Differential Evolution Algorithm Research Articles

As a typical swarm intelligence optimization method, the Differential Evolution (DE) algorithm exhibits excellent performance in solving high-dimensional complex problems; however, its parameter sensitivity and premature convergence issues still restrict its practical application effectiveness. Therefore, this paper proposes an improved Differential Evolution algorithm based on reinforcement learning, namely RLDE. First, it adopts the Halton sequence to realize the uniform initialization of the population space, which effectively improves the ergodicity of the initial solution set. Second, it establishes a dynamic parameter adjustment mechanism based on the policy gradient network, and realizes the online adaptive optimization of the scaling factor and crossover probability through the reinforcement learning framework. Furthermore, it classifies the population according to individual fitness values and implements a differentiated mutation strategy. To verify the effectiveness of the proposed algorithm, 26 standard test functions are used for optimization testing, and comparisons are conducted with multiple heuristic optimization algorithms in 10, 30, and 50 dimensions respectively. Experimental results demonstrate that the proposed algorithm significantly enhances the global optimization performance. Furthermore, by modeling and solving the Unmanned Aerial Vehicle (UAV) task assignment problem, the engineering practical value of the algorithm in real-world scenarios is verified from various indicators.

This study presents a multilayer surface plasmon resonance (SPR) biosensor engineered for the sensitive detection of Mycobacterium tuberculosis bacteria. The suggested configuration includes a prism combined with successive layers of phosphorus (BP). The optical properties were thoroughly investigated utilizing the transfer matrix method (TMM) and validated through finite element method (FEM) simulations. Through optimization of layer thicknesses, the sensor achieved a maximum angular sensitivity of 638°/RIU. Comparative analyses with alternative plasmonic metals, 2D materials, and prism substrates validated the superior sensing capability of the proposed design. Essential performance metrics comprised a minimum full width at half maximum (FWHM) of 3.55°, a peak signal-to-noise ratio (SNR) of 1.19, and a maximum quality factor of , indicating superior resolution and detection accuracy. The differential evolution (DE) algorithm was utilized to optimize essential structural dimensions, thereby improving the sensor's performance. This approach yielded an improved angular sensitivity of 654deg./RIU. Distinct from existing SPR configurations, the proposed sensor integrates a simple multilayer architecture with DE algorithm-based optimization, thereby achieving high sensitivity and an extensive refractive index detection range (1.25-1.35), which together enable precise, label-free identification of diverse biological and chemical analytes.

This study proposes an optimization-based methodology for predicting cutting forces in milling by eliminating the need for traditional offline calibration procedures. A mechanistic force model is employed, in which cutting force coefficients are identified using population-based metaheuristic algorithms, namely Genetic Algorithm (GA), Differential Evolution (DE), and Particle Swarm Optimization (PSO). Cutting force data collected during machining are utilized to optimize the model parameters directly. The performance of each algorithm is systematically evaluated through 30 independent trials to ensure statistical reliability. The DE algorithm demonstrated the best performance, converging in all 30 runs with an average of 197 iterations and 5.4 s, followed by PSO (363 iterations, 9.8 s), while GA exhibited lower reliability (18 successful runs, 2108 iterations, 62.9 s). The optimized coefficients were validated against experimental data, yielding mean prediction errors of 2.82 N (F x ) and 4.35 N (F y ). The proposed method offers a fast, accurate, and scalable solution for cutting force prediction, supporting adaptive process control, and contributing to the development of intelligent manufacturing systems.

