Adaptive Tuning Research Articles

Adaptive Search Algorithms: A Comprehensive Overview and Emerging Optimization Trends

Adaptive search algorithms have emerged as pivotal tools for addressing complex, high-dimensional, and nonlinear challenges across various domains. This paper provides a detailed review of adaptive search techniques, including evolutionary algorithms, swarm intelligence methods, and cutting-edge hybrid models, with a unique contribution of a systematic comparison that showcases quantifiable improvements—up to 50% reduction in computational overhead and a 30% increase in solution accuracy across diverse benchmarks. It delves into key methodologies such as genetic algorithms, particle swarm optimization, and differential evolution, highlighting recent breakthroughs in adaptive parameter tuning and multi-objective optimization frameworks. The research emphasizes significant advancements in practical applications like machine learning, engineering design, and logistics, where these algorithms have improved the balance between exploration and exploitation for more optimal outcomes. Furthermore, emerging trends such as bio-inspired models and the integration of reinforcement learning and quantum-enhanced optimization are discussed, promising to reshape the adaptive search landscape by equipping it with sophisticated tools to manage the growing complexity of optimization challenges. This paper aims to map the current state and guide future directions of adaptive search algorithms, fostering the development of more robust, efficient, and adaptable optimization strategies essential for ongoing academic and practical innovations.

Experience Replay Enhances Excitation Condition of Neural-Network Adaptive Control Learning

Neural-network adaptive control offers a promising avenue for managing dynamic uncertainties with unknown structures. Nevertheless, traditional neural-network adaptive control methodologies are plagued by challenges such as persistent excitation and learning singularities. This study introduces an enhanced neural-network adaptive control methodology that optimizes the use of empirical data and ensures a more even spatial distribution of samples. A neural-network adaptive controller is developed, incorporating an experience replay strategy that refines the adaptive tuning of network weights. The adaptation is fueled by the current tracking and historical model prediction errors, markedly improving the excitation condition. Furthermore, a data- selection strategy that prioritizes regions with sparse samples is introduced, effectively mitigating the issue of learning singularities. Additionally, an adaptive technique to circumvent control saturation is employed. The stability of the control system and the convergent behavior of the network weights are substantiated through theoretical analysis. Simulation outcomes corroborate the superiority of the proposed algorithm in terms of parameter convergence, learning efficacy, and control precision.

Angstrom Confinement-Triggered Adaptive Spin State Transition of CoMn Dual Single Atoms for Efficient Singlet Oxygen Generation.

To achieve high selectivity in the transformation from peroxymonosulfate to singlet oxygen, adaptive tuning of atomic spin state as the peroxymonosulfate structure varied is crucial. The angstrom confinement can effectively tune spin state, but developing an adaptive angstrom-confined atomic system is challenging. Angstrom-confined cobalt (Co) manganese (Mn) dual single atoms within flexible 2D carbon nitride interlayer are constructed to drive adaptive tuning of spin state by changing atomic coordination under angstrom confinement. The in situ characterizations and density functional theory calculations showed that medium-spin Co in Co─N4 absorbed electrons after the adsorption of peroxymonosulfate on CoMn dual single-atom sites and then cleaved O─H of peroxymonosulfate to facilitate *SO5 generation, while the introduction of *SO5 increased interlayer distance and then cleaved Co─N and Mn─N, resulting in the spin state transition from medium to high. Subsequently, the high-spin Co and Mn in Co─N2 and Mn─N2 desorbed the *O2 from *SO5, restoring the initial medium spin state. The adaptive spin state transition enhanced 38.6-fold singlet oxygen yield compared to the unconfined control. The proposed angstrom-confined diatomic strategy is applicable to serial diatomic catalysts, providing an efficient and universal design scheme for singlet oxygen-mediated selective wastewater treatment technology at the atomic level.

A Control Strategy for Salient-Pole Permanent Magnet Axial-Gap Self-Bearing Motors Considering Stator Inductance Variation

Axial-Flux Self-Bearing Motors (AFBMs) are distinguished by considerable variations in inductance resulting from axial displacement. Conventional controllers are frequently engineered to accommodate a wide range of inductance changes. Nevertheless, this methodology is often suboptimal, particularly when inductance is not accurately determined. This paper puts forward a novel inductance approximation function for the stator, designed to minimize modeling errors. Furthermore, an advanced control strategy is presented, based on a sliding mode controller with online adaptive parameter tuning. The experimental results demonstrate the efficacy of the proposed control strategy, exhibiting enhanced performance, stability, and robustness across a range of operating conditions.

