Keywords: Hybrid Systems, Linear Systems, Nonlinear Systems
Abstract: The paper aims to present and prove a general result regarding the usage of Lyapunov functions in testing the global exponential stability of equilibrium {0} for discrete-time arbitrary switching linear systems. The generality consists, on the one hand, in the possibility of working with both mode-dependent and mode-independent Lyapunov functions, and on the other hand, in the diversity of available concrete expressions (weighted vector norms). The flexible mathematical formulation allows the retrieval of several stability criteria that had been reported by separate articles, as various particular cases included in the generality of our new result. From this general result, individual criteria are derived by considering extended state-space representations for the description of systems dynamics, as well as several types of absolute norms in the definition of Lyapunov functions.

Keywords: Linear Systems, Industrial Applications, Control Systems Design
Abstract: It is considered the problem of stabilizing the torsional vibrations of a distributed flexible beam using boundary control. A Lyapunov functional obtained by modifying the energy functional is introduced, thus allowing to obtain exponential stability under quite mildly restrictive conditions. The bounded peaking problem is also discussed within the same framework. The paper ends by pointing other research problems which might be revisited within the aforementioned Lyapunov framework.

Keywords: Control Systems Design, System Identification and Modeling, Linear Systems
Abstract: In this paper the design and validation of ten control system (CS) structures for the cart position control of pendulum-cart systems (PCSs) are presented. In the first step the derivation of a Tensor Product (TP) model and four linear models for the PCS is conducted. In the second step, for each model a Single Input-Single Output CS structure is designed. The first CS structure includes a TP-based controller, and the other four CS structures include state feedback controllers. All controllers are tuned using the Linear Matrix Inequality & Parallel Distributed Compensation technique aiming the same system performance specifications. The five CS structures do not ensure zero steady-state control error; therefore they are included in cascade control system structures with Proportional-Integral controllers in the outer control loops. The same experimental scenario is used to test all CSs. Finally, a comparative analysis is conducted by computing five performance indices, namely the absolute value of the control error, the mean square error, the mean square control effort, the settling time and the overshoot.

Keywords: Computer Architectures, Computer Science Education and Training, Machine Learning
Abstract: This paper presents a systolic array architecture for General Matrix Multiplication. The system was designed and verified using the Verilog description language. The architecture was constructed for educational use aiming to complement the practical activities of Computer Architecture classes. The proposed solution can be used as a design space exploration tool for evaluation of matrix multiplication accelerators based on non-stationary systolic arrays. Both the width and format of the processed operands as well as the number of stages of the systolic array’s Processing Elements can be customized. As a case study, the design and performance of a systolic array for accelerating the multiplication of square matrices of signed, 8-bit integers, is presented. The overall architecture was synthesized for the Altera DE2 FPGA board in order to evaluate its performance.

Keywords: Automatic Control Education and Training
Abstract: This paper reports an easily reproducible experiment that can be used in class to illustrate fundamental concepts on Multi-Agent Systems (MAS) via human-robot interactions. In this context, the primary objective of this paper is to provide teaching resources to motivate students taking engineering courses on Robotics and MAS control that involve humans in the loop interactions. A leader-follower approach with a human being as a leader and a 4-degree of freedom robotic manipulator as a follower is proposed. An illustrative video is provided to highlight the first experimental results. Furthermore, the proposed experiments are further intended to popularize Robotics and Control to broad public during Open Days events in universities.

Keywords: Computer Science Education and Training, Software Engineering
Abstract: The effects of peer assessment on students’ learning experience over time are scarcely examined in the literature. The current research performs a longitudinal study that explores the use of LearnEval, a comprehensive peer assessment platform, in two computer science courses in two consecutive years. Both courses followed a Project-Based Learning (PBL) approach, allowing a peer assessment component to be easily integrated. In 2020-2021 academic year, 78 students participated in the process in a 3rd year Web Applications Design course. In the following year, a subset of the same students (54) participated in the activity in a 4th year Human-Computer Interaction course. A quantitative analysis of the outcomes in the two years was performed and the results were compared. Furthermore, the LearnEval usability was assessed. Therefore, the current work addresses two research questions: (1) How do the outcomes of the peer assessment activity in the 4th year course compare to the previous year when the students attended the process for the first time? (2) How did the students perceive LearnEval usability in the two years? The main findings show that the results were slightly improved in the second year of attending the peer assessment process in terms of involvement, reviewing skills, grades assigned and quality of the feedback provided.

