Parameter Optimization Algorithm Research Articles

High-Order Grid-Connected Filter Design Based on Reinforcement Learning

In grid-connected inverter systems, grid-connected filters can effectively eliminate harmonics. High-order filters perform better than conventional filters in eliminating harmonics and can reduce costs. For high-order filters, the use of multi-objective optimization algorithms for parameter optimization presupposes that the circuit structure must be known. To realize the design of the filter structure and related circuit parameters that meet the requirements of the grid-connected inverter system during the design process, this paper proposes a reinforcement learning (RL) method for designing higher-order filters. Our approach combines key domain knowledge with the characteristics of structural changes to obtain some constraints, which are then processed to obtain reward and are incorporated into RL strategy learning to determine the optimal structure and corresponding circuit parameters. The proposed method realizes the simultaneous design of parameters and structures in filter design, which greatly improves the efficiency of filter design. Simulation results for the corresponding grid-connected system setup show that the grid-connected filter designed by our method demonstrates a good performance in terms of filter dimension, harmonic rejection, and total harmonic distortion.

Metaheuristic algorithm for parametric optimization of liquid nitrogen pumping in hydrocarbon and allied fluids piping systems

Metaheuristic algorithm for parametric optimization of liquid nitrogen pumping in hydrocarbon and allied fluids piping systems

Neural Network Prediction of Locomotive Engine Parameters Based on the Dung Beetle Optimization Algorithm and Multi-Objective Optimization of Engine Operating Parameters

Neural Network Prediction of Locomotive Engine Parameters Based on the Dung Beetle Optimization Algorithm and Multi-Objective Optimization of Engine Operating Parameters

Ultra-short-term Forecasting Study of Power Load in Mega Steel Industry Based on Multi-stage Modeling

Background: In the large-scale steel industry, significant power load variability, especially during processes like steel smelting, poses challenges to power system safety. Although there is an abundance of research and patents related to load forecasting, studies and patents specifically addressing large industrial load forecasting are sparse. Hence, accurate ultra-short-term load forecasting becomes particularly crucial. Objective: This study proposes an innovative method for ultra-short-term load forecasting to improve prediction accuracy during peak periods and mitigate risks in high-load conditions. Methods: We introduce an LSTM-XGBoost model enhanced by a random forest network and an improved grey wolf optimization algorithm (IGWO) for feature selection and parameter optimization, respectively. Results: Compared to other advanced models, our method demonstrates superior performance across key indicators such as MAPE (1.93%), RMSE (220.81), and R2 coefficient (0.99), and the prediction error is lower during both peak and off-peak periods. For instance, the proposed model achieved a MAPE improvement of over 25% compared to traditional models. Validation with data from multiple time periods confirms the model's accuracy and robustness. Conclusion: The proposed forecasting method effectively tackles load fluctuations in the steel industry, supporting safe and economical power system operations. Future research will aim to further improve peak identification accuracy and enable continuous adaptive learning.

Optimal selection of machine learning algorithms for ciprofloxacin prediction based on conventional water quality indicators.

The long-term presence of antibiotics in the aquatic environment will affect ecology and human health. Techniques for determining antibiotics are often time-consuming, labor-intensive and costly, and it is desirable to seek new methods to achieve rapid prediction of antibiotics. Many scholars have shown the effectiveness of machine learning in water quality prediction, however, its effectiveness in predicting antibiotic concentrations in the aquatic environment remains inconclusive. Given that conventional water quality indicators directly or indirectly influence antibiotic concentrations, we explored the feasibility of predicting ciprofloxacin (CFX) concentrations based on conventional water quality indicators with the help of three commonly used machine learning algorithms and two parameter optimization algorithms. Then, we evaluated and determined the best model using four commonly used model performance evaluation metrics. The evaluation results showed that the generalized regression neural network (GRNN) model optimized by particle swarm optimization (PSO) had the best prediction among all the models under the conditions of six input variables, namely COD, NH4+-N, DO, WT, TN, and pH. The performance evaluations were R2= 0.936, NSE= 0.915, RMSE= 3.150 ng/L, and MAPE= 30.909 %. Overall, the CFX prediction models met the requirements for antibiotic concentration prediction accuracy, offering a potential indirect method for predicting antibiotic concentrations in water quality management.

