publications

2024

1. Comparing Feature and Trajectory-Based Remaining Useful Life Modeling of Electrical Resistance Heating Wires

   Simon Mählkvist, Wilhelm Söderkvist Vermelin, Thomas Helander, and 1 more author

   In , 2024

   Abs

   Industrial heating significantly contributes to global greenhouse gas emissions, accounting for a substantial portion of annual emissions. The transition to fossil-free operations in the heating industry is closely linked to advancements in industrial electrical heating systems, especially those using resistance heating wires. In this context, Prognostics and Health Management is crucial for enhancing system reliability and sustainability through predictive maintenance strategies. The integration of machine learning technologies into Prognostics and Health Management has significantly improved the precision and applicability of Remaining Useful Life modeling. This improvement enables more accurate predictions of component lifespans, optimizes maintenance schedules , and enhances operational efficiency in industrial heating applications. These developments are essential for reducing greenhouse gas emissions in the sector. This paper serves as a guide for conducting Remaining Useful Life modeling for industrial batch processes. It evaluates and compares two methodologies: deep learning-based approaches using full time-series data, such as recurrent neural networks and their variants, and feature-engineering-based methods, including random forest regression and support vector machines. Our results show that the feature-oriented approach performs better overall in terms of predictive accuracy and computational efficiency. The study includes a detailed sensitivity analysis and hyperparameter estimation for each method, providing valuable insights into developing robust and transparent Prognostics and Health Management systems. These systems are crucial in supporting the heating industry’s move towards more sustainable and emission-free operations. The findings reveal that feature-oriented methods are both performant and robust, particularly excelling in handling outliers. The random forest regression model, in particular, demonstrated the highest performance on the test dataset according to the chosen evaluation metrics. Conversely, trajectory-oriented methods exhibited less bias across varying levels of degradation, a helpful characteristic for Prognostics and Health Management systems. While feature-oriented methods tend to systematically underestimate Remaining Useful Life at high true values and overestimate it at low actual values, this issue is less pronounced in trajectory-oriented models. Overall, these insights highlight the strengths and limitations of each approach, guiding the development of more effective and reliable predictive maintenance strategies.

2. Collaborative Training of Data-Driven Remaining Useful Life Prediction Models Using Federated Learning

   Wilhelm Söderkvist Vermelin, Madhav Mishra, Mattias P. Eng, and 2 more authors

   International Journal of Prognostics and Health Management, Oct 2024

   Abs

   Remaining useful life prediction models are a central aspect of developing modern and capable prognostics and health management systems. Recently, such models are increasingly data-driven and based on various machine learning techniques, in particular deep neural networks. Such models are notoriously "data hungry", i.e., to get adequate performance of such models, a substantial amount of diverse training data is needed. However, in several domains in which one would like to deploy data-driven remaining useful life models, there is a lack of data or data are distributed among several actors. Often these actors, for various reasons, cannot share data among themselves. In this paper a method for collaborative training of remaining useful life models based on federated learning is presented. In this setting, actors do not need to share locally held secret data, only model updates. Model updates are aggregated by a central server, and subsequently sent back to each of the clients, until convergence. There are numerous strategies for aggregating clients’ model updates and in this paper two strategies will be explored: 1) federated averaging and 2) federated learning with personalization layers. Federated averaging is the common baseline federated learning strategy where the clients’ models are averaged by the central server to update the global model. Federated averaging has been shown to have a limited ability to deal with non-identically and independently distributed data. To mitigate this problem, federated learning with personalization layers, a strategy similar to federated averaging but where each client is allowed to append custom layers to their local model, is explored. The two federated learning strategies will be evaluated on two datasets: 1) run-to-failure trajectories from power cycling of silicon-carbide metal-oxide semiconductor field-effect transistors, and 2) C-MAPSS, a well-known simulated dataset of turbofan jet engines. Two neural network model architectures commonly used in remaining useful life prediction, long short-term memory with multi-layer perceptron feature extractors, and convolutional gated recurrent unit, will be used for the evaluation. It is shown that similar or better performance is achieved when using federated learning compared to when the model is only trained on local data.

