Kybernetika a robotika

Browse

Now showing 1 - 5 of 41

Simulation-Based Diagnosis for Cyber-Physical Systems - A General Approach and Case Study on a Dual Three-Phase E-Machine
(Schloss Dagstuhl – Leibniz-Zentrum für Informatik, 2024-11-26) Kaufmann, David; Kozovský, Matúš; Wotawa, Franz
This paper presents a simulation-based approach for fault diagnosis in cyber-physical systems. We utilize simulation models to generate training data for machine learning classifiers to detect faults and identify the root cause. The presented processing pipeline includes simulation model validation, training data generation, data preprocessing, and the implementation of a diagnosis method. A case study with a dual three-phase e-machine highlights the results and challenges of the simulation-based diagnosis approach. The e-machine simulation model provides a complex and robust system representation, including the capability to inject inter-turn short-circuit faults. The introduced validation procedures of the simulation model revealed limitations in signal similarity and distinguishability compared to real system behavior. Based on the discovered limitations, the overall best results are achieved by applying an Autoencoder model for anomaly detection, followed by a Random Forest classifier to identify the specific anomalies. Further, the focus is on identifying the affected e-machine phase rather than the exact number of faulty winding turns. The paper shows the challenges when applying a simulation-based diagnosis approach to time-series data and underlines the required analysis of simulation models. In addition, the flexible adaption in the diagnosis strategies enhances the efficient utilization of cyber-physical system models in fault diagnosis and root cause identification.
PMSM fault detection using unsupervised learning methods based on conditional convolution autoencoder
(IEEE, 2024-11-03) Kozovský, Matúš; Buchta, Luděk; Blaha, Petr
The challenges of fault detection and condition monitoring in powertrain systems have become increasingly prominent, particularly with the widespread adoption of failoperational systems. These systems are pivotal in diverse sectors, including the robotics, automotive industry, and various industrial applications. A critical attribute of such systems lies in their capability to identify non-standard behaviour of the system. This study describes a inovative conditional convolutional autoencoder-based fault detection algorithm for the permanent magnet synchronous motor. The study compares a train process of conditional convolutional autoencoder with a classical convolutional autoencoder. The presented autoencoder structure was designed to be implementable into the target microcontroller AURIX TC397 while providing sufficient recognition capabilities of the interturn short-circuit. Autoencoders are trained on data obtained during healthy motor operation and subsequently used to detect interturn short-circuit faults on the experimental dual three-phase permanent magnet synchronous motor with the possibility of emulating an interturn short-circuit fault. The paper provides insights into the achieved autoencoder inference times and the sensitivity in detecting the fault.
Design and Application of Neural Network for Compensation of VSI Output Voltage Nonlinearities
(IEEE, 2024-11-03) Buchta, Luděk; Kozovský, Matúš
Voltage source inverters (VSI) with modern power-switching elements are often used to control industrial AC motors. However, the non-linearities of the inverters, such as dead time, turn-on and turn-off switching delay times and voltage drops, are often behind the distortion of the phase currents of the controlled motor. The current distortions can be suppressed by appropriately calculated non-linear functions, which represent the compensation voltages and are consequently added to the control values of the current regulators in the field-oriented control (FOC) algorithm. An artificial neural network (ANN) was designed to identify the non-linear functions of the compensation voltages, which is presented in this paper. Only signals available in the FOC algorithm are used as ANN inputs. The learning process of the neural network takes place online during the running of the motor control algorithm. The learning pattern is generated in each step of the control algorithm from the control errors of the current controllers and the previous ANN outputs. It is not necessary to know the VSI parameters when learning the neural network. The proposed ANN and back-propagation learning algorithm were implemented on one core of the AURIX microcontroller TC397. The proposed strategy was validated through experiments on a real permanent magnet synchronous motor (PMSM), and experimental results prove the effectiveness of the ANN.
Online Monitoring of Interturn Short Circuit Current in PMSMs
(IEEE, 2025-03-10) Zezula, Lukáš; Blaha, Petr
This paper extends the previously published parameter estimation-based approach to interturn short circuit diagnostics in permanent magnet synchronous motors by real-time monitoring of hidden machine states after fault occurrence. The designed monitoring method relies on an adaptive formulation of the Kalman filter, which assumes interdependence between measurement and process noise variables. A variable forgetting factor not only mitigates the impact of the process model uncertainty but also facilitates the simultaneous operation of the monitoring algorithm and fault indicator estimation. Furthermore, contributions of fault current and healthy machine model to stationary reference frame currents are estimated from an advanced discrete-time motor description reflecting a stator winding arrangement inside a motor's case. The monitoring algorithm is validated in steady state, torque load transient, and velocity transient laboratory experiments with diverse fault severity values.
Parallel Computing Utilization in Nonlinear Model Predictive Control of Permanent Magnet Synchronous Motor
(IEEE, 2024-09-09) Kozubík, Michal; Veselý, Libor; Aufderheide, Eyke; Václavek, Pavel
Permanent Magnet Synchronous Motor (PMSM) drives are widely used for motion control industrial applications and electrical vehicle powertrains, where they provide a good torque-to-weight ratio and a high dynamical performance. With the increasing usage of these machines, the demands on exploiting their abilities are also growing. Usual control techniques, such as field-oriented control (FOC), need some workaround to achieve the requested behavior, e.g., field-weakening, while keeping the constraints on the stator currents. Similarly, when applying the linear model predictive control, the linearization of the torque function and defined constraints lead to a loss of essential information and sub-optimal performance. That is the reason why the application of nonlinear theory is necessary. Nonlinear Model Predictive Control (NMPC) is a promising alternative to linear control methods. However, this approach has a major drawback in its computational demands. This paper presents a novel approach to the implementation of PMSMs' NMPC. The proposed controller utilizes the native parallelism of population-based optimization methods and the supreme performance of field-programmable gate arrays to solve the nonlinear optimization problem in the time necessary for proper motor control. The paper presents the verification of the algorithm's behavior both in simulation and laboratory experiments. The proposed controller's behavior is compared to the standard control technique of FOC and linear MPC. The achieved results prove the superior quality of control performed by NMPC in comparison with FOC and LMPC. The controller was able to follow the Maximal Torque Per Ampere strategy without any supplementary algorithm, altogether with constraint handling.

Browse

Recent Submissions