Adaptive Optimization Research Articles

Federated learning for secure and privacy preserving data analytics in heterogeneous networks

Federated Learning (FL) has emerged as a groundbreaking paradigm enabling collaborative machine learning across distributed nodes without centralizing data, thus addressing critical concerns in security and privacy. This survey explores the application of FL for secure and privacy-preserving data analytics in heterogeneous networks, where diverse devices, data distributions, and network conditions present unique challenges. This paper provides a comprehensive review of recent advancements in FL, focusing on its efficacy in safeguarding sensitive information while enabling effective analytics across varied domains such as healthcare, finance, and IoT systems. The paper delves into key methodologies for achieving privacy preservation, including differential privacy, secure multi-party computation, and homomorphic encryption, while analyzing their performance in dynamic and resource-constrained environments. Additionally, this paper examines strategies for managing heterogeneity, including personalized FL, model aggregation techniques, and adaptive optimization algorithms. Challenges such as scalability, communication efficiency, and adversarial robustness are discussed alongside potential solutions and future research directions. This survey aims to provide researchers and practitioners with an in-depth understanding of the state-of-the-art in FL for secure and privacy-preserving data analytics, fostering innovation and addressing emerging needs in increasingly complex network ecosystems.

Multi-objective maintenance strategy for complex systems considering the maintenance uncertain impact by adaptive multi-strategy particle swarm optimization

Effective maintenance optimization strategies are crucial for improving the complex equipment reliability and reducing the maintenance costs. However, the effectiveness of the maintenance procedures applied to industrial equipment is affected by uncertainty, e.g. due to the professional skills of maintenance personnel, the actual condition of the equipment being maintained. The quantification of the uncertainty on the effect of maintenance has practical significance and must be accounted in the development of maintenance strategies for reducing equipment probability of failure. This paper proposes a multi-objective maintenance strategy considering the uncertain impact of the maintenance actions. The impact of maintenance actions on the reliability of components is first studied and a reliability assessment model is developed, which considers the skill of maintenance personnel and the actual condition of the equipment. To optimize the multi-objective maintenance strategy, a multi-strategy particle swarm optimization (MS-PSO) algorithm is proposed. Two case studies are considered to verify the effectiveness of the proposed approach for multi-objective maintenance strategy optimization. In the case studies considered, it turns out that the maintenance cost rate (MCR) is reduced throughout the system life cycle and the cumulative availability is improved.

A New Interpretable State‐of‐Health Assessment Model for Lithium‐Ion Batteries With Multidimensional Adaptability Optimization

ABSTRACTThe state‐of‐health (SOH) assessment of lithium‐ion batteries is critical to the development and optimization of maintenance strategies. To ensure the accuracy of the assessment results, it must not only address a variety of uncertainties but also rationalize and transparently conduct the assessment process, as well as make the results interpretable and traceable. These requirements are necessary to ensure that the battery operates safely and steadily. As an interpretable modeling method, belief rule base (BRB) has been widely used in lithium‐ion battery SOH assessment. However, current BRB‐based models face two problems: (1) The initial reference values provided by experts often have limited accuracy due to complex internal chemistry. (2) The multidimensionality of the parameters in the interpretable optimization process and the differences in their properties should be fully considered. Therefore, this paper proposes a new SOH assessment model for lithium‐ion batteries based on an interpretable BRB with multidimensional adaptability optimization (IBRB‐mao). First, an interpretable knowledge and data dual‐driven reference value generation method is proposed to address the issue of imprecise reference values. Expert knowledge is maintained when generating reference values using this method. Second, two interpretable multidimensional constraint strategies are proposed to ensure interpretability in the optimization process. Finally, the NASA lithium‐ion battery data set is taken as a case study to validate the effectiveness of the proposed method.