Keywords: Control Systems Design, Predictive Control, Nonlinear Systems
Abstract: This research study centers around the development of a nonlinear model predictive control (NMPC) model based on the special orthogonal group of dimension 3, mathrm{SO(3)}, to achieve precise tracking of a desired satellite orientation in the presence of exogenous disturbances. Furthermore, this approach aims to optimize energy and fuel consumption by effectively minimizing the tracking and configuration error functions and the control input, promoting sustainability in satellite missions. The critical contribution of our approach lies in integrating the mathrm{SO(3)} framework, which enables the construction of a comprehensive control model. This integration allows for an accurate representation of the satellite's orientation dynamics, avoiding singularities and enhancing the precision and robustness of the control system. Through extensive simulations, we provide compelling evidence of the effectiveness and practical viability of the proposed approach.

Keywords: Instrumentation, Signal Processing, Embedded Systems
Abstract: This paper presents a new approach for improving the attitude determination precision of small satellite systems by filtering the data acquired from their MEMS inertial sensors. The growth in the global space sector requires innovative solutions that can guarantee the successful integration of emerging technologies. While miniaturization may appear ideal for space missions because of the reduction in dimension and cost, the enhanced sensitivity to external factors usually compromises the accuracy of navigation information. The core of this research lies in finding a way to preserve the valuable attitude information from the noise-corrupted signals provided by a cluster of MEMS gyros. The proposed method for processing the perturbed signals was the powerful tool of wavelet multiresolution analysis (WMRA). The procedure was software implemented with the help of MATLAB apps Simulink and Wavelet Analyzer which ensured the implementation of a gyro model, an attitude determination algorithm and the wavelet decomposition of the simulated signals. To fully validate the use of wavelets in enhancing attitude estimation, the study was conducted in two distinct phases: local and global validation. The results of the simulations showed the effectiveness of each wavelet function and level of decomposition for noise reduction and attitude angles accuracy improvement.

Keywords: Automotive Control Systems, Adaptive Control, Signal Processing
Abstract: This paper addresses the relative orbit control problem and proposes an adaptive control approach using the estimation of the orbital parameters. The control architecture consists of three derivative-free model reference adaptive control systems and a Kalman filter to estimate the parameters of the elliptical orbit. The design also involves PD controllers, reduced-order observers, second-order reference models, and Lyapunov equations. The stability of the three control systems is briefly analyzed via the Lyapunov theory. The carried out numerical simulations prove that the rendezvous mission is successfully accomplished, the designed control architecture being able to accurately drive the chaser vehicle to the imposed position, as well to cancel the position, speed, and acceleration errors.

Keywords: Nonlinear Systems, Robust Control, Control Systems Design
Abstract: This paper presents a cascaded control structure based on a sliding-mode controller for the trajectory tracking control of a vapor compression cycle. For this purpose, an existing high-order system model is split into a linear as well as a nonlinear subsystem and both are described by reduced-order models. The linear outer subsystem is controlled by an output feedback control in conjunction with a disturbance observer. The nonlinear subsystem, however, is controlled via a sliding-mode controller to address model uncertainty. The stability of the internal dynamics of the input-output linearisation, which serves as the basis of the sliding-mode controller, is ensured by input constraints w.r.t. the outer control-loop. The proposed control structure is tested on the full high-order simulation model and compared to a previously presented controller based on input-output linearisation. The comparison indicated a significant improvement of control performance as well as robustness concerning disturbances and model uncertainty.

Keywords: Nonlinear Systems, Control Systems Design, System Identification and Modeling
Abstract: High gain observers are often used for the real-time estimation of the state of nonlinear systems. Several design methods are based on normal forms, which are based on differential geometric considerations. The embedding into a higher dimensional state could be advantageous regarding existence conditions and convergence. However, computation and implementation of such embedding observers is uasually significantly more challenging. In this paper, we circumvent some of these problems by employing machine learning. The required functions for the implementation of the observer are approximated by neural networks.

