Optimization Models Research Articles

A stochastic optimization procedure to design the fair aggregation of energy users in a Renewable Energy Community

Optimizing unit sizes and operation within a Renewable Energy Community (REC) can match intermittent renewable energy generation with variable user energy demands. These uncertain variables are often represented by pre-defined stochastic scenarios, without searching for the “best” scenarios and testing the optimization models with these scenarios. Moreover, little work both optimized RECs under uncertainty and distributed optimal life-cycle costs (investment and operation) among members. Thus, the objectives are: i) identifying the “best” set of stochastic scenarios of solar irradiance and user electricity demands and ii) assessing the accuracy of the “stochastic forecasts” of the total system costs and unit sizes, obtained by solving a stochastic programming model based on the “best” scenarios. The proposed novel procedure shifts the “present moment” back in time to split historical data into “past” and “future” periods used to identify the “best” scenarios and compare the “stochastic forecasts” with the utopic “perfect forecasts” based on the perfect knowledge of real data, respectively. The small errors between these forecasts in the optimal life-cycle costs (less than 2 %) and sizes (3–13 %) indicate good effectiveness of the suggested procedure. Also, the optimal life-cycle costs of “stochastic forecasts” are fairly distributed among users by applying the Shapley value mechanism.

How to optimize government R&D budget allocation? A review of recent studies and research gaps

Purpose Research and Development (R&D) activities are important for technological innovation and present opportunities for entrepreneurship. These activities depend on the flow of funding. This paper aims to review approaches used in R&D project selection and budget allocation. Design/methodology/approach This study conducts a systematic review, examining the content of 60 relevant papers (spanning 2000–2022) concerning public R&D budget allocation. The analysis focuses on allocation methodology, R&D output evaluation, budget allocation efficiency and the management of uncertainty in the allocation process. Findings The systematic review reveals different methods proposed for allocating government R&D budgets. These methods range from classical optimization, multi-criteria analysis and hierarchical analysis to techniques such as balanced scorecard, data envelopment analysis and analytic hierarchy process, including fuzzy approaches. Recent trends indicate an increase in the use of advanced optimization, integration and simulation algorithms. Performance indicators for reflecting R&D project outputs or goals can be categorized into four main groups: output (e.g. publications, patents, graduates), outcome, productivity (e.g. citations, patent references, articles and patents per capita) and sector-specific metrics. Practical implications Future research directions in government R&D budget allocation may include optimizing allocation to maximize social, economic and political benefits, developing ranking models, decision-making frameworks, simulations and evaluations of factors influencing allocation type and strategy. Additionally, there is a growing interest in novel budget allocation algorithms leveraging artificial intelligence and self-adjusting meta-heuristic algorithms. Originality/value The systematic review showed that some important research gaps in (government) R&D budget allocation could be considered in future studies; for example, long-term social, economic and political benefits in budget allocation optimization models, comprehensiveness of allocating government R&D budgets to universities, higher education and research institutes, R&D budget allocation to strategic technology development, e.g. renewable energy sector, supply chain issues and renewable energy value chain; new budget allocation algorithms based on artificial intelligence and self-adjusting meta-heuristic algorithms; methods for optimizing the structures of government budget allocation to R&D, considering executive and regulatory conflicts.

AI-Driven Innovation in CRM: Exploring the Potential of Generative Models for Customer Experience Optimization

AI-Driven Innovation in CRM: Exploring the Potential of Generative Models for Customer Experience Optimization

Bi-Level Operation Optimization and Performance Analysis for a Distributed Energy System with Energy Network

The increasing energy crisis and environmental problems promote the development of distributed energy systems (DESs) that utilize combined heat and power technology, renewable energy technology, and waste heat recovery technology to meet various load requirements. Although there is existing work and research on a variety of DESs, most previous studies tend to focus on energy production with little consideration of a distributed energy network and its energy loss, which results in large errors in energy-efficiency calculations and performance analyses. In this paper, a new DES model is proposed with full consideration of an energy network and the full use of solar energy, terrestrial heat, and exhaust gases. At the same time, an effective bi-level optimization method is also proposed for the daily operation of the DES in order to improve the system performance and benefits in the energy, the economy, and the environment. Specifically, the co-generation energy station and water heating network in the DES are optimized separately with two different optimization models. The first-level optimization model is used to seek the optimal values of water mass flow rate and water temperature of the water heating network, and the second-level optimization model is built to determine the optimal energy purchasing strategy and the optimal energy outputs of the co-generation energy station. In order to verify the advantages and effectiveness of the proposed system and method, a contrastive simulation study is undertaken to provide a comparison of the global optimization method. Simulation results show that energy loss, energy cost, and energy consumption of the DES using the bi-level optimization operation strategy only account for 22.51%, 33.42%, and 51.31% of the quantities of the global optimization method, respectively. The hourly curves of the optimal operating variables also demonstrate that the proposed bi-level optimization method can improve the operating stability of the DES better than the global optimization method.