Multi-agent reinforcement learning-driven adaptive controller tuning system for autonomous control of wastewater treatment plants: An offline learning approach

Multi-agent reinforcement learning-driven adaptive controller tuning system for autonomous control of wastewater treatment plants: An offline learning approach

Bi‐LORA: A Vision‐Language Approach for Synthetic Image Detection

ABSTRACTAdvancements in deep image synthesis techniques, such as generative adversarial networks (GANs) and diffusion models (DMs), have ushered in an era of generating highly realistic images. While this technological progress has captured significant interest, it has also raised concerns about the high challenge in distinguishing real images from their synthetic counterparts. This paper takes inspiration from the potent convergence capabilities between vision and language, coupled with the zero‐shot nature of vision‐language models (VLMs). We introduce an innovative method called Bi‐LORA that leverages VLMs, combined with low‐rank adaptation (LORA) tuning techniques, to enhance the precision of synthetic image detection for unseen model‐generated images. The pivotal conceptual shift in our methodology revolves around reframing binary classification as an image captioning task, leveraging the distinctive capabilities of cutting‐edge VLM, notably bootstrapping language image pre‐training (BLIP)2. Rigorous and comprehensive experiments are conducted to validate the effectiveness of our proposed approach, particularly in detecting unseen diffusion‐generated images from unknown diffusion‐based generative models during training, showcasing robustness to noise, and demonstrating generalisation capabilities to GANs. The experiments show that Bi‐LORA outperforms state of the art models in cross‐generator tasks because it leverages multi‐modal learning, open‐world visual knowledge, and benefits from robust, high‐level semantic understanding. By combining visual and textual knowledge, it can handle variations in the data distribution (such as those caused by different generators) and maintain strong performance across different domains. Its ability to transfer knowledge, robustly extract features and perform zero‐shot learning also contributes to its generalisation capabilities, making it more adaptable to new generators. The experimental results showcase an impressive average accuracy of 93.41% in synthetic image detection on unseen generation models. The code and models associated with this research can be publicly accessed at https://github.com/Mamadou‐Keita/VLM‐DETECT.

Advanced low-power filter architecture for biomedical signals with adaptive tuning.

This paper presents a low-power, second-order composite source-follower-based filter architecture optimized for biomedical signal processing, particularly ECG and EEG applications. Source-follower-based filters are recommended in the literature for high-frequency applications due to their lower power consumption when compared to filters with alternative topologies. However, they are not suitable for biomedical applications requiring low cutoff frequencies as they are designed to operate in the saturation region. The major contribution in this work are the filter is made to operate in the weak inversion zone to reduce the area needed for the capacitor and the amount of power dissipated. Process variation is one of the major issues in the weak inversion regime. To overcome this, a unique method of compensating against fluctuations in process, voltage, and temperature is put forth based on magnitude comparison is another contribution. Key findings from post-layout simulations and experimental measurements demonstrate that the filter achieves a tunable cutoff frequency range of 0.5 Hz to 150 Hz, with a total power dissipation of only 6nW at 150 Hz. The design occupies a compact silicon area of 0.065 mm2 and offers a dynamic range of 75 dB. The measured results indicate that for a 300 mVpp signal swing, the top bound on THD is -40 dB. The filter's robustness against process, voltage, and temperature variations is validated through on-chip tuning using a current steering DAC, ensuring stable performance across different operating conditions. These results make the proposed filter a promising candidate for low-power biomedical devices. The recommended filter is developed and implemented using UMC-0.18μm CMOS technology with a 1.0V supply, and the IC is tapped out using an MPW run of Euro practice IC services.

Adaptive Parameter Tuning and Virtual Impedance Injection Control for Coupled Harmonic Mitigation of Photovoltaic Converter

Adaptive Parameter Tuning and Virtual Impedance Injection Control for Coupled Harmonic Mitigation of Photovoltaic Converter

Adaptive dynamic programming for trajectory tracking control of a tractor‐trailer wheeled mobile robot

AbstractTractor‐trailer wheeled mobile robots (TTWMRs) possess complex nonlinear dynamics that make their precise trajectory tracking control challenging. This paper explores an adaptive dynamic programming (ADP) approach that utilizes critic neural networks to improve tracking control for continuous‐time TTWMRs. To achieve this, the decoupled kinematic and dynamic loops of the TTWMR are considered, and ADP controllers are proposed aimed at integrated trajectory and velocity tracking. Tis study defines two tracking error systems related to the kinematic and dynamic control loops, which reduces the computational load compared to previous research. The two critic neural networks approximate the optimal cost functions and enable the adaptive tuning of the control policies. Theoretical analysis demonstrates both closed‐loop stability and convergence. Simulation results indicate that the proposed method offers superior tracking performance compared to earlier techniques, exhibiting lower errors and reduced control efforts. This underscores the advantages of using ADP to optimize the control of TTWMRs, even in the presence of partially unknown dynamics.