Keywords: Manufacturing Systems, Industrial Applications, Nonlinear Systems
Abstract: This paper deals with an input-output linearization for a nonlinear ice clamping system based on a Peltier element used in an innovative manufacturing system. For this model-based design, a third-order state-space model is derived based on physical principles, and an input-output linearization is proposed with an asymptotically stable internal dynamics. Accordingly, a robust sliding-mode tracking control is designed to accurately track a desired cold side temperature even in the presence of disturbances. Unknown states as well as disturbances are estimated by a gain-scheduled nonlinear state and disturbance observer that can be assigned a linear error dynamics. Meaningful simulation results are shown that point out the effectiveness of the proposed control approach.

Keywords: Industrial Applications, Robotics, Nonlinear Systems
Abstract: This paper shows the Lagrangian-based modelling of a fully actuated three-axis robot and a control application using a variant of Super Twisting Sliding Mode Control (STSMC) in which the robot is subject to disturbances and model uncertainties to demonstrate the robustness of the proposed method. The stability of the control system is proven with the help of the direct Lyapunov method. In particular, the robustness of the results is validated through computer simulations in which impulsive disturbances are simulated together with a model mismatch due to an uncompensated mechanical friction.

Keywords: System Identification and Modeling, Nonlinear Systems, Signal Processing
Abstract: This paper proposes a robust extended Kalman smoothing approach in the case that there is a modeling mismatch between the given nominal nonlinear state-space model and the actual one. The resulting algorithm follows a linearization process and takes the form of a robust fixed-lag smoother. At each time step the computation of the smoothing gain of the linearized model is required. The latter is derived from a dynamic game: one player selects the least favorable model in a prescribed ``ball'' centered about the nominal linearized model in Kullback-Leibler (KL) divergence topology, and the other selects the optimal smoothing gain. Finally, some numerical experiments show the effectiveness of the proposed algorithm.

Keywords: Automatic Control Education and Training
Abstract: This paper focuses on the Workshop ``(Re)Creative Mobile Robotics for Kids -- (Re)CreativeRobot'' for popularizing Control and Mobile Robotics for children. Inspired from the ``Girls in Control'' workshop, this education activity allows children to individually implement basic Control algorithms on mobile robots via Scratch programming. Its first occurrence was held during the French science festival ``Fête de la Science'' 2022 edition. A second occurrence was held in a primary school in the Paris region, allowing small groups of kids to collaboratively develop Scratch algorithms for mobile robots. This paper shows some insights and perspectives of this perennial interactive Control Education activity.

Keywords: Software Engineering, Computer Science Education and Training, Theory of Algorithms
Abstract: The present work aims to find computationally-efficient models for solving discretized partial differential equations. To accomplish that, we implement and compare the performance of a series of algorithmic models, both sequential and parallelized (using OpenMP libraries), which may be used for solving convergence problems. For demonstrating the results upon a real-world case scenario, the work introduces a CFD (computational fluid dynamics) Poisson process, which is numerically discretized and approximated using Jacobi and Gauss-Seidel algorithms. Performance of the parallelized implementations will be measured against a baseline, and the consequent model optimizations will be described. Finally, the scalability of the parallelized algorithms will be compared together, with regards to CPU thread count and memory handling. The results show that, compared with a sequential version, a parallelized Jacobi algorithm shows good performance and scalability when introducing more threads. At the same time, a parallelized version of the Gauss-Seidel algorithm did not present substantial increases in performance over its sequential counterpart.