Adaptation Time to Climate-Induced Extreme Events—Impact of Trend, Seasonality, and Interest Rate Stochasticity

Climate-induced catastrophes such as bushfires, hurricanes, floods, and storm surges have caused significant damage in many regions. The frequency and severity of these events may further increase with climate change, making adaptation imperative. In this article, we introduce a new modeling framework to analyze regional adaptation to catastrophic risk. Using extreme value theory to model climate risk, we analyze the impact of seasonality and risk aversion on investment decisions. We also analyze the impact of stochastic interest rates on optimal adaptation timing. In a case study of New York City’s flood risk management, we find that neglecting seasonality or risk aversion can underestimate investment values and cause unwarranted adaptation delays. In addition, our findings highlight that allowing for stochastic interest rates can lead to earlier climate change adaptation.

Two-Stage Estimation and Variance Modeling for Latency-Constrained Variational Quantum Algorithms

The quantum approximate optimization algorithm (QAOA) has enjoyed increasing attention in noisy, intermediate-scale quantum computing with its application to combinatorial optimization problems. QAOA has the potential to demonstrate a quantum advantage for NP-hard combinatorial optimization problems. As a hybrid quantum-classical algorithm, the classical component of QAOA resembles a simulation optimization problem in which the simulation outcomes are attainable only through a quantum computer. The simulation that derives from QAOA exhibits two unique features that can have a substantial impact on the optimization process: (i) the variance of the stochastic objective values typically decreases in proportion to the optimality gap, and (ii) querying samples from a quantum computer introduces an additional latency overhead. In this paper, we introduce a novel stochastic trust-region method derived from a derivative-free, adaptive sampling trust-region optimization method intended to efficiently solve the classical optimization problem in QAOA by explicitly taking into account the two mentioned characteristics. The key idea behind the proposed algorithm involves constructing two separate local models in each iteration: a model of the objective function and a model of the variance of the objective function. Exploiting the variance model allows us to restrict the number of communications with the quantum computer and also helps navigate the nonconvex objective landscapes typical in QAOA optimization problems. We numerically demonstrate the superiority of our proposed algorithm using the SimOpt library and Qiskit when we consider a metric of computational burden that explicitly accounts for communication costs. History: Accepted by Giacomo Nannicini, Area Editor for Quantum Computing and Operations Research. Accepted for Special Issue. Funding: This material is based upon work supported by the U.S. Department of Energy, Office of Science, National Quantum Information Science Research Centers and the Office of Advanced Scientific Computing Research, Accelerated Research for Quantum Computing program under contract number DE-AC02-06CH11357. Y. Ha and S. Shashaani also gratefully acknowledge the U.S. National Science Foundation Division of Civil, Mechanical and Manufacturing Innovation Grant CMMI-2226347 and the U.S. Office of Naval Research [Grant N000142412398]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2024.0575 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2024.0575 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

Adaptive bi-level path optimization for deep-sea mining vehicle in non-uniform grids considering ocean currents and dynamic obstacles

The navigation safety of Deep-sea Mining Vehicle (DSMV) is significantly threatened by unstructured seabed terrain and dynamic obstacles. To address this challenge, this paper proposes an adaptive bi-level path optimization (ABLPO) method that adaptively adjusts cost function weights for optimal global path planning and effective local obstacle avoidance in dynamic environments. Initially, a non-uniform grid-based traversability map is established using sonar data integrated with a seabed current model. Dense grids in high-risk areas enhance safety through detailed calculations, while sparse grids in low-risk areas improve computational efficiency. Based on this non-uniform grid map, the ABLPO method is implemented for bi-level path planning: the upper level utilizes the adaptive-weight A∗ (AWA∗) algorithm for global path planning noted for its computational efficiency, while the lower level employs the adaptive dynamic window approach (ADWA) for local obstacle avoidance, providing efficient navigation in dynamic environments. Throughout the planning process, ABLPO adaptively adjusts the weights for path distance, traversability, and energy consumption to determine the optimal path. Simulation results demonstrate that ABLPO, with its adaptive weight adjustment, notably outperforms traditional methods in terms of convergence time, path distance, energy consumption, and traversability. The experiments further confirm the feasibility of the ABLPO method.

