Master Thesis - Nonlinear Filtering and Sensor Fusion for Clamp Force Estimation

Level of thesis project: Master thesis
Education requirement: Mechatronics, Mechanical or Electrical Engineering
Functional area: Atlas Copco Industrial Technique AB and Tightening technique and lab development
Number of students for this project: 1
Start: Flexible, early 2026 preferred

Your role

Target:
The goal of this master thesis is to develop and validate advanced algorithms for real-time estimation of clamp force in bolted joints. By combining nonlinear filtering and sensor fusion techniques, the project aims to improve accuracy and robustness beyond what is achievable with standard torque–angle methods. The results will contribute to Atlas Copco’s next generation of intelligent tightening systems with improved quality assurance and process transparency.

Background:
Clamp force is the primary quality indicator in bolted joints, but it cannot be directly measured during tightening. Instead, industry relies on indirect measurements such as torque and angle, which are affected by variations in friction and joint stiffness. Recent research at Atlas Copco has demonstrated that dynamic responses—such as resonance frequencies of the bolt–joint system—contain valuable information about preload. By fusing these signals with torque–angle data through model-based estimation, the accuracy and observability of the tightening process can be significantly improved. This project builds on ongoing research within Atlas Copco and Linköping University on model-based clamp-force estimation.

Mission:
The thesis will formalize and extend the existing estimation framework for clamp-force prediction.

Key tasks include:

• Implement and compare nonlinear filtering methods such as Extended and Unscented Kalman Filters, particle filters, and hybrid approaches.
• Quantify estimator performance through Monte Carlo simulations, sensitivity analysis, and Cramér–Rao lower bounds.
• Integrate additional sensing (e.g., accelerometers, microphones) to extract eigenfrequency-based features correlated with preload.
• Fuse these measurements with torque–angle data to improve robustness and accuracy.
• Build and test an experimental prototype demonstrating real-time clamp-force estimation on a standard tightening tool.
• The outcome will be a validated proof-of-concept showing how multimodal sensing and nonlinear filtering can enhance clamp-force estimation in industrial tightening applications.

Take the opportunity to write your master's thesis with our engaged team and contribute to our research on future tightening techniques!

To succeed, you will need

• Knowledge of control systems, estimation theory, and/or signal processing
• Experience with MATLAB/Simulink (or similar simulation tools)
• Knowledge of mechanical systems, preferably threaded fastener mechanics
• Knowledge of data acquisition and instrumentation
• Interest in statistical methods (e.g., Monte Carlo simulations, Cramér–Rao lower bound (CRLB), sensitivity analysis)
• Interest in hands-on engineering (test-rig development)
• Interest in implementing and test real-time algorithms

Job location

This role requires you to work on-site at our office in Stockholm, Sweden (SE).

Contact information

Recruiting Team:  Nils Dressler, Erik Persson

Phone: +46721424923

Email:

Uniting curious minds
Behind every innovative solution, there are people working together to transform the future. With careers sparked by initiative and lifelong learning, we unite curious minds, and you could be one of them.

Din första dag på Atlas Copco är den första dagen på en spännande resa. Hos oss finns en mängd olika utmanande jobb inom ett otal branscher världen över. Hos oss kan dina idéer ge människor en bättre livskvalitet. Kom till de industriella idéernas hemvist.