Eliminating Network Depth: Genetic Algorithm for Parameter Optimization in CNNs

Eliminating Network Depth: Genetic Algorithm for Parameter Optimization in CNNs

Surrogate-assisted decomposition multi-objective evolutionary algorithm for parameters optimization in polyester fiber polymerization process

Surrogate-assisted decomposition multi-objective evolutionary algorithm for parameters optimization in polyester fiber polymerization process

Gradient Method with Step Adaptation

The paper solves the problem of constructing step adjustment algorithms for a gradient method based on the principle of the steepest descent. The expansion of the step adjustment principle, its formalization and parameterization led the researchers to gradient-type methods with incomplete relaxation or over-relaxation. Such methods require only the gradient of the function to be calculated at the iteration. Optimization of the parameters of the step adaptation algorithms enables us to obtain methods that significantly exceed the steepest descent method in terms of convergence rate. In this paper, we present a universal step adjustment algorithm that does not require selecting optimal parameters. The algorithm is based on orthogonality of successive gradients and replacing complete relaxation with some degree of incomplete relaxation or over-relaxation. Its convergence rate corresponds to algorithms with optimization of the step adaptation algorithm parameters. In our experiments, on average, the proposed algorithm outperforms the steepest descent method by 2.7 times in the number of iterations. The advantage of the proposed methods is their operability under interference conditions. Our paper presents examples of solving test problems in which the interference values are uniformly distributed vectors in a ball with a radius 8 times greater than the gradient norm.

AI-Driven Optimization of Renewable Energy Storage Systems in Smart Cities Using Improved Binary Genetic Algorithm

Today, the utilization and management of renewable energy have become integral to the development of smart cities. This paper explores the application of Artificial Intelligence (AI) in analyzing energy storage and renewable energy systems within smart city contexts. We introduce a joint optimization method that combines two-stage input feature selection and parameter training in traditional prediction models. This method integrates physical, statistical, and AI modeling techniques. Additionally, we conduct comprehensive experiments on feature input selection using a Binary Genetic Algorithm (BGA) and two-stage model parameter optimization based on a Natural Evolution Strategies (NES). The experimental results provide strong support for the conclusions of this study. Finally, we simulate the full charge process of energy storage in a smart city and predict the full charge data for batteries B0005, B0006, B0007, and B0029. The results indicate that the NES-BGA-ELM approach outperforms the BGA-ELM method in terms of error reduction.

Optimization of Time-Varying Temperature Profiles for Enhanced Beer Fermentation by Evolutive Algorithms

Beer is one of the most popular alcoholic beverages globally, leading to continuous efforts to enhance its production methods. Raw materials and the production process are crucial in the brewing industry, with fermentation being a vital stage that significantly impacts beer quality. The aim of this study is to optimize the beer fermentation process by maximizing the ethanol concentration while minimizing species that adversely affect the organoleptic properties of beer. A novel optimization approach has been developed to derive an optimal, smooth, and continuous temperature profile that can be directly applied in real-world processes. This method integrates Fourier series and orthogonal polynomials for control action parameterization, in combination with evolutionary algorithms for parameter optimization. A key advantage of this methodology lies in its ability to handle a reduced parameter set efficiently, resulting in temperature profiles that are continuous and differentiable. This feature eliminates the need for post-smoothing and is particularly advantageous in biotechnological applications, where abrupt changes in temperature could negatively affect the viability of microorganisms. The optimized profiles not only enhance fermentation efficiency, but also improve the ethanol yield and reduce undesirable flavor compounds, providing a substantial improvement over current industrial practices. These advancements present significant potential for improving both the quality and consistency of beer production.