Keywords: Computer Science Education and Training
Abstract: Correlating the education and training system with the evolution of today's technologies is a complex but necessary process for training, improvement, and lifelong learning. Digital technologies are present today in all fields, and have become increasingly necessary in higher education institutions, not only to attract students, to bring them into a familiar environment, but also to initiate and improve their skills in what is an evolution in science. The digitalization of education services is necessary to have a more accessible, interactive, and personalized educational process and this also involves the use of digital technologies, elements that contribute to people's self-education, and education outside of schools’ walls. In this paper we present aspects related to the digitalization of education, both from the perspective of educators and learners. It is a study carried out in two Romanian universities, the University of Craiova and the “Constantin Brâncuși” University of Târgu Jiu. The purpose of this work is to emphasize the necessity of digitalization, its importance, but also to emphasize the stage in which the two universities are now, in relation to what is meant by digitalization, who should be involved in the digitalization of the institution, what activities should be digitalized, knowledge of the digital signature concept or whether respondents use machine learning tools such as ChatGPT, or others. The results of this study reflect the opinion of the teachers and students of the two universities.

Keywords: Virtual and Augmented Reality, Computer Science Education and Training, Human - Computer Interface
Abstract: The integration of technology in education has provided a more interactive and engaging experience through the creation of Serious Games. By using 3D virtual environments and Virtual Reality, interactive experiences which simulate real-life situations and environments can be created. The development process of such applications often proves to be slow and cumbersome. However, the addition of a trigger system or scenario editor could enable educators to easily create scenarios for students without requiring much programming knowledge. Designing the trigger system should be based on the principles of reusability, applicability and extensibility. This paper focuses on the creation of such a trigger system and its uses throughout the development of a 3D VR Serious game called Visit UPT. This game enables students to explore the university campus and simulate small scientific experiments.

Keywords: Embedded Systems, Digital Design
Abstract: Understanding spoken language evolution can offer important insight that is needed to devise intelligent systems with speech capabilities. However, existing work mostly focuses on mathematical and software models that are insensitive to naturalistic conditions, even though such conditions are essential in the evolution of real languages. This paper discusses the implementation of Talking Tiny Cognitive Architecture proposed to study language evolution in more naturalistic conditions. The architecture can generate new words that are verbally communicated to describe environment conditions. It also attempts to understand and learn words verbally communicated by another architecture. Experiments describe the performance of the implementation as an embedded system.

Keywords: Automotive Control Systems, Nonlinear Systems
Abstract: This work tackles the relative orbital control problem and presents the development of an H∞ controller in terms of the Riccati algebraic matrix equations. The nonlinear Tschauner-Hempel equations associated with rendezvous missions involving elliptical orbits are easily transformed into a linear time-varying dynamics, the resulting optimal control problem being solved by designing an H∞ robust controller that guarantees the asymptotical stability regardless of the uncertainties and existence conditions. The theoretical validation of the novel controller is accompanied by simulations proving the successful tracking of the target-chaser relative position and the cancel of the relative position and speed errors.

Keywords: Automotive Control Systems, Adaptive Control, Nonlinear Systems
Abstract: This work deals with development of a software and hardware technological platform, with applications to the control of the launch vehicles. The software and hardware platform has two main experimental modules: 1) a virtual model of the launch vehicle, implemented as software simulator in a dedicated computer having communication capabilities; 2) a control unit (single-board computer), with specific hardware and dedicated software, which implements the control laws based on dynamic inversion technique and neural networks. The validation of the experimental platform was performed both by numerical simulations and tests in lab conditions as MIL – Model in the Loop (the space vehicle simulator as well as the control system run on the same computer) and HIL – Hardware in the Loop (the control system is embedded in an experimental control unit that communicates with the simulator.

Keywords: Machine Learning, Real Time Applications, Software Engineering
Abstract: This paper discusses a model for detecting burned areas based on a time series of multi-spectral optical images. The model employs a two-class Gaussian Naive-Bayes classification algorithm on a set of literature proven burned area indexes, e.g. Mid-Infrared Burned Area Index. The area of study, training and validation mainly comprises geographically diverse sites, scattered in and around the Romanian territory; Nevertheless, several over the seas sites are included, e.g. Evia Greece. We show that the model provides fast and reliable detection of burned areas. despite its computational simplicity.