Efficient irregular texture nesting, which is necessary for improving the efficiency of texture mapping and 3D model rendering, especially for large-scale 3D reconstruction tasks, has emerged as a critical research topic in the fields of photogrammetry, computer graphics, and computer vision. However, persistent inefficiencies and high computational costs in existing texture nesting algorithms pose significant challenges when dealing with vast quantities of irregularly shaped texture patches. To solve this problem, this work presents an efficient and well structured texture nesting for reorganizing irregular textures in a space efficient and time efficient way. More specifically, a hybrid optimization approach that integrates an enhanced no fit polygon (NFP) method with an improved simplified atavistic differential evolution (SADE) algorithm is proposed. The canonical SADE is reformulated, tailored for texture nesting optimization, and a novel self-adaptive container resizing strategy is used to surpass traditional NFP approaches in polygon processing efficiency. The experimental results demonstrate that the proposed method significantly improves irregular texture nesting efficiency, achieving speed improvements of up to 5.44 times compared with the common genetic algorithm–based method and 5.21 times over the simulated annealing–based method. Furthermore, it consistently improves space use by approximately 6.56%, indicating a more effective layout strategy and optimized resource use. Code is available at https:// github. com/louliyuan/NFP-SADE-With-Adaptive-Container-Resizing.

A stochastic linear neural network-based differential evolution algorithm for optimizing reluctance actuator

DXMODE: A dynamic explorative multi-operator differential evolution algorithm for engineering optimization problems

Surrogate-assisted differential evolutionary algorithm with dynamic region exploration for expensive optimization problems

The research on ceramic powders has always been one of the main research directions of laser cladding technology. This article mainly focuses on the numerical simulation method of temperature field and intelligent optimization of preparation parameters in the laser cladding preparation process of iron-based ceramic layers, in order to optimize the overall process of preparing iron-based ceramic coatings. Firstly, the generation and transfer of laser heat, as well as the heat absorption process of materials, were analyzed and constructed. After determining the Gaussian three-dimensional heat source model, the temperature field of the heat source model was simulated using ANSYS software to obtain appropriate laser power and other parameters. Then, based on various process parameters such as coating material, material supply method, pre coating thickness, laser power scanning rate, and spot size during the preparation process, a multivariate nonlinear model of processing parameters and various performance measurements was constructed with melting power and feed rate as inputs, dilution rate and overlap rate as responses. Nonlinear regression analysis was then conducted, and an improved differential evolution algorithm was used to solve the optimal solutions of various parameters in the layer preparation process. Finally, simulation experimental results were presented to verify the effectiveness and feasibility of the proposed method in this paper.

Water resource management and disaster risk reduction depend on accurate rainfall-runoff modeling (RRM). Time-series data frequently exhibits fine details and local variations that are difficult for traditional LSTM models to capture. To overcome these problems, we introduce an improved RRM model that utilizes a spatial attention-enhanced transductive LSTM (TLSTM) network. By using transductive learning on data points that closely resemble the test set, this model improves performance and captures subtle temporal differences. With the fusion of a spatial attention mechanism, the model can focus on the most important parts of the input. It also includes a more sophisticated differential evolution (DE) algorithm to facilitate complex hyperparameter tuning. For the DE algorithm, we used a mutation strategy that finds a significant cluster using K-means clustering. We applied the Catchment Attributes and Meteorology for Large Sample Studies (CAMELS) dataset and used individual and regional RRM throughout our assessment of the model. There were very positive results for individual basins with an 8-day runoff prediction of 0.728 Nash-Sutcliffe efficiency (NSE). For regional assessment, the model had an NSE of 0.878. This method supports that combining TLSTM with spatial attention and sophisticated DE algorithms increases the accuracy and reliability of rainfall-runoff (RR) predictions, providing opportunities for improved planning and disaster management of water resources. The source code is publicly available at https://github.com/ZhaoChenchina/RRM .

