistent results across centralized and federated models. Adhesion and combined flank wear and adhesion contain fewer samples, so their MAE values are more affected by individual prediction errors. This is visible in both modalities, where the smaller wear types show larger variation than flank wear. Therefore, differences for adhesion and combined wear should be interpreted together with the number of available test samples. From a federated systems perspective, the results support the use of data- local collaborative learning for industrial monitoring. Each client performs com- putation locally, while the server coordinates communication rounds and aggre- gates model parameters. The server does not receive raw operational data, and the clients do not need to exchange datasets with one another. This makes the approach relevant for industrial environments where machining data may be dis- tributed across machines, production cells, factories, or organizations, and where direct data pooling may be costly, restricted, or undesirable. 7 Limitations and Future Directions The federated learning setup supports privacy by design because raw opera- tional data remain at the client sites and are not transferred to the server or other clients. Stronger privacy protection would require additional mechanisms such as secure aggregation or differential privacy, since model updates may still contain information about local data. The evaluation is based on a controlled fed- erated simulation using a benchmark dataset. In larger deployments with more clients, the data distribution may become more heterogeneous, especially when clients differ in machine type, cutting conditions, workpiece materials, tooling, data volume, or wear progression. Such non-IID behaviour can make federated optimization more difficult and may affect the stability of FedAvg. In these cases, alternative aggregation and optimization strategies, such as FedProx or personalized federated learning, may become useful for handling stronger client heterogeneity. The results also show that the two modalities contribute differ- ently across wear types. The sensor modality gives strong performance for flank wear, where process signals capture the cutting behaviour associated with pro- gressive tool degradation. The image modality is more effective for adhesion, which is a visual surface phenomenon caused by material sticking to the cutting edge. This suggests that sensor and image data provide complementary infor- mation. A promising direction is therefore multimodal federated learning, where both modalities are used jointly within the federated framework to combine their advantages while still keeping raw operational data local to each client. Federated Learning for Distributed CNC Tool Wear Prediction15 8 Conclusion In this work, we study the applicability of federated learning to tool wear predic- tion with real-world, naturally distributed CNC milling data containing multiple modalities in scenarios where data cannot be centralized. The problem is formu- lated to fit into the FL paradigm and solved by preprocessing to harmonize local data distributions and by a federated training protocol using FedAvg. Through experimentation on the MATWI dataset, it is demonstrated that the federated method outperforms local-only models, therefore justifying the increased com- plexity of federated learning compared to local-only machine learning. Future research can focus on improving prediction accuracy through the application of more involved aggregation methods or by applying multimodal federated learn- ing to learn a single model from multiple, complementary modalities. References 1. Daubechies, I.: The wavelet transform, time-frequency localization and signal anal- ysis. IEEE transactions on information theory 36(5), 961–1005 (1990) 2. de Melo Rosa, G.L., Gwose, A., Ganser, P., Bergs, T.: Security-enhanced cutting tool wear segmentation with federated learning. Procedia CIRP 138, 415–420 (2026). https://doi.org/https://doi.org/10.1016/j.procir.2026.01.072, https://w.sciencedirect.com/science/article/pii/S2212827126000727,18th CIRP Conference on Intelligent Computation in Manufacturing Engineering 3. De Pauw, L., Jacobs, T., Goedemé, T.: Matwi: A multimodal automatic tool wear inspection dataset and baseline algorithms. In: International Conference on Com- puter Vision Systems. p. 255–269. Springer (2023) 4. Huang, G.Y., Lee, C.H.: Industrial federated learning algorithm (p-pfedsgd) for tool wear estimation. Future Generation Computer Systems 158, 150– 157(2024).https://doi.org/https://doi.org/10.1016/j.future.2024.04.026, https://w.sciencedirect.com/science/article/pii/S0167739X24001547 