In German Language: Dieses Buch stellt die modernen technologischen Fortschritte und Revolutionen im biomedizinischen Sektor vor. Fortschritte in der zeitgenössischen Sensorik, dem Internet der Dinge (IoT) und bei Maschinenlernalgorithmen und -architekturen haben neue Ansätze im mobilen Gesundheitswesen eingeführt. Eine kontinuierliche Beobachtung von Patienten mit kritischer Gesundheitssituation ist erforderlich. Sie ermöglicht die Überwachung ihres Gesundheitszustandes während alltäglicher Aktivitäten wie Sport, Gehen und Schlafen. Dank moderner IoT-Rahmenbedingungen und drahtloser biomedizinischer Implantate, wie Smartphones, Smartwatches und Gürtel, ist dies realisierbar. Solche Lösungen befinden sich derzeit in der Entwicklung und in Testphasen durch Gesundheits- und Regierungsinstitutionen, Forschungslabore und biomedizinische Unternehmen. Die biomedizinischen Signale wie Elektrokardiogramm (EKG), Elektroenzephalogramm (EEG), Elektromyographie (EMG), Phonokardiogramm (PCG), bei chronisch-obstruktiver Lungenkrankheit (COP) und Elektrookulographie (EoG), Photoplethysmographie (PPG), Positronenemissionstomographie (PET), Magnetresonanztomographie (MRI) und Computertomographie (CT) werden nicht-invasiv erfasst, gemessen und über die biomedizinischen Sensoren und Gadgets verarbeitet. Diese Signale und Bilder repräsentieren die Aktivitäten und Zustände des menschlichen kardiovaskulären, neuralen, visuellen und zerebralen Systems. Eine Mehrkanalerfassung dieser Signale und Bilder mit einer angemessenen Granularität ist für eine effektive Überwachung und Diagnose erforderlich. Sie erzeugt ein großes Datenvolumen, und seine Analyse ist manuell nicht machbar. Daher sind automatisierte Gesundheitssysteme in der Entwicklung. Diese Systeme basieren hauptsächlich auf der Erfassung und Sensorik von biomedizinischen Signalen und Bildern, Vorverarbeitung, Merkmalsextraktion und Klassifizierungsstufen. Die zeitgenössischen biomedizinischen Signal-Sensorik, Vorverarbeitung, Merkmalsextraktion und intelligente maschinelle und tiefgreifende Lernalgorithmen für die Klassifizierung werden beschrieben.

This study improves the mathematical modeling for the potential of atmospheric water harvesting (AWH) using desiccant materials. This research is crucial in highlighting the sustainable water management strategies of AWH systems in converting atmospheric moisture into a vital water source. The primary focus of our research methodology was the analytical derivation of sorption isotherms, essential for the hygrothermal simulation of desiccant materials. This was accomplished through the application of two established models, namely, the Guggenheim-Anderson-de Boer (GAB) and Van Genuchten (VG). Experimental data on various anhydrous salts from existing literature have been used. An in-depth comparative analysis of these models reveals that the VG model aligns more closely with the experimental data, thus asserting its superiority in enhancing the selection and efficiency of desiccant materials in AWH systems. By confirming the VG model’s superiority in accurately modeling sorption isotherms, our research not only improves the model of AWH systems but also, importantly, contributes to the development of advanced water harvesting technologies.