3. A Link Between the Lab and the Real World - A Setup for Accelerated Aging of Power Electronics Using Mission Profiles from the Field

   Mattias P. Eng, Madhav Mishra, Wilhelm Söderkvist Vermelin, and 2 more authors

   In 2024 25th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE), Apr 2024

   Abs

   To generate data used for developing schemes and models for CM, PHM, and for estimating RUL of power electronic devices, accelerated aging experiments in the form of power cycling are often performed. In these experiments, a set current is passed through the power devices and is turned on and off in regular cycles. Due to the mismatch in CTEs of the materials in the devices , the on/off cycles will generate thermally induced stress in the various material interfaces, which is the main cause of failures. Most of the power cycling setups that are currently used can only manage a single set on-state current level and fixed on/off times (which is also the common standard for lifetime testing); a condition that is very far from most real applications. The experimental setup described here is based on a Gamry Reference 3000AEpotentiostat/galvanostat/ZRA working with a Gamry 30k Booster , which can be programmed to generate a variable load current profile and will thus enable the application of more realistic conditions for accelerated aging of power electronic devices in the lab. This will improve prognostics model development and provide excellent use cases for evaluating the capabilities of the prognostics algorithms for generalization to field conditions. The application of variable load profiles from the field, instead of the regular on/off cycles traditionally used, is not compatible with the commonly used method of using the chip itself as a temperature sensor. Instead, we here present a novel method of estimating the junction temperature using a device specific derivation of thermal parameters from the measured cooling block temperature, case temperature, and dissipated power in conjunction with simulations using the PySpice simulation package implemented in Python. The setup coupled with the new junction temperature estimation is an important step in enabling predictive maintenance of power devices that is currently missing from the power electronics community.

2023

1. Data-Driven Remaining Useful Life Estimation of Discrete Power Electronic Devices

   Wilhelm Söderkvist Vermelin, Andreas Lövberg, Maciej Misiorny, and 2 more authors

   In Proceedings of the 33rd European Safety and Reliability Conference (ESREL 2023), Apr 2023

   Abs Bib

   Robust and accurate prognostics models for estimation of remaining useful life (RUL) are becoming an increasingly important aspect of research in reliability and safety in modern electronic components and systems. In this work, a data driven approach to the prognostics problem is presented. In particular, machine learning models are trained to predict the RUL of wire-bonded silicon carbide (SiC) metal-oxide-semiconductor field-effect transistors (MOSFETs) subjected to power cycling until failure. During such power cycling, ON-state voltage and various temperature measurements are continuously collected. As the data set contains full run-to-failure trajectories, the issue of estimating RUL is naturally formulated in terms of supervised learning. Three neural network architectures were trained, evaluated, and compared on the RUL problem: a temporal convolutional neural network (TCN), a long short-term memory neural network (LSTM) and a convolutional gated recurrent neural network (Conv-GRU). While the results show that all networks perform well on held out testing data if the testing samples are of similar aging acceleration as the samples in the training data set, performance on out-of-distribution data is significantly lower. To this end, we discuss potential research directions to improve model performance in such scenarios.

   @inproceedings{data-driven-sic-mosfet-rul-estimation,
     author = {Söderkvist Vermelin, Wilhelm and Lövberg, Andreas and Misiorny, Maciej and P. Eng, Mattias and Brinkfeldt, Klas},
     publisher = {European Safety and Reliability Conference},
     booktitle = {Proceedings of the 33rd European Safety and Reliability
                    Conference (ESREL 2023)},
     title = {Data-Driven Remaining Useful Life Estimation of Discrete Power
                Electronic Devices},
     year = {2023},
     volume = {33},
     doi = {10.3850/978-981-18-8071-1-4procd},
     url = {https://urn.kb.se/resolve?urn=urn:nbn:se:ri:diva-67109},
   }