Adaptive Partition Linearization Global Optimization Algorithm and Its Application on the Simultaneous Heat Exchanger Network and Organic Rankine Cycle Optimization

Adaptive Partition Linearization Global Optimization Algorithm and Its Application on the Simultaneous Heat Exchanger Network and Organic Rankine Cycle Optimization

A Talent Cultivation and Performance Evaluation Model Based on a Fuzzy Control Algorithm

The Internet era has raised the demand for skilled workers in businesses and for economic expansion. Businesses find it difficult to expand when trained personnel are shortages. Evaluating and managing exceptionally gifted persons have become more difficult due to technological developments, which have slowed down technological progress and business expansion. The study suggests a technique to quantify how firms develop and employ talent to address this issue. An algorithm called the fuzzy optimized talent cultivation engine (FOTCE) is introduced to evaluate the effectiveness of talent development. The algorithm combines an adaptive neuro-Fuzzy Inference System with Adaptive Hybrid Particle Swarm Optimization. By considering factors including work-life balance, training, involvement, job performance, and satisfaction, the suggested FOTCE algorithm hopes to improve talent strategies and boost organizational competitiveness. The FOTCE algorithm assesses talent development on three levels: overall efficacy, major categories (such as job satisfaction and involvement), and individual indicators within each category. Fuzzy logic handles subjective viewpoints and unclear data in the study, allowing for a more comprehensive assessment. The technique uses fuzzy math to show different levels of employee retention by giving different factors different weights. The results show that the FOTCE technique helps evaluate and upgrade talent development programs and provides valuable insights for better people management when tested with accurate HR data from a company. The findings demonstrate that the suggested FOTCE algorithm outperformed state-of-the-art models like IAA-NRM (83% retention rate) and FAHP (87% retention rate) by a wide margin. Comparatively, FAHP had an EEI of 83% and DEA had 86%. The EEI for this organization achieved 95%. In addition, the FOTCE model showed a higher Talent Development Efficiency Score (TCDES) of 93%, proving that it boosts organizational competitiveness and employee happiness.

Distribution-Aware Dual-LLM Collaborative Framework for Human Preference Prediction

This paper proposes a novel Distribution-aware Dual-LLM Collaborative Framework (D2CF) for human preference prediction in large language model dialogue systems. Through data analysis in the Kaggle LMSYS Chatbot Arena competition, we innovatively selected two complementary base models: Gemma-2-9b and Llama-3.1-8b. The framework's main technical innovations include: (1) A model complementarity quantification method based on Wasserstein distance, optimizing model selection from a data distribution perspective; (2) A parameter-efficient QLoRA improvement strategy that reduced computational overhead by 42.6% through adaptive rank adjustment and quantization optimization; (3) A validation set-driven dynamic weight fusion mechanism that achieves adaptive feature fusion through attention mechanisms. In the competition evaluation, this solution achieved stable performance on both public and private test sets, ultimately winning a silver medal, validating the effectiveness and robustness of distribution-aware strategies in practical applications. The significant performance improvement from public to private test sets demonstrates the superiority of this method in handling different data distributions. This paper details the technical principles and implementation details of the solution, providing reproducible engineering practice references for human preference prediction tasks in large-scale language models.

Prediction of Sunspot Activity Based on Differential Evolution Algorithm and BP Neural Network Model

In this research report, the prediction method of sunspot activity is deeply discussed, and a variety of statistical and data science models are used to make comprehensive prediction. Research focuses include the start time and duration of the solar maximum in the solar cycle, as well as the prediction of the number and area of sunspots. By capturing the periodicity of solar activity, Pearson correlation coefficient is used to analyze the relationship between the maximum solar activity and the number of sunspots, and the adaptive multivariate nonlinear regression-BP neural network model and particle swarm optimization BP neural network are combined to make high-precision prediction. The research results provide a scientific basis for the prediction of space weather, ionospheric state and communication system reliability.