Keywords: Software Engineering, Other Topics, Modeling, Simulation and CAD Tools
Abstract: The development of CubeSat technology provided the possibility for universities to design and launch small satellites in Earth’s orbit in order to run scientific experiments. Typically, CubeSats are small satellites, in the form of a cube with a side of 10cm and a weight of about 1 kg and don’t normally have their own propulsion. Once placed in orbit around our planet, usually at altitudes under 700 km, their trajectory is influenced by various external forces, one being the effect of the solar radiation reflected by Earth called Albedo perturbation. While often ignored for larger satellites due to their small value compared to other forces, the albedo perturbation has a greater effect on CubeSats because they are in closer proximity to Earth. This paper presents an early version of a software component that computes the effect of the Albedo perturbation on small CubeSats on different Low Earth Orbits.

Keywords: Linear Systems, Optimization
Abstract: This paper proposes a Distributed Moving Horizon Estimation (DMHE) with an Extended Kalman Filter (EKF)-based pre-estimation to solve the constrained cooperative localization problem for a Multi-Agent System (MAS) using nonlinear measurements. The proposed DMHE strategy uses a fused arrival cost obtained by a consensus among neighbors to efficiently spread the relevant estimation information across the communication network. The EKF pre-estimation enables to reduce the number of optimization variables and, thus, the computation time of the constrained nonlinear optimization problem over the horizon length, while preserving the accuracy of the estimation. A simulation case study of cooperative localization of a fleet of Unmanned Aerial Vehicles (UAVs) is proposed. Comparison with existing distributed estimation methods is carried out to confirm the effectiveness of the proposed DMHE algorithm in terms of estimation accuracy, computation time, and constraints handling.

Keywords: Computer Vision, Machine Learning, Other Topics
Abstract: This paper investigates the impact of incorporating Optical Character Recognition (OCR) information into object detection models for extracting key information fields from invoices. We propose a method that adds a fourth channel to the input images, representing text presence, derived from two OCR models. Our experiments show that while larger models do not benefit from the additional text localization information, smaller models exhibit significant accuracy improvements and accelerated learning. In particular, we observe a substantially higher mean average precision (mAP) by epoch 20 out of 100 when including the fourth OCR information channel. This research demonstrates the potential benefits of incorporating OCR information into object detection models, particularly for smaller models with limited resources, by enhancing not only accuracy but also the speed of convergence during training.

Keywords: Computer Vision, Machine Learning
Abstract: Person re-identification (Re-ID) has attracted considerable attention within the field of biometrics in recent times, driven by an escalating demand in public communities, marketing and banking endeavors, and social media networks. Moreover, the rapid advancement of deep learning has introduced a wide range of effective tools for addressing person re-identification tasks. The result is more precise and faster processing compared to the video surveillance performed manually by human operators. This article provides an overview of recent Re-ID studies by highlighting the challenges and issues, proposing a taxonomy of Re-ID methods, and comparing the performances of the most representative deep learning-based approaches.

Keywords: Computer Vision, Optimization, Intelligent Systems
Abstract: Representing relevant information from volume data sets is a problem often faced in visualization. Generating meaningful images from highly-complex volume data sets is a challenging, tedious task requiring specialized knowledge of the distribution and properties of the data. Traditionally, this task has been carried out manually via specialized user interfaces. We propose a volume visualization pipeline which facilitates the automatic generation of high-quality images from volume data sets. Our method involves a direct volume renderer which generates images from volume data based on visual mappings provided by a transfer function. Central to our approach is a quality-focused descriptor which exploits the properties of the distribution of gradient orientations of an alpha-bounded surface within the volume. This feature is useful for determining transfer functions that result in the rendering of corresponding images depicting various details from the volume. We show that by using this feature as an optimization objective, the generation of high quality images can be automated. Using simple genetic algorithms, we can automatically generate sets of images illustrating coherent, easily-distinguishable and high-quality surfaces of relevant structures from volume data.

Keywords: Biomedical Engineering, Computer Vision, Machine Learning
Abstract: People with dietary restrictions caused by certain gastrointestinal pathologies and by certain physiological conditions need permanent monitoring of their food intake. Besides this, they are also required to meet certain nutritional values during the day for keeping themselves healthy. Hence, many such diagnosed patients use mobile apps and other intelligent systems for food detection, tracking, and nutritional value intake. We propose an intelligent mobile application that uses a Deep Neural Network (DNN) for on-device food classification. Therefore, we protect the user's privacy and never send the images taken by the user to the cloud server (e.g., conform with GDPR regulations). Our results show over 70% accuracy on Food101 dataset using two state-of-the-art DNNs, i.e., ResNet50 and EfficientNet-b0, the latter one being more suitable for on-device deployment due to it's reduced complexity. The proposed mobile application is a simple yet effective food monitoring app with minimal input from the user which can suggest avoiding and/or help the user better estimate the nutritional values of the served dishes.