The prognosis of organic light-emitting diodes (OLEDs) not only requires early detection of a bearing defect, but also the capability to predict their life data under all operational scenarios. The use of sophisticated machine learning (ML) algorithms is undoubtedly becoming an increasingly exciting research direction, as these algorithms can yield high predictive models with minimal domain expertise. The central question of this perspective is: how well can ML models advance our ability to forecast the lifetime of OLEDs compared to the physics-based models? In this paper, data-driven methods, feed-forward neural networks (FFNN), support vector machines (SVMs), k-nearest neighbors (KNNs), partial least squares regression (PLSR), and decision trees (DTs), are used to predict the lifetime and reliability of OLEDs through analyzing the lumen degradation data collected from the accelerated lifetime test. The final predicted results indicate that both the data-driven and our physics-based OLED lifetime models fit well the experimental data. The main drawback of the former method is that their efficacy is highly contingent on the quantity and quality of the operational dataset. Among all these methods, much more reliability information (time to failure) and the highest prediction accuracy can be achieved by FFNN.The prognosis of organic light-emitting diodes (OLEDs) not only requires early detection of a bearing defect, but also the capability to predict their life data under all operational scenarios. The use of sophisticated machine learning (ML) algorithms is undoubtedly becoming an increasingly exciting research direction, as these algorithms can yield high predictive models with minimal domain expertise. The central question of this perspective is: how well can ML models advance our ability to forecast the lifetime of OLEDs compared to the physics-based models? In this paper, data-driven methods, feed-forward neural networks (FFNN), support vector machines (SVMs), k-nearest neighbors (KNNs), partial least squares regression (PLSR), and decision trees (DTs), are used to predict the lifetime and reliability of OLEDs through analyzing the lumen degradation data collected from the accelerated lifetime test. The final predicted results indicate that both the data-driven and our physics-based OLED lifetime models fit well the experimental data. The main drawback of the former method is that their efficacy is highly contingent on the quantity and quality of the operational dataset. Among all these methods, much more reliability information (time to failure) and the highest prediction accuracy can be achieved by FFNN.

As much as the development of smart grid systems, accurate and reliable solar irradiance is crucial. In smart grid systems, forecasting solar irradiance remains a vital solution to mitigate the best management of energy.This study presents a new approach using a hybrid deep learning model, Conv-LSTM-AE, enhanced by Bayesian algorithm optimization. In four seasons of the years, the model was evaluated based on the accuracy using different metrices errors, and time computing in training. The model exhibits exceptional adeptness as compared to other deep learning models, showing that the Conv-LSTM-AE with Bayesian optimization is a standout performer with remarkable accuracy. This development represents a substantial step towards utilizing state-of-the-art technology for the efficient use of solar power within intelligent grid systems.

The increasing deployment and exploitation of distributed renewable energy source (DRES) units and battery energy storage systems (BESS) in DC microgrids lead to a promising research field currently. Individual DRES and BESS controllers can operate as grid-forming (GFM) or grid-feeding (GFE) units independently, depending on the microgrid operational requirements. In standalone mode, at least one controller should operate as a GFM unit. In grid-connected mode, all the controllers may operate as GFE units. This article proposes a consensus-based energy management system based upon Model Predictive Control (MPC) for DRES and BESS individual controllers to operate in both configurations (GFM or GFE). Energy management system determines the mode of power flow based on the amount of generated power, load power, solar irradiance, wind speed, rated power of every DG, and state of charge (SOC) of BESS. Based on selection of power flow mode, the role of DRES and BESS individual controllers to operate as GFM or GFE units, is decided. MPC hybrid cost function with auto-tuning weighing factors will enable DRES and BESS converters to switch between GFM and GFE. In this paper, a single hybrid cost function has been proposed for both GFM and GFE. The performance of the proposed energy management system has been validated on an EU low voltage benchmark DC microgrid by MATLAB/SIMULINK simulation and also compared with Proportional Integral (PI) & Sliding Mode Control (SMC) technique. It has been noted that as compared to PI & SMC, MPC technique exhibits settling time of less than 1μsec and 5% overshoot.

Cancer is distinguished by the presence of abnormal cellular proliferation and growth, both of which serve as signs and symptoms for this kind of illness. Computer vision, deep learning, and metaheuristics optimization techniques are increasingly important for solving complex medical Artificial Intelligence (AI) problems such as cancer detection. This paper introduces a new methodology for training the Multi-Layer Perceptron (MLP) using the optimization algorithm known as Hunger Games search Optimization technique (HGO) and apply this method to classify the cervical cancer. The main goal of this method is to reduce the error and enhance the classification rate of cervical cancer. The outcomes show that the MLP with HGO algorithm performed better than other algorithms in terms of classification efficacy and accuracy rate. Simulation outcomes indicate that the proposed strategy performs better than previously published research in terms of effectiveness for the classification optimization methods.