A Signal Generation Fault Diagnosis Method for Planetary Reducer of High‐Pressure Grinding Roll

ABSTRACTThis paper proposes a signal generation fault diagnosis method for the challenge of insufficient training samples in the planetary reducer of high‐pressure grinding roll (HPGR). The expression of the vibration response signal is derived, and a fault vibration signal model is established for the planetary reducer of HPGR. To generate the expanded samples, an adaptive fox optimization (AFO) algorithm is employed for optimizing the parameters of the fault vibration signal model, so that the simulated signal matches the measured signal. Before that, the measured vibration signal of the planetary reducer of HPGR is preprocessed by a tunable Q‐factor wavelet transform (TQWT). Finally, the residual convolution neural and long short‐term memory network with attention mechanism (ResCNN‐LSTM‐ATT) is utilized for feature extraction from collected signals, and the hierarchical fault classifier is developed for the fault classification. The experimental results show that the AFO algorithm has better performance, and the signals generated by the optimized fault vibration signal model are in good agreement with the measured signals. The classification accuracy of normal signals reaches 99.70%, which shows good diagnosis accuracy in both known fault classification and unknown fault identification. Compared to other methods, the proposed method has better diagnostic accuracy.

A novel multi-objective robust optimization method based on improved Gray Wolf optimizer and Kriging model

A novel multi-objective robust optimization method for mechanical structure products is introduced. Firstly, the method introduces Bernoulli Dynamic Adaptive Gray Wolf Optimizer (BDAGWO) to address inherent limitations within the original Gray Wolf Optimizer (GWO), such as falling into local optimal solutions and low convergence rate. Subsequently, BDAGWO is utilized to search optimal correlation coefficients of Kriging. The proposed surrogate model exhibits high fitting accuracy on different test cases, with coefficients of determination reaching above 0.99 on the test set, and mean relative error, mean absolute error, mean absolute percentage error and root mean squared error are all close to 0. To solve multi-objective optimization problems, an improved NSGA-II is introduced to amplify search capabilities. Then based on robust optimization method, combine BDAGWO-Kriging and improved NSGA-II to establish a multi-objective robust optimization framework with high accuracy and high solution efficiency. The proposed method is implemented to an EMU bogie frame, demonstrating that the robust optimization scheme reduces the maximum equivalent stress and mass, and exhibits lower fluctuation compared to deterministic optimization. This substantiates the method’s effectiveness and provides an optimization approach of large and complex mechanical products.

Multiobjective Secure Trust Aware Redundant Array Shifting Encryption and Clustering Based Routing in Mobile Ad Hoc Networks

ABSTRACTMobile ad hoc networks (MANETs) are characterized by their self‐configuring and decentralized nature in wireless networking. In MANETs, energy‐constrained devices face increased packet loss due to malicious nodes, leading to higher delays and energy consumption. Addressing this issue is crucial to improve network performance and efficiency. Therefore, to overcome these issues, a multiobjective secure trust‐aware redundant array shifting encryption adaptive sparrow search optimization (MORS‐ASO) was introduced in this manuscript for secure data transmission in MANETs. This encryption enhances security by distributing data across multiple nodes and dynamically shifting it, making it more resistant to attacks and unauthorized access. The hybrid integrated cat swarm optimization is the combination of integrated optimization algorithm and cat swarm optimization. Adaptive sparrow search optimization is applied to carry out the data transmission securely. The proposed method helps to offer security against various routing disruption attacks, genuine and hurtful attack. The MORS‐ASO achieves a packet delivery ratio of 94.3%, an end‐to‐end delay of 3.362 s, a throughput of 2952.7548 bps, and a jitter of 1.245 ms. Also, it consumes 32 joules of energy, supports a network lifetime of 3500 s, and has a packet loss ratio of 19%.

Seasonal dynamics of water-use strategies and response to precipitation in different habitats of Nitraria L.