Explainable Thyroid Cancer Diagnosis Through Two-Level Machine Learning Optimization with an Improved Naked Mole-Rat Algorithm.

Modern technologies, particularly artificial intelligence methods such as machine learning, hold immense potential for supporting doctors with cancer diagnostics. This study explores the enhancement of popular machine learning methods using a bio-inspired algorithm-the naked mole-rat algorithm (NMRA)-to assess the malignancy of thyroid tumors. The study utilized a novel dataset released in 2022, containing data collected at Shengjing Hospital of China Medical University. The dataset comprises 1232 records described by 19 features. In this research, 10 well-known classifiers, including XGBoost, LightGBM, and random forest, were employed to evaluate the malignancy of thyroid tumors. A key innovation of this study is the application of the naked mole-rat algorithm for parameter optimization and feature selection within the individual classifiers. Among the models tested, the LightGBM classifier demonstrated the highest performance, achieving a classification accuracy of 81.82% and an F1-score of 86.62%, following two-level parameter optimization and feature selection using the naked mole-rat algorithm. Additionally, explainability analysis of the LightGBM model was conducted using SHAP values, providing insights into the decision-making process of the model.

Dynamic model adjustment of a honeycomb sandwich panel using response surface methods and parametric optimization algorithms

The aeronautical and aerospace industries have been evolving in technologies that improve the efficiency of their structures, where, in this context, sandwich panels with honeycomb cores are one of the frequently employed technologies in aircraft and nano satellites. This type of structure presents superior mechanical properties compared to traditional panels, as well as a low structural weight due to its porous core. However, they exhibit complex characteristics both geometrically and behaviorally, requiring the use of more sophisticated computational tools such as finite element analysis packages. Precise numerical models are required to allow for reliable analysis during the design process of aerospace structures. The most commonly employed approach is to create a numerical model and refine it through adjustments using experimental data. This study presents the modal numerical analysis of a sandwich panel with a honeycomb core using finite elements and proposes an adjustment of the obtained numerical model. The panel is composed of Aluminum 2024 skins and an Aluminum 5056 honeycomb core. This panel had its vibration modes and frequencies previously identified through experimental tests. Once the numerical and experimental values are compared, a numerical-experimental adjustment is proposed using the optimization algorithms NLPQL, MISQP, and Genetic, which are applied through two methodologies, with and without the use of a metamodel, where the metamodel is constructed using the kriging algorithm. The optimization problem was formulated considering the minimization of a function defined as the sum of squared differences between experimental and numerical frequencies. The ANSYS Workbench software package was utilized for the numerical modeling, which also provided the necessary optimization algorithms. The results demonstrate good agreement with the experimental data, and the numerical adjustments made using the optimization algorithms proved to be effective.

Effects of reducing redundant parameters in parameter optimization for symbolic regression using genetic programming

Gradient-based local optimization has been shown to improve results of genetic programming (GP) for symbolic regression (SR) – a machine learning method for symbolic equation learning. Correspondingly, several state-of-the-art GP implementations use iterative nonlinear least squares (NLS) algorithms for local optimization of parameters. An issue that has however mostly been ignored in SR and GP literature is overparameterization of SR expressions, and as a consequence, bad conditioning of NLS optimization problem. The aim of this study is to analyze the effects of overparameterization on the NLS results and convergence speed, whereby we use Operon as an example GP/SR implementation. First, we demonstrate that numeric rank approximation can be used to detect overparameterization using a set of six selected benchmark problems. In the second part, we analyze whether the NLS results or convergence speed can be improved by simplifying expressions to remove redundant parameters with equality saturation. This analysis is done with the much larger Feynman symbolic regression benchmark set after collecting all expressions visited by GP, as the simplification procedure is not fast enough to use it within GP fitness evaluation. We observe that Operon frequently visits overparameterized solutions but the number of redundant parameters is small on average. We analyzed the Pareto-optimal expressions of the first and last generation of GP, and found that for 70% to 80% of the simplified expressions, the success rate of reaching the optimum was better or equal than for the overparameterized form. The effect was smaller for the number of NLS iterations until convergence, where we found fewer or equal iterations for 51% to 63% of the expressions after simplification.