5. Kairouz, P., McMahan, H.B., Avent, B., Bellet, A., Bennis, M., Nitin Bhagoji, A., Bonawitz, K., Charles, Z., Cormode, G., Cummings, R., D’Oliveira, R.G.L., Eichner, H., El Rouayheb, S., Evans, D., Gardner, J., Garrett, Z., Gascón, A., Ghazi, B., Gibbons, P.B., Gruteser, M., Harchaoui, Z., He, C., He, L., Huo, Z., Hutchinson, B., Hsu, J., Jaggi, M., Javidi, T., Joshi, G., Khodak, M., Konecný, J., Korolova, A., Koushanfar, F., Koyejo, S., Lepoint, T., Liu, Y., Mittal, P., Mohri, M., Nock, R., Özgür, A., Pagh, R., Qi, H., Ramage, D., Raskar, R., Raykova, M., Song, D., Song, W., Stich, S.U., Sun, Z., Suresh, A.T., Tramèr, F., Vepakomma, P., Wang, J., Xiong, L., Xu, Z., Yang, Q., Yu, F.X., Yu, H., Zhao, S.: Advances and open problems in federated learning. Found. Trends Mach. Learn. 14(1–2), 1–210 (Jun 2021). https://doi.org/10.1561/2200000083, https://doi.org/10.1561/2200000083 6. Kaleli, I.S., Unal, P., Deveci, B.U., Albayrak, O., Ozbayoglu, A.M.: A domain- aware federated learning study for cnc tool wear estimation. In: Younas, M., Awan, I., Petcu, D., Feng, B. (eds.) Mobile Web and Intelligent Information Systems. p. 250–265. Springer Nature Switzerland, Cham (2024) 7. Kennedy, C.H., Hilal, A., Momeni, M.: The role of federated learning in improving financial security: A survey. In: 2025 IEEE Global Conference on Artificial Intelligence and Internet of Things (GCAIoT). p. 1–8 (2025). https://doi.org/10.1109/GCAIoT68269.2025.11275535 16A. Khan et al. 8. Li, X.: 2010 phm society conference data challenge (2021). https://doi.org/10.21227/jdxd-y51, https://dx.doi.org/10.21227/jdxd-y51 9. Lin, Y.R., Lee, C.H., Lu, M.C.: Robust tool wear monitoring sys- tem development by sensors and feature fusion. Asian Journal of Con- trol 24(3), 1005–1021 (2022). https://doi.org/https://doi.org/10.1002/asjc.2741, https://onlinelibrary.wiley.com/doi/abs/10.1002/asjc.2741 10. Mallat, S.G.: A theory for multiresolution signal decomposition: the wavelet repre- sentation. IEEE transactions on pattern analysis and machine intelligence 11(7), 674–693 (1989) 11. Mativenga, P., Schoop, J., Jawahir, I., Biermann, D., Kipp, M., Kilic, Z.M., Özel, T., Wertheim, R., Arrazola, P., Boing, D.: Engineered design of cutting tool material, geometry, and coating for optimal performance and customized ap- plications: A review. CIRP Journal of Manufacturing Science and Technology 52, 212–228 (2024). https://doi.org/https://doi.org/10.1016/j.cirpj.2024.06.001, https://w.sciencedirect.com/science/article/pii/S1755581724000798 12. Mehta, S., Aneja, A.: Securing data privacy in machine learning: The fedavg of federated learning approach. In: 2024 4th Asian Conference on Innovation in Tech- nology (ASIANCON). p. 1–5. IEEE (2024) 13. Noor, A., Chaman, U.M., Kabir, M.S.: Federated learning in healthcare: A comprehensive survey on privacy, scalability and clinical applications. ICT Express (2026). https://doi.org/https://doi.org/10.1016/j.icte.2026.05.011, https://w.sciencedirect.com/science/article/pii/S2405959526000834 14. Patel, D., Muthuswamy, S.: Tool wear state recognition in CNC milling using reliable labels, robust domain features, and lightweight AI models. J. Intell. Manuf. 37(5), 2129–2164 (May 2026) 15. Purkayastha, A.A., Aggarwal, S.: Federated Learning for Predic- tive Maintenance: A Survey of Methods, Applications, and Chal- lenges. In: 2024 IEEE 67th International Midwest Symposium on Cir- cuits and Systems (MWSCAS). p. 238–242. IEEE, Springfield, MA, USA (Aug 2024). https://doi.org/10.1109/MWSCAS60917.2024.10658900, https://ieeexplore.ieee.org/document/10658900/ 16. Sana, T.Z., Abdulla, S., Nag, A., Das, A., Hassan, M.M., Fiza, Z.Z., Karim, A., Kabir, S.R.R.: Advancing Federated Learning: A Systematic Literature Review of Methods, Challenges, and Applications. IEEE Ac- cess 13, 153817–153844 (2025). https://doi.org/10.1109/ACCESS.2025.3605165, https://ieeexplore.ieee.org/document/11146653/ 17. Shah, R., Pai, N., Thomas, G., Jha, S., Mittal, V., Shirvni, K., Liang, H.: Machine learning in wear prediction. Journal of Tribology 147(4), 040801 (11 2024) 18. Soori, M., Arezoo, B., Dastres, R.: Machine learning and artificial intelligence in cnc machine tools, a review. Sustainable Manufacturing and Service Economics 2, 100009 (2023). https://doi.org/https://doi.org/10.1016/j.smse.2023.100009, https://w.sciencedirect.com/science/article/pii/S2667344423000014 19. Yang, C., Zhou, J., Li, E., Wang, M., Jin, T.: Local-feature and global-dependency based tool wear prediction using deep learning. Scientific Reports 12(1), 14574 (2022) 20. Zhu, H., Xu, J., Liu, S., Jin, Y.: Federated learning on non- IID data: A survey. Neurocomputing 465, 371–390 (Nov 2021), https://w.sciencedirect.com/science/article/pii/S0925231221013254