Keywords: Machine Learning, Computer Vision
Abstract: In this paper, we approach the important topic of video surveillance systems which automatically identify and track moving vehicles in different environments. A comparative evaluation has been performed on four algorithms which have been adapted for tracking and counting vehicles passing through a line, using both contour detection and feature detection techniques over a region of interest in the input video frames. YOLO and Haar Cascades were also considered in view of the efficiency comparison of Machine Learning algorithms. The proposed algorithms have been implemented and tested on a set of video sequences and an accuracy measure was computed to determine the most effective algorithm. The results show that the YOLO-based algorithm has the best average accuracy, while Algorithm 1, which is based on frame difference moving object detection and tracking, with an initial calibration phase, has the most consistent performance.

Keywords: Control Systems Design, Linear Systems, Other Topics
Abstract: The focus of this paper is to control a distributed parameter cracking process by utilizing state and error feedback regulators. The main aim is to track a desired output while dealing with disturbances originating from a distributed parameter exosystem. At first, a stabilizing regulator with state feedback is developed to drive the process output towards the reference trajectory. The second goal is to develop a dynamic controller that takes the tracking error as input, for which it is proven that the closed-loop system is exponentially stable and that the tracking error gradually approaches zero over time. The effectiveness of the regulators is demonstrated through numerical simulations.

Keywords: Control Systems Design, Optimization, Linear Systems
Abstract: In this paper, we study linear quadratic regulator (LQR) design subject to linear equality constraints in the controller parameters. Necessary solvability conditions are provided, and a method for choosing the weighting matrices in the quadratic objective function minimized by the constrained LQR is presented. To this end, the problem at hand is transformed into a set of polynomial inequalities that can be solved using Bernstein polynomials. We explicitly show how the requirement of input-output decoupling can be transformed into a set of linear equations in the controller parameters. All control structures that can be transformed into a set of linear equality constraints, e.g. output feedback control, decentralized control, or combinations thereof, can be determined with our method. We demonstrate the proposed method by designing structured optimal controllers for a three-tank system.

Keywords: Control Systems Design, Linear Systems
Abstract: A series of recent works have highlighted the interest of multiplicity varieties in the characterization of the exponential decay rate for the solution of linear dynamical systems represented by delayed differential equations. In fact, it has been shown in the case of a single delay that a sufficiently high multiplicity spectral value tends to be dominant, in what is now known as the multiplicity-induced-dominancy property (MID). Despite the many existing results in such a configuration, only a few concern the case of multiple delays. Through this work, we propose a first characterization of the MID property for second-order systems controlled by a two-delay "block". As an application of the results obtained, we consider the problem of stabilization of the classical pendulum with exclusive access to the delayed position.

Keywords: Real Time Applications, Industrial Applications, Control Systems Design
Abstract: Real-time (RT) implementation of control systems for non-linear processes raises numerous theoretical and practical challenges. Multi-model structures (MM) represent one of the successful solutions for these processes. However, switching of control algorithms constitutes the specific problem of MM structures. Even if reducing the sampling period can increase the precision in the control of some processes, for the case of algorithms switching in the MM structures is not improving significantly. The tests carried out in this work confirm this behavior for some numeric PIDs algorithms. Moreover, these tests can be an example for reducing the computational effort by changing the sampling period up to a maximum limit, specific to the process. However, superior performances in switching control algorithms can be obtained by using a common database for all switched algorithms in the respective MM structure. The architecture proposed in the paper and the tests performed on a laboratory implementation prove these solutions.