This article proposes a framework based on differential evolution algorithm and bidirectional long short-term memory network to optimize Transformer, providing an intelligent solution for 5G base station RF module fault prediction. Through variable correlation analysis, it can be concluded that the modulation efficiency (0.19) has the strongest positive correlation with fault_type, while the operation time (-0.08) has the strongest negative correlation with fault_type. After comparing the framework with models such as decision tree, random forest, CatBoost, XGBoost, LightGBM, and ExtraTrees, it was found that DE BiLSTM Transformer had significantly better comprehensive performance in 5G base station RF module fault classification prediction: its accuracy reached 0.956, higher than XGBoost and LightGBM's 0.925, ExtraTrees' 0.921, random forest's 0.904, decision tree's 0.864, and CatBoost's 0.798, indicating a significant lead in overall classification accuracy; The recall rate is 0.956, which can more comprehensively identify actual fault samples; The accuracy rate is 0.958, only slightly lower than the 0.919 of random forest, far higher than the 0.9 of XGBoost and 0.86 of LightGBM, demonstrating outstanding ability to accurately identify fault types; The F1 value of 0.941 and AUC value of 0.979 are also the highest, showing better performance in balancing accuracy and recall, distinguishing between faulty and non faulty samples. This advantage stems from the synergistic effect of optimizing model parameters with differential evolution algorithm, capturing bidirectional temporal features with BiLSTM, and focusing on key features with Transformer self attention mechanism, making it more suitable for complex feature correlation scenarios in RF module fault prediction.

ABSTRACT More and more companies have realised that implementing a joint replenishment and delivery (JRD) strategy can lead to significant cost savings. This paper presents a practical JRD problem for heterogeneous products, taking into account resource constraints. An enhanced hiking optimization technique called BHHOA is proposed. BHHOA incorporates a bound heuristic to refine the delivery frequency boundaries and introduces two new population generation methods based on the differential evolution algorithm and attention mechanism. Experimental results demonstrate that BHHOA outperforms six other algorithms with a lower average total cost. The JRD model presented in this study is effectively solved using the BHHOA algorithm. This study provides a practical technical method for companies to implement the JRD strategy.

Advancing biomedical waste classification through a hybrid ensemble of deep Learning, reinforcement Learning, and differential evolution algorithms.

This paper proposes two improved algorithms, the DE-PSO algorithm, which combines differential evolution and phased strategy, and the hybrid particle swarm optimization algorithm integrating whale algorithm (WOAPSO), which combines the whale optimization mechanism. Compared to traditional calibration methods (such as the Newmark three- and two-parameter methods), which rely on empirical simplified models, adapting them to the complex seismic nonstationarity and multipeak characteristics is difficult. However, although intelligent optimization algorithms, such as particle swarm optimization (PSO) and differential evolution (DE) have improved calibration accuracy in recent years, insufficient convergence stability and low computational efficiency, among other problems, persist. Therefore, based on experiments, the performances of these algorithms were compared with those of standard PSO, traditional DE, and other algorithms. The results demonstrate the significant superiority of DE-PSO and WOAPSO. In 50 repeated experiments, the fitness standard deviation (STD) was significantly reduced, and the algorithms achieved rapid convergence by the mid-iteration stage, which effectively resolves the issues of premature convergence and local oscillation tendencies inherent in the standard Particle Swarm Optimization algorithm. Regarding the key parameters (Tg, βmax, γ) of the standard, the STD of the improved algorithm approached zero, verifying its strong adaptability to multimodal optimization problems. Furthermore, the DE-PSO algorithm had the best performance in balancing computational efficiency and stability, with a convergence speed that is three times faster than that of standard DE algorithm while maintaining the lowest parameter volatility. This study provides an efficient algorithmic tool for the rapid analysis of strong motion records and the efficient calibration of design response spectra, which has implications for the seismic optimization design of complex structures and can be guided by regulations, contributing to engineering seismic practice.

For numerous years, researchers have extensively explored real parameter single-objective optimization by evolutionary computation. Among the various types of evolutionary algorithms, Differential Evolution (DE) performs outstandingly. Recently, the academic community has began concerning itself with long-term search. IMODE is a good DE algorithm for long-term search. The algorithm is based on two primary mutation strategies and one secondary. Within the population, the control ratio of each mutation strategy is determined by their respective performance outcomes. Sequential Quadratic Programming (SQP), an iterative method for continuous optimization, is employed on the best individual in the final stage of IMODE at a dynamic probability as a local search method. Based on the DE algorithm, we propose Differential Evolution with Secondary Mutation Strategies (SMSDE). In the proposed algorithm, more secondary mutation strategies are added, in addition to the original one used in IMODE. In each generation, just one of the secondary mutation strategies is activated based on history performance to cooperate with the two primary mutation strategies. In addition, at a dynamic probability, SQP is now called not only for the best individual in the final stage, but also for the worst individual among old ones in each generation. The experimental results demonstrate that SMSDE performs better than a number of state-of-the-art algorithms, including IMODE.