Robust and Adaptive Tuning of PI Current Controllers for Grid-Forming Inverters

Robust and Adaptive Tuning of PI Current Controllers for Grid-Forming Inverters

Mobile SoC Power Optimization: Redefining Performance with Machine Learning Techniques

Power optimization is a critical challenge in System-on-chip (SoC) design due to the growing demand for high-performance mobile devices with extended battery life. This paper explores novel applications of machine learning (ML) technologies to readdress the question of power optimization strategies in mobile SoCs. In order to balance performance and power consumption in real time, ML takes advantage of predictive modeling, dynamic power management, and energy efficient workload scheduling. This study makes key contributions to integrating ML models into SoC architectures, including a novel ML-driven adaptive voltage scaling and frequency tuning workflow that demonstrates the potential for significant power savings, as well as empirical analysis. Given our findings, we believe that ML offers a transformative potential for the realization of sustainable and efficient mobile computing solutions.

A fuzzy state-dependent Riccati equation control: adaptive tuning of the state weighting matrix

A fuzzy state-dependent Riccati equation control: adaptive tuning of the state weighting matrix

An Improved SOGI-Higher-Order Sliding Mode Observer-Based Induction Motor Speed Estimation

Abstract This article presents a novel adaptive gain tuning second-order generalised integrator (SOGI)-higher-order sliding mode (HOSM) observer for robust speed estimation for an induction motor’s entire speed range. This article introduces a hyperbolic tangent function and a varying gain exponent that ensures accurate speed estimation under noisy conditions and significantly reduces chattering observed in conventional sliding mode observers (SMOs). The robustness of the proposed speed estimation method is verified through simulations conducted on MATLAB/Simulink R2024a developed by MathWorks, demonstrating its capability to effectively track the motor’s actual speed even under varying load torque conditions, parameter variations and additional sensor noise. The proposed approach’s superiority and robustness were compared with the conventional SOGI-frequency locked loop (FLL) and super twisting algorithm (STA) SMO.

Enhanced Metabolic Control in a Pediatric Population with Type 1 Diabetes Mellitus Using Hybrid Closed-Loop and Predictive Low-Glucose Suspend Insulin Pump Treatments.

Insulin pumps coupled with continuous glucose monitoring sensors use algorithms to analyze real-time blood glucose levels. This allows for the suspension of insulin administration before hypoglycemic thresholds are reached or for adaptive tuning in hybrid closed-loop systems. This longitudinal retrospective study aims to analyze real-world glycemic outcomes in a pediatric population transitioning to such devices. We evaluated children with type 1 diabetes mellitus (T1D) admitted to the Pediatric Diabetes Department from a major University Hospital in Bucharest, Romania, who transitioned to hybrid closed-loop or predictive low-glucose suspend system from either non-automated insulin pumps or multiple daily injections. The primary outcome was assessing the change in glycated hemoglobin (HbA1c) after initiating these devices. Secondary outcomes analyzed changes in glucose metrics from the 90 days prior to the baseline and follow-up visit. 51 children were included (58.8% girls), the mean age was 10.3 ± 3.7 years, and the mean follow-up duration was 13.2 ± 4.5 months. The analyzed parameters, such as HbA1c (6.9 ± 0.7% vs. 6.7 ± 0.6%, p = 0.023), time in range (69.3 ± 11.2% vs. 76 ± 9.9%, p < 0.001), time in tight range (47.4 ± 10.9% vs. 53.7 ± 10.7%, p < 0.001), time below range (5.6 ± 2.9% vs. 3.5 ± 1.9%, p < 0.001), time above range (25 ± 11.2% vs. 20.4 ± 9.4%, p = 0.001), and coefficient of variation (37.9 ± 4.8% vs. 35.6 ± 4.6%, p = 0.001), showed significant improvements. The application of these sensor-integrated insulin pumps can significantly enhance metabolic control in pediatric populations, minimizing glycemic variations to mitigate complications and enrich the quality of life.

Method of oscillation excitation for investigation of inconsistency of coating deposition on long parts

A method is proposed for studying the uniformity of electrochemical coating on the outer wall of a thin metal rod by determining the parameters of natural oscillations of this rod, which is clamped on one side. The parameters are studied by analyzing the natural frequencies of the rod. The uneven coating of the rod wall causes a change in the frequency of transverse modes and the shape of the bending line of the rod axis. The mechanical vibrations of the rod are excited by the interaction of an external magnetic field with the Ampere force arising from the current flowing through the rod. A method of adaptive tuning of the spectral components of the excitation current is developed by modifying the coefficients of a digital filter through which a noise-like signal pass. The coefficients of the digital filter are changed adaptively, based on the analysis of the signal spectrum obtained in response to the excitation in the first step. The oscillation is excited by mixing the noise-like signal and the signal received in response to the excitation in the previous step after averaging with a sliding window, the coefficients of which are obtained from the resonance characteristics of the mechanical oscillatory system. A gradual change in the ratio of the amplitude of the noise-like and stationary components of the excitation signal causes a smooth increase in the amplitude of oscillation at a frequency approaching the frequency of the rod's natural oscillations. A numerical and full-scale experiment is carried out to excite mechanical vibrations of a thin metal rod, the end of which is clamped on one side.

