Publications

Academia Overview
for full Abstracs and download, please see below
- Energy-Aware Speed Regulation in Electrical Drives: A Load-Agnostic
Motor Control Approach via Reinforcement Learning
June 2024, Automotive, Infineon Technologies AG, Munich, Germany,
Electronic Systems TU Eindhoven, Netherlands; Electronic Design Automation, TU Munich, Munich, Germany - Artificial Intelligence for Predictive Maintenance Applications: Key Components, Trustworthiness, and Future Trends
January 2024, R&D Department Albayrak Makine Elektronik A.Ş., Fırat University, Türkiye - Towards Rapid Exploration of Heterogeneous TinyML Systems using Virtual Platforms and TVM’s UMA
Workshop on Compilers, Deployment, and Tooling for Edge AI (CODAI)
September 2023, University of Munich, Germany - Entwicklung einer Toolchain zur automatisierten Codeerzeugung von Steuerung/Regelung von permanenterregten Synchronmaschinen mit neuronalen Netzen
November 2023, University of Ilmenau, Moteon GmbH, Germany - Machine Learning based Diagnostic Approach to Condition Monitoring of Railway Platform Screen Door Systems
April 2023, Dept. of Electrical and Electronics Engineering, Eskişehir Technical University, Türkiye, Dept. of Computer Engineering, Elazığ Fırat University, Türkiye, Albayrak Makine Elektronik San. Tic. A.Ş. Eskişehir, Türkiye - Modeling and Simulation of Platform Screen Door (PSD) System using MATLAB-SIMULINK
October 2022, R&D Department Albayrak Makine Elektronik A.Ş., Eskişehir Vocational School, Eskisehir Osmangazi University, Department of Computer Engineering, Fırat University, Türkiye - A New Fault Detection and Classification Approach for Platform Screen Door Systems Using Artificial Neural Networks
October 2022, R&D Department Albayrak Makine Elektronik A.Ş., Eskişehir Vocational School, Eskisehir Osmangazi University, Department of Computer Engineering, Fırat University, Türkiye

Commercial
Use Cases
Electric Drive Systems and Predictive Maintenance
Please visit also our YouTube Channel to access our videos about our use cases, test benches and project goals.
Use Cases
Electric Drive Systems and Predictive Maintenance