Geomembranes (GM) have been extensively used for waterproofing applications, and often they are in contact with soil materials or other geosynthetics for mechanical protection. Strength evaluation at the interface between the GM and the contact material is fundamental to ensure a good design that guarantees a low probability of this interface failure. This paper analyses and compares four Artificial Neural Network (ANN) models (varying the number of inputs and hidden layers) and the Random Forest (RF) technique to predict sand/GM interface shear strength based on 495 results from previous investigations. All models were optimized with the Differential Evolution (DE) algorithm. The Coefficient of Determination (R2) and root mean squared error (RMSE) were set as evaluation criteria for the accuracy of the developed models. The results show that RF performs best as a prediction tool for the data analysed. Data correlation and RF feature importance analysis were also conducted, establishing GM asperity height as the most significant variable for the collected data. The results show the great potential of Machine Learning applications for predicting the interface shear strength between sand and geomembranes in geotechnical engineering constructions.

Tower cranes are very important in construction works as they are the devices that can lift heavy loads to very high places and take up to a required place. Nevertheless, their mechanical nature, particularly the load swinging and power usage, are sources of big problems in ensuring the safety of both the personnel and the equipment as well as in the running of the system. Mostly, traditional analytical models have difficulties in fully describing the nonlinearity and complexity of the relationships among crane parameters and performance and this is kind of a challenge that calls for data-driven approaches. This paper suggests a mixed strategy of neural networks and metaheuristic optimization to accomplish the prediction and the optimization of the power consumption in tower crane systems. To build up the model, the Container Crane Controller dataset, which is open to the public, containing speed, load angle, and corresponding power was used. As the database was a small-sample one, the authors resorted to data augmentation to extend the training set and enhance the model's ability to generalize. A multilayer perceptron (MLP) network with two hidden layers (32 and 16 neurons, respectively) was set up to learn the nonlinear mapping that takes speed and angle as inputs and power as output.On testing the network secured a high value of determination coefficient (R²), which is a strong confirmation of its capability to reflect the basic dynamics accurately. After the training, the neural network was combined with the Differential Evolution (DE) algorithm to figure out the best operating points for maximum power output.A study of convergence illustrated that DE operations got stable solutions very quickly and did not encounter the problem of local minima. Among the visualizations, such as contour plots and three-dimensional surfaces, the interaction between speed, angle, and power was unveiled, while the convergence curve gave an insight into the stability and reliability of the optimization process.The results confirm that the proposed hybrid approach not only enhances predictive accuracy but also provides practical insights into safe and energy-efficient operation of tower cranes. This paper gives a proof of neural network and evolutionary optimization combination as a very effective method to achieve better crane performance. Moreover, this performance-enhancing methodology can be further employed in the future studies to development of other construction machinery.

This paper proposes a computational and optimization model for determining the effective shielding duration of enemy missiles by smoke countermeasure deployment from unmanned aerial vehicles (UAVs), focusing on the application of a multi-decision variable nonlinear optimization method based on the Differential Evolution (DE) algorithm. First, by analyzing the flat trajectory of smoke countermeasures and the detonation point location, combined with missile trajectory equations and line-of-sight distance criteria, the effective shielding duration is derived. Second, with maximizing shielding duration as the objective, parameters such as heading angle, flight speed, smoke grenade deployment time, and detonation delay are optimized. The DE algorithm performs a global search in continuous variable space through mutation, crossover, and selection operations. Finally, the approach was extended to scenarios involving a single UAV deploying three smoke grenades and a three-UAV deployment. In both cases, the DE algorithm optimized multiple decision variables to derive optimal masking strategies. This model enables precise quantification and optimization of smoke interference effects. The DE algorithm ensures the acquisition of globally optimal solutions, adapts to multi-scenario parameter optimization, and effectively enhances the effectiveness of countermeasures against enemy missiles.