An Enhanced Crowned Porcupine Optimization Algorithm Based on Multiple Improvement Strategies

The Crowned Porcupine Optimization (CPO) algorithm exhibits certain deficiencies in initialization efficiency, convergence speed, and adaptability. To address these issues, this paper proposes an enhanced Crowned Porcupine Optimization algorithm (ICPO) based on multiple improvement strategies. ICPO optimizes the initialization process by introducing Logistic chaotic mapping, thereby expanding the search space. It accelerates convergence through an elite retention strategy and enhances global search capability by integrating stochastic operations, mutation-like operations, and crossover-like operations to increase population diversity. Additionally, adaptive step tuning based on fitness values is employed to comprehensively improve the algorithm’s performance. To verify the effectiveness of ICPO, 23 standard functions were used for a comprehensive evaluation, and its practicality was further validated through optimization of actual engineering design problems. The experimental results demonstrate significant improvements in convergence speed, solution quality, and adaptability with ICPO.

Study on Self-tuning of Robot Parameters for EMC Vehicle Steering Test

With the continuous advancement of electric vehicles and smart internet technologies, ensuring vehicle safety through electromagnetic compatibility (EMC) testing in complex electromagnetic environments has become increasingly critical. However, due to the significant variability in vehicle response characteristics under electromagnetic interference, traditional PID control methods for steering robots struggle to meet the high-precision requirements of such tests. In this study, a novel fuzzy PID parameter self-tuning method is proposed, leveraging a Sparrow Search Algorithm-Back Propagation (SSA-BP) neural network. This method optimizes the fuzzy controller's quantization factor by constructing a neural network system where the expected motor angle serves as the input and the quantization factor as the output. The quantization factor is then calibrated online through iterative training.The proposed approach enables the steering robot to achieve real-time, adaptive tuning of the PID parameters for the drive motor by adjusting the steering torque according to different vehicle characteristics, thereby enhancing the robot's anti-interference capability and robustness in EMC testing. The effectiveness of this method is validated through Matlab/Simulink simulations, experiments conducted on the Sensodrive platform, and tests performed in an EMC anechoic chamber. The results indicate that the method offers substantial improvements in control accuracy and anti-interference capabilities, highlighting its strong potential for practical application.

Enhanced Multi-Objective Grey Wolf Optimization using Adaptive Diversity Tuning and Levy Flights

This study proposes an enhanced Multi-Objective Grey Wolf Optimizer (MOGWO) using adaptive population diversity tuning and levy flight theories (EMOGWO-ADTLF). It addresses the issues of parameter tunning by balancing exploration and exploitation. Using MATLAB and Python library Pymoo, the study implemented and evaluated the performance of EMOGWO-ADTLF using multi-objective test problems. The results were compared to high-performing algorithms like MOGWO, Non-Dominated Sorting Grey Wolf Optimizer (NSGWO), Dynamic Chaos MOGWO (DCMOGWO), Multi-Objective Mayfly Algorithm (MMA), Multi-Objective Antlion Algorithm (MOALO) and Multi-Objective Dragonfly Algorithm (MODA). In this work, inverted generational distance (IGD) and hypervolume (HV) were the metrics used to measure the performance of algorithms. The metrics measure the diversity, coverage, and spread of solutions. The results obtained showed the potency of EMOGWO-ADTLF in approximating the Pareto fronts. It ranks first in overall average scores in IGD and HV, with total rank scores of 17 and 18, respectively.

Graphene based THz filter design using open-loop resonator for space applications

This paper introduces a graphene-based tunable terahertz bandpass filter with a dual-mode open-loop resonator, designed to operate at a resonant frequency of 6.35 THz. Leveraging a graphene layer between the conductor and dielectric layers, the filter enables the propagation of surface plasmons, allowing dynamic tunability essential for terahertz applications. Key advantages of this design include precise bandwidth control, achieved by manipulating the resonator's physical parameters, and adaptable center frequency tuning through variations in graphene's chemical potential, providing a tuning range of approximately 0.16 THz. These reconfigurable properties make the filter highly suitable for integration into terahertz space communication systems, where adaptability and high performance are critical. The proposed design's ability to fine-tune both bandwidth and center frequency establishes it as a versatile and efficient solution for emerging terahertz-based applications, demonstrating significant potential for enhancing the flexibility and functionality of future communication technologies.

FedEco: Achieving Energy-Efficient Federated Learning by Hyperparameter Adaptive Tuning

FedEco: Achieving Energy-Efficient Federated Learning by Hyperparameter Adaptive Tuning