Nitraria L. is a dominant shrub in arid areas and its survival is hampered by low and unpredictable precipitation and uncertain future water conditions. However, little is known about the shrub’s water requirements or its response to precipitation. We determined the water utilization strategies of Nitraria L. species in habitats with different soil textures using the isotopic composition of xylem water and those of potential water sources (groundwater and vadose zone soil water). We used the MixSIAR model to quantify the relative contribution of potential water sources to shrub water in different soil textures. Our results showed that (1) The dynamic characteristics of water use of Nitraria L. species differed in different soil textures. In gravel soils, water in all soil layers was recharged by precipitation, and the shrub water source was controlled by precipitation with significant seasonal changes. In sandy and clay soils, shallow soil water was recharged by precipitation infiltration, but deep soil water was recharged by capillary rise. Nonetheless, the shrub water sources exhibited considerable seasonal fluctuations. During wet seasons, the shrub’s primary water sources were in shallow and mid-soil depths. However, during the dry season, the shrubs relied on groundwater, with more than half of their water originating in deep soil layers. (2) Nitraria L. species generally responded significantly to precipitation events, and those that survived in three soil textures were able to rapidly switch water sources to varying degrees. In particular, in sandy and gravelly soils, the proportion of deep soil water (24.3 %) and groundwater (16.2 %) used by Nitraria L. plants decreased significantly after a large precipitation event (e.g., 18.8 and 11.9 mm), shifting to a predominantly transient use of shallow soil water (48.5 %). Nitraria L. could shift water-absorbing soil layers according to the availability of potential water sources in different soil textures. This flexibility allows them to access the most readily available water more rapidly. Such optimal ecological adaptation can ensure that the plants will have an advantage in future predicted water shortage conditions.

Multiple strategies improved spider wasp optimization for engineering optimization problem solving

The Spider Wasp Optimization (SWO) algorithm is a swarm intelligence optimization technique inspired by the collective behaviors of social animals. This algorithm, designed to address optimization challenges, emulates the unique hunting, nesting, and mating behaviors of female spider wasps. It offers several advantages, including rapid search speed and high solution accuracy. However, when tackling complex optimization problems, it can encounter issues such as getting trapped in local optima, slow early convergence, and the need for manual adjustment of the “Trade-off Rate” (TR) parameter for different problems.To improve the performance and versatility of the SWO algorithm, a Multi-strategy Improved Spider Wasp Optimizer (MISWO) is proposed. Firstly, the Grey Wolf Algorithm is integrated into the initialization phase to enhance early convergence and improve the fitness of the initial population, thereby boosting the algorithm’s global optimization capabilities.Secondly, an adaptive step size operator and Gaussian mutation are introduced during the search phase to automatically adjust the search range at different optimization stages. This enhancement increases both the optimization accuracy and the algorithm’s ability to avoid local optima. The Trade-off Rate (TR) is dynamically selected to better accommodate a variety of problems. Finally, a dynamic lens imaging reverse learning strategy is employed to update optimal individuals, further improving the algorithm’s capacity to escape local optima. To validate the effectiveness of MISWO, it was tested on 23 benchmark functions and 7 engineering optimization problems, and compared with several state-of-the-art algorithms. Experimental results show that MISWO outperforms other algorithms in terms of optimization capability, stability, and adaptability across diverse problems.

Application of the AVMD-AIMCKD method to the enhancement of rolling bearing fault features

Abstract Rolling bearings are prone to failure due to harsh working environments, which can affect production efficiency. Given the background environmental noise interference, it is difficult to extract the fault characteristics of rolling bearings in the early stages of weak fault signals. Based on adaptive VMD combined with adaptive IMCKD (AVMD-AIMCKD), a rolling bearing fault diagnosis method is proposed in this paper. The vibration signal from the rolling element bearing is processed by adaptive variational modal decomposition (AVMD) to extract and select the time-frequency domain signal characteristics. The Adaptive Improved maximum correlated kurtosis deconvolution (AIMCKD) algorithm is used for noise reduction optimization to obtain the best extraction results. Because the traditional method requires input parameter values for VMD and IMCKD, it is not adaptive. The particle swarm optimization algorithm (PSO) is utilized in this study to optimize two variables: the penalty factor and decomposition mode number of variable mode decomposition (VMD). The filter length and shift order of the improved maximum correlated kurtosis deconvolution (IMCKD) are optimized to make it adaptive. The viability of the method is substantiated through simulations, and the efficacy and precision are validated by comparative studies. The experimental results show that the AVMD-AIMCKD method has high accuracy and robustness in rolling bearing fault diagnosis and provides an accurate reference for rolling bearing health monitoring in engineering practice. The AVMD-AIMCKD method effectively extracts the early fault features through adaptive optimization, overcomes the restrictions of traditional methods, and provides a more accurate and reliable tool for bearing fault signal diagnosis.