2. A Simple Hybrid Model for Estimating Remaining Useful Life of SiC MOSFETs in Power Cycling Experiments

   Mattias P Eng, Andreas Lövberg, Maciej Misiorny, and 3 more authors

   In PHM Society Asia-Pacific Conference, Apr 2023

   Abs Bib

   Recording and prediction of the accumulated damage, which will eventually lead to the failure of power electronic modules, is an aspect of high importance for power electronic systems design and, in particular, for development of Prognostic and Health Management (PHM) schemes for in-field applications. To this end, this paper presents a simple and cost-effective prognostic method for predicting the remaining useful life (RUL) of TO-247 packaged silicon carbide (SiC) metal-oxide semiconductor field-effect transistors (MOSFETs) subjected to power cycling experiments. The model assumes that the major failure mode is bond-wire lift-off and uses a damage accumulation scheme based on Paris’ crack law. The only inputs to the model are historical data on the average junction temperature swing and the temperature-compensated drain-source ON-state resistance at the peak temperature of the current cycle. Using only these two input values, the model is shown to predict RUL with surprising accuracy for the range of constant current loads determining cycling conditions under which the test data series have been acquired. This work is a first step in an ongoing project towards building more elaborate prognostic schemes for RUL-determination of SiC power MOSFETs in actual working conditions, using physics-informed neural networks (PINNs).

   @inproceedings{eng2023simple,
     title = {A Simple Hybrid Model for Estimating Remaining Useful Life of SiC
                MOSFETs in Power Cycling Experiments},
     author = {Eng, Mattias P and L{\"o}vberg, Andreas and Misiorny, Maciej and Vermelin, Wilhelm S{\"o}derkvist and Brinkfeldt, Klas and Mishra, Madhav},
     booktitle = {PHM Society Asia-Pacific Conference},
     volume = {4},
     number = {1},
     year = {2023},
   }

2022

1. Self-supervised Learning for Efficient Remaining Useful Life Prediction

   Wilhelm Söderkvist Vermelin, Andreas Lövberg, and Konstantinos Kyprianidis

   In Vol. 14 No. 1 (2022): Proceedings of the Annual Conference of the PHM Society 2022, Apr 2022

   Abs Bib

   Canonical deep learning-based Remaining Useful Life (RUL) prediction relies on supervised learning methods which in turn requires large data sets of run-to-failure data to ensure model performance. In a large class of cases, run-to-failure data is difficult to collect in practice as it may be expensive and unsafe to operate assets until failure. As such, there is a need to leverage data that are not run-to-failure but may still contain some measurable, and thus learnable, degradation signal. In this paper, we propose utilizing self-supervised learning as a pretraining step to learn representations of the data which will enable efficient training on the downstream task of RUL prediction. The self-supervised learning task chosen is time series sequence ordering, a task that involves constructing tuples each consisting of n sequences sampled from the time series and reordered with some probability p. Subsequently, a classifier is trained on the resulting binary classification task; distinguishing between correctly ordered and shuffled tuples. The classifier’s weights are then transferred to the RUL-model and fine-tuned using run-to-failure data. We show that the proposed self-supervised learning scheme can retain performance when training on a fraction of the full data set. In addition, we show indications that self-supervised learning as a pretraining step can enhance the performance of the model even when training on the full run-to-failure data set. To conduct our experiments, we use a data set of simulated run-to-failure turbofan jet engines.

   @inproceedings{ssl-rul,
     author = {S{\"o}derkvist Vermelin, Wilhelm and L{\"o}vberg, Andreas and Kyprianidis, Konstantinos},
     booktitle = {Vol. 14 No. 1 (2022): Proceedings of the Annual Conference of the
                    PHM Society 2022},
     doi = {10.36001/phmconf.2022.v14i1.3222},
     eid = {3222},
     institution = {M{\"a}lardalen University, Sweden},
     keywords = {self-supervised learning, self-supervised, deep learning, machine
                   learning, remaining useful life, unsupervised learning, neural
                   networks},
     title = {Self-supervised Learning for Efficient Remaining Useful Life
                Prediction},
     volume = {14},
     year = {2022},
     url = {https://doi.org/10.36001/phmconf.2022.v14i1.3222},
   }