Optimization of Micro-EFP charge structures based on PSO-SVM algorithm

Abstract To achieve a high-performance design of micro Explosive Formed Projectile (EFP) structures, a method based on the combination of the Particle Swarm Optimization (PSO) algorithm and Support Vector Machine (SVM) with numerical simulation technology is proposed for the optimization and design of EFP structures. The efficient optimization of the EFP structure is achieved by establishing the mapping relationship between the EFP structure and performance and using the PSO algorithm for parameter optimization. The results show that the optimal parameters of the structure are R out/D k=1.347, R in/D k=1.552, δ/D k=0.045, σ/D k=0.060, and N (L/D k) =1.082. Under this optimization option, the EFP velocity is predicted to be 2362.6 m/s, and the numerical simulation results are 2385 m/s, with an error of only 0.95%. An effective optimization approach for the design of micro explosively formed projectiles is provided.

A Novel ANN-PSO Method for Optimizing a Small-Signal Equivalent Model of a Dual-Field-Plate GaN HEMT.

This study introduces a novel method that integrates artificial neural networks (ANNs) with the Particle Swarm Optimization (PSO) algorithm to enhance the efficiency and precision of parameter optimization for the small-signal equivalent model of dual-field-plate GaN HEMT devices. We initially train an ANN model to predict the S-parameters of the device, and subsequently utilize the PSO algorithm for parameter optimization. Comparative analysis with the NSGA2 and DE algorithms, based on convergence speed and accuracy, underscores the superiority of the PSO algorithm. Ultimately, this ANN-PSO approach is employed to automatically optimize the internal parameters of a 4 × 250 μm dual-field-plate GaN HEMT equivalent circuit model within the frequency range of 1-18 GHz. The method's effectiveness under varying bias conditions is validated through comparison with traditional physical formula analysis methods. The results demonstrate that the ANN-PSO method significantly enhances the automation and efficiency of parameter optimization while maintaining model accuracy, providing a reference for the optimization of other device models.

Differential evolution-based time domain decomposition method for multi-impact vibration signals of reciprocating machinery

Abstract To solve the problem of extracting the impact component from the complex time-domain vibration signal of reciprocating machinery vibration signals, a differential evolution (DE)-based time domain decomposition method is proposed to achieve adaptive extraction of impact components. The method establishes new decomposition window containing three adjustment parameters to adapt to multiple forms of impact components. Furthermore, with the optimization objectives of minimizing reconstruction loss, amplitude moment loss, and similarity loss, a decomposition parameter optimization algorithm based on DE is established to achieve the optimization process of decomposition parameters. The results of processing simulated and actual vibration signals of diesel engines show that the new method can adaptively and accurately identify the impact component and impact time center in the vibration component, with a signal reconstruction loss of less than 2.5% and a decomposition time of only 54.1 s.

Prediction of building HVAC energy consumption based on least squares support vector machines

Air conditioning, as an essential appliance in daily life, has the function of ensuring comfortable room temperature, but it is also accompanied by a large amount of power consumption. Consequently, the study suggests an energy consumption prediction model based on improved genetic algorithm—least squares support vector machine—to accurately predict the energy consumption of building heating, ventilation, and air conditioning. This model uses the improved genetic algorithm for regularization parameter and kernel parameter optimization to prevent overfitting and underfitting issues. According to the testing results, the least squares support vector machine, an upgraded genetic algorithm, may accomplish convergence faster than other algorithms, taking only 0.2 milliseconds to finish. In addition, the average relative error of the improved genetic algorithm- least squares support vector machine did not exceed 0.6%. In the energy consumption prediction for the whole year of 2022, the average error of the improved genetic algorithm-least squares support vector machine was only 2.0 × 106 kWh, and the prediction accuracy could reach up to 97.2%. The above outcomes revealed that the energy consumption prediction model can accurately predict the air conditioning energy consumption, which provides a strong support for the control and optimization of the air conditioning system.

