Abstract:Ultrasonic metal welding (UMW) is a key joining technology with widespread industrial applications. Condition monitoring (CM) capabilities are critically needed in UMW applications because process anomalies significantly deteriorate the joining quality. Recently, machine learning models emerged as a promising tool for CM in many manufacturing applications due to their ability to learn complex patterns. Yet, the successful deployment of these models requires substantial training data that may be expensive and time-consuming to collect. Additionally, many existing machine learning models lack generalizability and cannot be directly applied to new process configurations (i.e., domains). Such issues may be potentially alleviated by pooling data across manufacturers, but data sharing raises critical data privacy concerns. To address these challenges, this paper presents a Federated Transfer Learning with Task Personalization (FTL-TP) framework that provides domain generalization capabilities in distributed learning while ensuring data privacy. By effectively learning a unified representation from feature space, FTL-TP can adapt CM models for clients working on similar tasks, thereby enhancing their overall adaptability and performance jointly. To demonstrate the effectiveness of FTL-TP, we investigate two distinct UMW CM tasks, tool condition monitoring and workpiece surface condition classification. Compared with state-of-the-art FL algorithms, FTL-TP achieves a 5.35%--8.08% improvement of accuracy in CM in new target domains. FTL-TP is also shown to perform excellently in challenging scenarios involving unbalanced data distributions and limited client fractions. Furthermore, by implementing the FTL-TP method on an edge-cloud architecture, we show that this method is both viable and efficient in practice. The FTL-TP framework is readily extensible to various other manufacturing applications.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges faced by condition monitoring (CM) in ultrasonic metal welding (UMW), especially the lack of generalization ability and adaptability of existing machine - learning models when the process configuration changes. Specifically: 1. **High - cost and time - consuming data collection**: To train effective CM models, a large amount of training data is required, and the collection of these data is both expensive and time - consuming. 2. **Poor model generalization ability**: Most of the existing CM methods are for fixed configurations (i.e., specific fields). When the UMW configuration (such as material type, welding parameters, etc.) changes, the performance of the trained model will decline significantly. 3. **Data privacy issues**: Improving the generalization ability of models through data sharing across manufacturers or factories is a potential solution, but data sharing will cause critical data privacy issues, which is a major obstacle in the industry. To solve these problems, the paper proposes a framework named **Federated Transfer Learning with Task Personalization (FTL - TP)**. This framework aims to achieve domain generalization ability in distributed learning while ensuring data privacy. By effectively learning a unified representation from the feature space, FTL - TP can adapt to the work tasks of different clients, thereby enhancing its overall adaptability and performance. ### Main contributions: 1. **Domain generalization ability**: Compared with the baseline federated learning algorithms (such as FedAvg and FedProx), FTL - TP shows higher accuracy in new, unseen target domains, with an improvement of 5.35% - 8.08%. 2. **Data privacy protection**: By implementing federated learning, this research first integrates data privacy into data - driven condition monitoring in UMW, solves the widespread data security and privacy problems in the industry, and lays the foundation for large - scale cooperation between enterprises using UMW for production. 3. **Verification in practical applications**: FTL - TP is implemented in a real - world network manufacturing environment and its computational efficiency is evaluated. By running on an edge - cloud architecture, using Raspberry Pi as an edge device and a message broker for weight transmission, it is proved that this method is not only data - efficient but also time - efficient and has the potential to be widely used in intelligent manufacturing. In conclusion, through the FTL - TP framework, this paper solves the problems of difficult data collection, poor model generalization ability and data privacy in UMW condition monitoring, and provides an effective and practical solution for industrial applications.

Federated Transfer Learning with Task Personalization for Condition Monitoring in Ultrasonic Metal Welding

Privacy-preserving federated transfer learning for defect identification from highly imbalanced image data in additive manufacturing

Maximizing model generalization for machine condition monitoring with Self-Supervised Learning and Federated Learning

Personalized Federated Learning for Multi-task Fault Diagnosis of Rotating Machinery

Data Privacy Protection: A Novel Federated Transfer Learning Scheme for Bearing Fault Diagnosis

Deep Spatial-Temporal Feature Extraction and Lightweight Feature Fusion for Tool Condition Monitoring

Federated learning-based collaborative manufacturing for complex parts

Self-adaptive fusion of local-temporal features for tool condition monitoring: A human experience free model

Deep transfer learning for tool condition monitoring under different processing conditions

Investigation on domain adaptation of additive manufacturing monitoring systems to enhance digital twin reusability

Aggregation Strategy on Federated Machine Learning Algorithm for Collaborative Predictive Maintenance

WeldMon: A Cost-effective Ultrasonic Welding Machine Condition Monitoring System

Task incremental learning-driven Digital-Twin predictive modeling for customized metal forming product manufacturing process

Privacy-preserving Anomaly Detection in Cloud Manufacturing via Federated Transformer

A federated learning approach to mixed fault diagnosis in rotating machinery

Federated transfer learning for auxiliary classifier generative adversarial networks: framework and industrial application

FedDGA: Federated Multi-Task Learning Based on Dynamic Guided Attention

Model and Feature Aggregation Based Federated Learning for Multi-sensor Time Series Trend Following.

A novel approach to tool condition monitoring based on multi-sensor data fusion imaging and an attention mechanism

Federated Transfer Learning for Bearing Fault Diagnosis With Discrepancy-Based Weighted Federated Averaging