2019

1. Discovering Patterns in Driving Data

   Karen Baca Mendoza, and Wilhelm Söderkvist Vermelin

   Apr 2019

   Master’s thesis in Engineering Mathematics and Computational Science

   Abs Bib

   Modern vehicles are equipped with numerous sensors for providing feedback to the control unit. These measurements hold a substantial amount of information about the driver’s action, environmental and traffic conditions. In this thesis, we investigate using various machine leaning techniques to analyze driving data for discovering repetitive patterns when facing similar traffic situations. To this end, we first use unsupervised learning and data mining techniques to find driving patterns and to develop a labeling scheme. This last point consist of finding patterns in individual signals which are then combined to find patterns describing more complex behaviors. The discovered patterns and labels are used in the second part of the thesis to develop a classifier for recognizing the current driving situation. The classifier is designed such that it can be implemented in an Electric Control Unit of a production vehicle. After the analysis, we were able to discover intelligible driving scenarios and we focused on some of them to label our data. We used this labeling to train and compare four different neural network architectures commonly used in time series classification. The models are trained by simulating an online situation where data comes in a form of data stream. The results show that online classification is feasible. Implementing the classifier in the vehicle software could be beneficial for aiding the control unit in deciding gear shifts, energy recuperation and propulsion system. This may lead to a more efficient vehicle and a better driving experience.

   @misc{driving-patterns,
     author = {Baca Mendoza, Karen and Söderkvist Vermelin, Wilhelm},
     institution = {Mathematical Sciences, Chalmers University of Technology},
     keywords = {pattern discovery, machine learning, clustering, data mining,
                   classification, neural networks, time series data, traffic
                   scenarios, vehicle control},
     note = {Master's thesis in Engineering Mathematics and Computational Science},
     school = {Chalmers University of Technology},
     title = {Discovering Patterns in Driving Data},
     year = {2019},
     url = {https://hdl.handle.net/20.500.12380/300095},
   }

2016

1. 3+1 Approach to Cosmological Perturbations : Deriving the First Order Scalar Perturbations of the Einstein Field Equations

   Wilhelm Söderkvist Vermelin

   Apr 2016

   Bachelor’s thesis in Physics

   Abs Bib

   Experimental data suggest that the universe is homogeneous and isotropic on sufficiently large scales. An exact solution of the Einstein field equations exists for a homogeneous and isotropic universe, also known as a Friedmann-Lemaı̂tre-Robertson-Walker (FLRW) universe. However, this model is only a first approximation since we know that, locally, the universe has anisotropic and inhomogeneous structures such as galaxies and clusters of galaxies. In order to successfully introduce inhomogeneities and anisotropies to the model one uses perturbative methods. In cosmological perturbations the FLRW universe is considered the zeroth order term in a perturbation expansion and perturbation theory is used to derive higher order terms which one tries to match with observations. In this thesis I present a review of the main concepts of general relativity, discuss the 3+1 formalism which gives us the Einstein field equations in a useful form for the perturbative analysis, and lastly, I derive the first order scalar perturbations of the Einstein field equations.

   @misc{3+1-approach,
     author = {Söderkvist Vermelin, Wilhelm},
     institution = {Karlstad University, Faculty of Health, Science and Technology
                      (starting 2013)},
     keywords = {physics, astronomy, theoretical physics, perturbation, cosmology,
                   cosmological perturbations, 3+1 formalism, first order perturbation
                   , perturbation theory, universe, fysik, astronomi, teoretisk fysik,
                   st{\"o}rning, st{\"o}rningsr{\"a}kning, kosmologi, kosmologiska st{
                   \"o}rningar, kosmologisk st{\"o}rningsr{\"a}kning, 3+1 formalismen,
                   f{\"o}rsta ordningens st{\"o}rningar, st{\"o}rningsteori, universum
                   },
     pages = {57},
     note = {Bachelor's thesis in Physics},
     school = {Karlstad University, Faculty of Health, Science and Technology
                 (starting 2013)},
     title = {3+1 Approach to Cosmological Perturbations : Deriving the First Order
                Scalar Perturbations of the Einstein Field Equations},
     year = {2016},
     url = {https://urn.kb.se/resolve?urn=urn:nbn:se:kau:diva-43257},
   }