Wingsuit Flying Search Optimization Algorithm Strategy for Combinatorial T-Way Test Suite Generation

Advancement in software development has resulted in complex software applications that encompass various functional and non-functional requirements. Such a complex system usually consists of many inputs either directly from users or from other connected systems or devices. Here, there is a potential for the system to go wrong due to certain combinations of inputs. Combinatorial Testing (or T-Way Testing) is effective in tackling the issue. Numerous studies have proposed strategies in generating T-Way test suite, and current trend indicates that researchers often incorporate metaheuristic algorithms in their proposed strategies. Many recent studies employ parameter optimization algorithms such as (Whale Optimization Algorithm, Particle Swarm Optimization, Gravitational Optimization Algorithm) in generating an optimized T-Way test suite. Often, researchers need to tune the parameters involved in the algorithm before the algorithm can be used for test suite generation. Since the system under test (SUT) can come in various numbers of input, it is impossible to find a single best value for every algorithm parameter. As a result, this paper proposed a T-Way Test suite generator utilizing Wingsuit Optimization Algorithm (a parameter free optimization algorithm) for combinatorial test suite generation. The algorithm learnt by itself as the optimization process progresses and hence eliminates the need for control parameters. Statistical analysis shows that WFS produces a smaller test suite compares to most T-Way strategies and in some cases, the difference between test suite size produce by WFS and other T-Way strategies are insignificant.

Data-driven parameter identification of an equivalent mechanical model for large amplitude liquid sloshing

The accurate extraction of model parameters is vital to ensure that the equivalent mechanical model can precisely describe the dynamic behavior of large amplitude liquid sloshing. In this paper, the parameters of the moving pulsating ball model (MPBM), which is an equivalent mechanical model used to represent the large amplitude liquid sloshing, are extracted using a combination of computational fluid dynamics (CFD), experimental data, and data-driven algorithms. The arbitrary Lagrangian-Eulerian finite element method (ALE-FEM) is adopted to simulate three-dimension large amplitude liquid sloshing in the tank with high precision. The calculated results from the presented algorithm are compared with the experimental data to verify the reliability and validity. The typical parameters required by the MPBM mainly include equivalent sloshing mass, equivalent ball radius, and damping coefficient. These parameters are extracted by using a data-driven parameter optimization algorithm, which is based on the numerical calculation results of ALE-FEM. The accuracy of the MPBM with the equivalent model parameters extracted by using data-driven parameter optimization algorithm is investigated under two types of excitations: harmonic excitation and step excitation. The results show that the MPBM with equivalent model parameters extracted by a data-driven parameter optimization algorithm can precisely imitate the large amplitude liquid sloshing behavior and the method presented can provide significant reference for the overall design of spacecraft dynamics system.

An Offset Parameter Optimization Algorithm for Denoising in Photon Counting Lidar.

In the case of a weak signal from a photon counting lidar and strong noise from the solar background, the signal is completely submerged by noise, potentially resulting in the appearance of multiple peaks in the denoising algorithm of photon counting entropy. Consequently, a clear distinction between the signal and noise may become challenging, leading to significant fluctuation in the ranging error. To solve this problem, this paper proposes an improved offset parameter optimization algorithm under the framework of photon counting entropy, aiming to effectively eliminate peak interference caused by noise and enhancing ranging accuracy. The algorithm includes two aspects. First, we introduce the solar irradiance prediction of an MLP network and least squares linear conversion to accurately estimate the noise rate of the solar background noise. Then, we propose the offset parameter optimization method to effectively mitigate the interference caused by noise. In simulation and experimental analyses, the ranging error of our proposed method is within 5 and 30 cm, respectively. Compared with the denoising method of photon counting entropy, the average ranging error is increased by 81.99% and 73.76%. Furthermore, compared to other anti-noise methods, it exhibits superior ranging capability.