With the increasing complexity and scale of digital VLSI designs, ensuring reliability in IC design necessitates effective fault detection processes during the pre-silicon stage. Many fault detection algorithms lead to significant computational time due to the problem of search space explosion. To handle the ever-growing volume of data, deep learning algorithms, a subset of machine learning techniques, can be employed. In this paper, we propose two novel approaches for fault detection (FD) of digital VLSI circuits, specifically targeting stuck-at faults. The first proposed model is semi-supervised FD model that aims to mitigate the search space explosion issue by leveraging both unsupervised and supervised learning processes. The second presented model is based on an optimizer for finding the appropriate configurations for detecting stuck-at faults in digital circuits. The initial proposed model achieves maximum validation accuracy of approximately 98% applied to circuits from ISCAS’85. This model yields a higher accuracy compared to the second approach that achieves maximum accuracy of around 95% when applied to the same circuits from the ISCAS'85 benchmark.

Each photovoltaic (PV) system demonstrates a specific point on the current-voltage (IV) curve characteristics where the power generated reaches its maximum value which is referred to as the maximum power point (MPP). The maximum power point tracking (MPPT) system is designed to sample the output of solar cells and adjust the load in order to achieve maximum power output under various environmental conditions. In this paper, a design of a boost converter with MPPT based on a PID controller is proposed. The goal is to optimize the extraction of the highest attainable power from photovoltaic systems. The extracted power is subsequently directed to the load through a boost converter, which elevates the voltage to the necessary level. The PID controller is calibrated by Neural Gas Network (NGN) and utilized to achieve a consistent output voltage irrespective of fluctuations in the power source or the connected load. Moreover, an NGN technique is applied to generate the optimal PID parameters before using the PID. The comprehensive system design depicted has been simulated using MATLAB Simulink to verify the functionality of the system.

Due to the explosive demand and rapid development of the electronics industry, the integration and complexity of electronic devices have significantly grown. Fault detection has played a vital role in identification of faults in electronic circuits, ensuring normal operation and reliability of systems. Given traditional methods of fault detection are often inaccurate and time-consuming, artificial intelligence techniques are a growing interest for researchers of this field. In this paper, various artificial intelligence-based fault detection techniques in analog and digital VLSI circuits are described. In addition, two distinct models for fault detection of digital circuits based on deep learning are proposed. The primary goal of the first approach is utilization of a stacked sparse autoencoder to avoid the search space explosion problem. The second proposed method utilizes an optimization method for detecting the best model configuration. The proposed models deliver maximum validation accuracy of 97.7% and 95.7% respectively, implemented on digital circuit from ISCAS’85.

Wireless sensor networks (WSN) generate an enormous quantity of data, which necessitates preprocessing at the source to lower the entire amount of data collected for transmission and storage. The finite impulse response (FIR) filter is extensively utilized as a signal preprocessing phase in WSNs. The triangular common subexpressions elimination framework is suggested in which the number of logical operators (LO) and logical depths (LD) in FIR filter implementations has been significantly reduced. In the presented Triangular CSE approach, the occurrences of common Triangular subexpressions are first examined across the complete set of filter coefficients. The proposed method considerably reduces the computational burden, which is again reduced by using Vertical common expressions and Horizontal common expressions. The paper includes a detailed illustration of the algorithm and compares existing algorithms. The proposed architecture minimizes the delay units, structural adders, and adder circuits in the FIR filter's multiplier blocks (MBAs) to decrease the overall complexity of the hardware. Furthermore, the design is improved using Parallel Prefix Adder Based on Kogge-Stone Tree. The LD is not increased during the reduction in LO. Energy consumption is also investigated, along with hardware expenses. Compared to other methods, the proposed solution decreases the number of structural adders while slightly increasing the number of delay elements.