Keywords: Biomedical Engineering, Nonlinear Systems, Control Systems Design
Abstract: The main objective of this paper is to design a suitable control strategy which solves the the so-called Artificial Pancreas Problem, which affects patients with Type 1 Diabetes Mellitus. The theoretical background consists in using the exact feedback linearization technique. Using a suitable change of coordinates paired with a designed input signal create a linear map between input and output which is further used to design a control law. Moreover, the input is computed to perform a disturbance decoupling. The resulting linear model is used within the optimal control approach. Conclusively is presented a comparison between a patient using the open loop control and a patient with the present control law.

Keywords: Agent - Based Systems, Distributed Systems, Networked Control
Abstract: The Hegselmann--Krause (HK) model belongs to the class of bounded confidence opinion dynamics. The concepts of iterative averaging and homophily that characterize the HK model can be used for the design of synchronization strategies in multi-agent systems. In this paper, the application of this technique for the synchronization of an electronic circuit is proposed. The experimental setup consists of electronic integrators, representing the agents of the model, which interact through a microcontroller that implements the iterative averaging process. Similarly to the agents in the HK model, two integrators interact if the difference between their output voltages, i.e., the agents' opinions, does not exceed a constant confidence bound. Experimental results show transient and steady-state behaviors of the system and sensitivity to confidence thresholds, by also discussing on the robustness of corresponding numerical solutions. Experiments confirm the typical phenomenon of decreasing in the number of clusters and convergence time by increasing the confidence thresholds.

Keywords: Adaptive Control, System Identification and Modeling, Nonlinear Systems
Abstract: When working with real world systems, their usually nonlinear behavior results in great challenges. Hence, effects like hysteresis and friction are often not or only partly modeled. This causes the need for suitable system identification methods that result in analytical models that are easy to analyze and interpret with known control methods. Moreover, robust controllers are needed to guarantee a sufficient performance at any working point. Hence, this work uses neural networks in a simplified additive nonlinear autoregressive exogenous (SANARX) structure. Those can be reformulated in a linear parameter-varying model. Afterwards, H∞ synthesis is used to get a robust gain-scheduled controller. The suitability of this method is evaluated on a hydraulic benchmark system.

Keywords: Adaptive Control, Linear Systems, Robust Control
Abstract: This paper presents a novel approach to compensate periodic perturbations that occur in a vast variety of production processes. Typically, these perturbations interrupt the steady-state motion of mechanical systems, resulting in production inaccuracies. To address that problem, an adaptive torque feedforward algorithm is employed to systematically reduce unknown perturbations affects on the mechanical system. The algorithm is inherent stable and purely based on stochastic values. Therefore, the adaptation process uses only gradient-free optimization methods. All presented results can be, without any means, incorporated into a state-of-the-art programmable logic controller. Correctness of the stated algorithm is verified by simulation on the widely used two non-rigid coupled rotational masses spring-damper system corrupted by periodic perturbations on the load-end.

Keywords: Signal Processing, Digital Design, Modeling, Simulation and CAD Tools
Abstract: The paper analyzes the efficiency of fixed-point implementation and the effect of quantization of coefficients and signals in the implementation of numerical filters on the STM32 Nucleo-64P development board. Discretization methods are also analyzed. By modelling the effects of quantization, it is possible to indicate how the system responds. Proper quantization can increase the performance. We model the quantization error as stochastic noise. The results show that the quantization of the coefficients and the fixed-point processing minimally change the response of the digital filter compared to the analog one, thus achieving very good results at high sampling frequencies. Also, through this analysis, the most efficient implementation can be chosen, taking into account the system and the characteristics of the development board.

Keywords: Modeling, Simulation and CAD Tools, Nonlinear Systems
Abstract: This paper deals with an analysis regarding the effect of Sewer Network operation on the performances of the integrated wastewater collecting and treatment system. The analysis is done by considering various control actions for the output storage tank of the sewer network, comparing a series of performance indicators defined for both subsystems with their “no control” value. Two cases have been considered regarding the control actions for the output storage tank of the sewer network: 1. keeping the output valve at a constant opening and 2. keeping the output valve at the maximum opening possible such as the outflow is limited to a constant value. While the second case did not prove advantageous, the first one showed that lowering the sewer network output valve opening would improve the integrated system performances.