ECOMAI Project Overview

Academia Abstract Downloads
Energy-Aware Speed Regulation in Electrical Drives: A Load-Agnostic
Motor Control Approach via Reinforcement Learning
Authors
Steven Klotz 1, Thorsten Bucksch 1, Dip Goswami 2 and Daniel Mueller-Gritschneder 3
1 Automotive, Infineon Technologies AG, Munich, Germany (steven.klotz,thorsten.bucksch)@infineon.com
2 Electronic Systems, TU Eindhoven, Eindhoven, Netherlands
d.goswami@tue.nl
3 Electronic Design Automation, TU Munich, Munich, Germany
ABstract
Abstract—Robotic and automotive platforms are rapidly expanding in features and are incorporating more and more electric motor components. Consequently, the energy efficiency
of motor control systems emerges as a major design challenge. The process of formulating and fine-tuning specialized speed regulation strategies for each application becomes progressively more laborious and expensive. A reinforcement learning agent specialized in electrical motor dynamics, capable of generalizing across a wide range of possible end-use applications, presents a promising and convenient solution.
In this article, we introduce a novel design of a reinforcement learning agent, grounded in time series analysis, intended for application-agnostic electric motor control that optimizes
both speed regulation and energy efficiency. Trained on the motor’s internal dynamics, the agent provides operating pointspecific control inputs, eliminating the need for manual tuning and application system-identification. Compared to application tuned classical control methods, the agent exhibited on-par or improved speed regulation performance and demonstrated advanced capability to save energy, showcasing its potential for
future applications.
Artificial Intelligence for Predictive Maintenance Applications: Key Components, Trustworthiness, and Future Trends
Authors
Aysegul Ucar 1, Mehmet Karakose 2, Necim Kırımça 3
1 Mechatronics Engineering Department, Elazığ Fırat University, Türkiye
2 Computer Engineering Department, Elazığ Fırat University, Türkiye
3 R&D Department, Albayrak Makine Elektronik A.S., Eskisehir 26110, Türkiye
aBstract
Predictive maintenance (PdM) is a policy applying data and analytics to predict when
one of the components in a real system has been destroyed, and some anomalies appear so that maintenance can be performed before a breakdown takes place. Using cutting-edge technologies like data analytics and artificial intelligence (AI) enhances the performance and accuracy of predictive maintenance systems and increases their autonomy and adaptability in complex and dynamic working environments. This paper reviews the recent developments in AI-based PdM, focusing on key components, trustworthiness, and future trends. The state-of-the-art (SOTA) techniques, challenges, and opportunities associated with AI-based PdM are first analyzed. The integration of AI technologies into PdM in real-world applications, the human robot interaction, the ethical issues emerging from using AI, and the testing and validation abilities of the developed policies are later discussed. This study exhibits the potential working areas for future research, such as digital twin, metaverse, generative AI, collaborative robots (cobots), blockchain technology, trustworthy AI, and Industrial Internet of Things (IIoT), utilizing a comprehensive survey of the current SOTA techniques, opportunities, and challenges allied with AI-based PdM.
Towards Rapid Exploration of Heterogeneous TinyML Systems using Virtual Platforms and TVM’s UMA
Authors
Samira Ahmadifarsani, Rafael Stahl, Philipp van Kempen, Daniel Mueller-Gritschneder, Ulf Schlichtman; TU Munich, Germany
Abstract
Workshop on Compilers, Deployment, and Tooling for Edge AI (CODAI2023)
The rapid setup of deep learning compilation toolchains for heterogeneous TinyML systems with a processor and dedicated ML accelerator is still at an early stage. Here, achieving the most optimal combination of targets for a TinyML application on ultra-low-power edge devices demands additional benchmarking solutions to estimate the final performance.
ApacheTVM’sUniversalModularAccelerator (UMA) interface as an easy-to-use API is a promising speed-up approach to this scope. In this paper, we integrate a simple custom dedicated accelerator into TVM using UMA to offload the quantized convolution operators in order to demonstrate such an approach. Furthermore, we leverage MLonMCU tool and its capability of virtual prototyping to estimate and explore the performance improvement achieved by the accelerator
Machine Learning based Diagnostic Approach to Condition Monitoring of Railway Platform Screen Door Systems
Authors
Şükrü Görgülü*1, İsa Koç 2, Necim Kırımça 3, Ömer Mermer 4, Mehmet Karaköse 5, Mehmet Tankut Özgen 6
1,6 Dept. of Electrical and Electronics Engineering, Eskişehir Technical University, Türkiye
2 Dept. of Electrical and Electronics Engineering, Eskişehir Osmangazi University, Türkiye
5 Dept. of Computer Engineering, Elazığ Fırat University, Türkiye
1,2,3,4 Albayrak Makine Elektronik San. Tic. A.Ş. Eskişehir, Türkiye
Abstract
Paper presented at 2nd International Conference on Engineering, Natural and Social Sciences on April 4-6, 2023 : Konya, Turkey
This study focuses on the use of machine learning-based fault classification to investigate the condition monitoring of Platform Screen Door (PSD) Systems in the railway industry. PSDs are safety systems installed in subway or train stations to prevent passengers from falling onto the tracks. The electromechanical system consists of sliding doors or gates that run along the platform edge and only open when the train is properly aligned with the platform.
Modeling and Simulation of Platform Screen Door (PSD) System using MATLAB-SIMULINK
Authors
R&D Department Albayrak Makine Elektronik A.Ş.
Eskişehir Vocational School, Eskisehir Osmangazi University
Department of Computer Engineering, Fırat University
Abstract
In this paper, electromechanical modeling of an industrial platform screen door (PSD) system has been developed. The system incorporates several sub-units such as PMSM, motor controller, inverter, and mechanical unit.
Moreover, a newly designed motor control system based on real operation conditions is introduced. In addition, modeling procedures are described, and simulation results are presented.
Furthermore, numerical simulation in accordance with specific velocity and position references are provided for verifying the effectiveness of the proposed modeling. Furthermore, the results show that the developed electromechanical model of PSD system is effective in managing the parameters that affect the actual operating state of the PSD sliding door system. These modelling and simulation will play an important role.
A New Fault Detection and Classification Approach for Platform Screen Door Systems Using Artificial Neural Networks
Authors
R&D Department Albayrak Makine Elektronik A.Ş.
Department of E&E Engineering Eskişehir Osmangazi University
Department of Computer Engineering, Fırat University
Abstract
Platform Screen Door (PSD) system, which is called a safety-critical system, is a sliding barrier door installed on the sides of a platform in many modern metros and RBT (Rapid Bus Transit) stations. Failures that may occur in the PSD system will seriously affect the availability of train transportation as well as the passenger’s safety. In this study, data-driven fault detection and classification method have been studied on the PSD system to ensure the safe and reliable operation. An artificial neural network (ANN) is preferred because of its powerful capabilities. Different operating conditions (normal and faulty) were created artificially on the PSD and the motor current, motor voltage, and door speed signals were used as input dataset. Datasets were collected over 1000 on/off cycles and related 18 features were calculated for each operating condition. Different parameters (features, neuron numbers, and input signals) were investigated and performance metrics such as accuracy, sensitivity, and precision were calculated comparatively. According to the results, ANN with three layers (input-hidden-output) and the number of neurons 12-9-7, respectively, show the best performance. In addition, the highest accuracy value (%97.1) is obtained when the motor current and motor voltage are taken together as the input signal. Consequently, it is observed that the ANN structure is a useful AI tool in fault detection on the PSD system.