Quadrupole mass spectrometers are highly sensitive and specific analytical instruments, widely used in pharmaceuticals, clinical diagnostics, and other fields. Their performance depends on a tuning process to optimize key parameters, which has traditionally relied on engineers’ expertise or simple univariate search methods. This paper proposes an automatic tuning method using an improved differential evolution algorithm. This algorithm introduces a ranking and subpopulation classification for individuals, enabling distinct mutation strategies. Validation on the CEC-2017 benchmark functions confirms the superiority of the improved algorithm. In automatic tuning experiments, it achieved a 25.3% performance gain over the univariate search method and also surpassed both the classical differential evolution algorithm and standard particle swarm optimization algorithm. This method proves to be an effective approach for enhancing the performance of quadrupole mass spectrometers.

The ability to generate dynamic, expressive dance routines that adapt to various musical compositions has broad applications in activity recognition, performance arts, entertainment, virtual reality, and interactive media, offering new avenues for creative professionals and audiences alike. In this article a deep learning framework is developed for music-synchronized dance choreography through modified vision transformers and graph convolutional networks based on Mexican hat wavelet function for position quantization and motion forecasting. More explicitly high-dimensional pose characteristics are extracted from dance video frames using modified vision transformer to generate a skeletal graph, while modified graph convolutional network captures the spatial and temporal relationships between human joints. The process of discretizing continuous pose data is performed by using K-mean clustering and vector quantized variational autoencoders, respectively. The music synchronization beat-aligned loss was optimized, and the best-tuned weight coefficients were found using two variants of the differential evolution algorithm, based on controlled mutation factors :mathcal{F} =log-sigmoid () and :mathcal{F} =rand(). The proposed architecture with :mathcal{F} =log sigmoid () achieves the lowest Fréchet inception distance (FIDk = 32.451, FIDg = 11.219) and music motion correlation of 0.341 demonstrating enhanced motion synthesis in comparison to existed state of art techniques. The mean fitness value of 6.0294 × 10–10 is obtained with an overall classification accuracy of 97.019% in 0.8431G FLOPs for differential evolution algorithm with :mathcal{F} log-sigmoid (). The framework may be utilized in AI-generated choreography, virtual dance instruction, and interactive entertainment.

This article introduces a novel hybrid optimization method, the Young’s Double-Slit Experiment-Differential Evolution (YDSE-DE) technique, which is based on the integration of the YDSE approach with the robust capabilities of Differential Evolution algorithms. This integration enables the precise estimation of parameters in photovoltaic (PV) models, enhancing the modeling and simulation of solar energy systems. Employing a comparative assessment strategy, the authors demonstrate that the proposed YDSE-DE algorithm outperforms existing methods such as Ant Lion Optimizer (ALO) and Sooty Tern Optimization Algorithm (STOA) in terms of accuracy and computational efficiency. The effectiveness of this method is confirmed through rigorous testing, which shows improvements in the root mean square error (RMSE) metrics by significant margins across single-diode, double-diode, and three-diode PV models. RMSE values obtained using the YDSE-DE algorithm for the single-diode, double-diode, and three-diode models are documented as 0.0059117, 0.0015218, and 0.0018409, respectively. The present study underscores the considerable potential of the YDSE-DE algorithm in augmenting the precision and effectiveness of photovoltaic system simulation and design. The results develop and enhance the existing methodologies for PV parameter estimation and can be applied to optimize other complex systems requiring reliable and precise simulation. The novelty and scientific contribution of this work lie in the methodical synthesis of physical principles from optics with evolutionary computation, presenting a significant advancement in the field of renewable energy technology.