Workload-based adaptive decision-making for edge server layout with deep reinforcement learning

Mobile edge computing (MEC) is crucial in applications such as intelligent transportation, innovative healthcare, and smart cities. By deploying servers with computing and storage capabilities at the network edge, MEC enables low-latency services close to end users. However, the configuration of edge servers needs to meet the low-latency requirements and effectively balance the servers’ workloads. This paper proposes an adaptive layout and dynamic optimization method, modeling the edge server layout problem as a Markov decision process. It introduces a workload-based server placement rule that adjusts the locations of edge servers according to the load of base stations, enabling the learning of low-latency and load-balanced server layout strategies. Experimental validation on a real dataset from Shanghai Telecom shows that the proposed algorithm improves average latency performance by about 40% compared to existing technologies, and enhances workload balancing performance by about 17%.

Adaptive Optimization of Wind Environment in Coastal Village Spatial Forms of Western Guangdong

Adaptive Optimization of Wind Environment in Coastal Village Spatial Forms of Western Guangdong

Fully distributed adaptive optimization event-triggered/self-triggered synchronization for multi-agent systems

This paper investigates the distributed adaptive optimization synchronization problem of multi-agent systems (MASs) with general linear dynamics on undirected graphs. The goal is to fulfill the synchronization among agents and synergistically optimize the team cost function formed by a family of local convex functions. The tracking servo signal is first generated by sampling the implicit state of the axillary system, and its sampling events are governed by the triggering mechanism I. Meanwhile, the disagreement vector is sampled if the triggering mechanism II is violated. An adaptive event-triggered scheme is then constructed by the gradient term whose input is the tracking servo signal and the relative sampling information among followers, which fulfills the synchronization as the desired one and minimizes the team cost function. It proves that Zeno behavior is excluded under the triggering mechanisms I and II, respectively. Moreover, a self-triggered strategy is leveraged that depends only on the partial derivative of the local cost function in the implicit state sampling and the relative sampling information of itself and its neighbors; thus, continuously monitoring the information of neighbors is avoided. It is noted that the proposed scheme incorporates adaptive event-triggered control, which makes it possible to implement the fully distributed control manner. The efficacy and advantage of the presented theoretical results are finally demonstrated using a non-trivial simulation example.

Adaptive Optimization and Dynamic Representation Method for Asynchronous Data Based on Regional Correlation Degree.

Event cameras, as bio-inspired visual sensors, offer significant advantages in their high dynamic range and high temporal resolution for visual tasks. These capabilities enable efficient and reliable motion estimation even in the most complex scenes. However, these advantages come with certain trade-offs. For instance, current event-based vision sensors have low spatial resolution, and the process of event representation can result in varying degrees of data redundancy and incompleteness. Additionally, due to the inherent characteristics of event stream data, they cannot be utilized directly; pre-processing steps such as slicing and frame compression are required. Currently, various pre-processing algorithms exist for slicing and compressing event streams. However, these methods fall short when dealing with multiple subjects moving at different and varying speeds within the event stream, potentially exacerbating the inherent deficiencies of the event information flow. To address this longstanding issue, we propose a novel and efficient Asynchronous Spike Dynamic Metric and Slicing algorithm (ASDMS). ASDMS adaptively segments the event stream into fragments of varying lengths based on the spatiotemporal structure and polarity attributes of the events. Moreover, we introduce a new Adaptive Spatiotemporal Subject Surface Compensation algorithm (ASSSC). ASSSC compensates for missing motion information in the event stream and removes redundant information, thereby achieving better performance and effectiveness in event stream segmentation compared to existing event representation algorithms. Additionally, after compressing the processed results into frame images, the imaging quality is significantly improved. Finally, we propose a new evaluation metric, the Actual Performance Efficiency Discrepancy (APED), which combines actual distortion rate and event information entropy to quantify and compare the effectiveness of our method against other existing event representation methods. The final experimental results demonstrate that our event representation method outperforms existing approaches and addresses the shortcomings of current methods in handling event streams with multiple entities moving at varying speeds simultaneously.