Analysis of optimization models under different approaches to deal with uncertainty regarding pre-disaster planning in food bank supply chains

Purpose: Pre-positioning is a crucial choice in pre-disaster humanitarian logistics planning that consists of deciding in advance how much aid and where should it be located to enable effective and prompt operations in the case of an emergency. To support managers making such decisions, we propose four mathematical formulations that, considering the uncertainty on the demand to satisfy, seek to optimize aid prepositioning (before the event) and further distribution (after the event) in order to minimize unmet demand (MUD). The purpose of this paper is to evaluate and compare the performance of these formulations on a real case to discuss when and why should each approach be applied.Design/methodology/approach: The two first formulations adopt the cardinality-constrained (CC) approach to handle uncertainty. These formulations differ in their objective functions, the first formulation’s objective seeks to MUD, whilst the second incorporates equity in the way that demand is satisfied. The two remaining formulations are scenario-based (SB) and as in the previous two formulations, seek to MUD with and without equity considerations, respectively.Findings: Applying our formulations to a case study, we compare the differences between the solutions produced by the proposed formulations and the solutions that would have been produced without uncertainty (perfect information) to have a better understanding of their performance and their behavior. A discussion of the strengths and weaknesses of each model is provided to help managers choose the model that best suits their needs.Originality/value: The formulations are applied to a case study where a food bank is faced with the arrival of a hurricane in Mexico. As far as our knowledge, it is the first work in literature to deal with humanitarian logistics under a cardinality-constrained approach.

Optimization modelling for a sustainable closed-loop supply chain network using IoT: multiobjective metaheuristic algorithms

Optimization modelling for a sustainable closed-loop supply chain network using IoT: multiobjective metaheuristic algorithms

Oxidation Potential of 2,6-Dimethyl-1,4-dihydropyridine Derivatives Estimated by Structure Descriptors

Linear relationships, expressing the electrochemical properties of molecules as functions of structure, give insight into the associated electrochemical process and are a tool for prediction. Many biological activities rely on water-based dissociation, making electrochemical properties a bridge between structure and activity. Motivated by a previous study, a replica is made here on a different dataset in order to validate/invalidate the previously reported results. There are several methods for obtaining structure-based descriptors. Some of the methods have been devised to account for molecular topology, some to account for molecular geometry, and others to account for both. Two methods are involved here to derive structure-based descriptors and further obtain structure–property relationships (FMPI and ChPE). In order to express structure descriptors, both FMPI and ChPE express first the topology of the molecule, using the heavy atoms identity matrix and the heavy atoms adjacency matrix, both square symmetric matrices in the belief that symmetry is one major factor of molecular stability. A set of 2,6-dimethyl-1,4-dihydropyridine derivatives with oxidation peak potentials and coulometrically determined number of electrons experimental data is subjected to the search for structure–activity relationships. Even if the 2,6-dimethyl-1,4-dihydropyridine is a symmetric compound (of Cs point group), their derivatives are generally not symmetric (9 out of 24 are asymmetric). The dataset is subjected to descriptive and inferential statistics in order to filter out the most relevant structure–activity relationship. The geometry is built using three levels of theory (one from molecular mechanics and two others from density functionals, of which one accounts for the interaction with water as solvent). One challenge of picking one out of two reported measured values is dealt with by calculating the likelihood associated with the two choices. Relevant structure–activity models are extracted and discussed. The use of in vivo (in water, SM8 model) models in geometry optimization (from MMFF94 and B3LYP and to M06 + Water SM8) results in a precision gain, but this is, in most of the cases, not statistically significant, and this can be considered a negative result.

Hybrid quantum annealing decomposition framework for unit commitment

Quantum computing is an emerging and promising technology that has overwhelming quantum advantages compared to its classical counterparts. Unit commitment (UC) is a critical issue in the power system, and it becomes more challenging with the integration of intermittent renewable energy. Therefore, this paper proposes an innovative decomposition and coordination optimization framework to accelerate the solution of UC, in which the interaction between an adiabatic quantum computer and a classical computer is designed to harness the immense computational power of quantum computers effectively. First, decomposition methods considering the requirements of quantum computers are introduced to decompose UC into small-scale models. Then, the paper presents a quadratic unconstrained binary optimization modeling method to transform UC problems into the form of quantum computing. Furthermore, due to the limitations of quantum computing resources, a reductive variable technique is proposed to reduce the number of slack variables in the optimization model and ensure that it remains feasible for quantum computers. Case studies conducted in test systems with a quantum annealing simulator and a real quantum annealing computer illustrate the feasibility and effectiveness of the method and demonstrate its potential in the era of quantum computing.

Hybrid extreme gradient boosting regressor models for the multi-objective mixture design optimization of cementitious mixtures incorporating mine tailings as fine aggregates

The design of cementitious mixtures incorporating mine tailings as fine aggregates is a multi-objective optimization (MOO) problem, in which both the uniaxial compressive strength (UCS) and cost of the mixtures need to be considered simultaneously. Given that data-driven methods have shown promising results when solving similar MOO problems, this study developed an extreme gradient boosting regressor (XGBR) model on a dataset extracted from the literature to predict the UCS. Among the efforts taken to improve the models, a genetic algorithm (GA)-based XGBR model demonstrated the optimal prediction performance, with an R2 of 0.959. Next, the GA-XGBR model and a cost equation were used as objective functions in the MOO problem. The non-dominated sorting genetic algorithm with elite strategy (NSGA-II) was selected to solve the optimization problem. A case study was conducted, generating mixture designs that offered improved trade-offs between cost and UCS compared to experimental designs. Finally, a graphical user interface was developed to provide access to the prediction model and optimization method. Overall, this work can be used as a guide for optimal mixture designs as it facilitates informed decision-making before the actual applications.

Techno-economic analysis of arsenic (III) removal using spiral wound polyethersulfone nanofiltration membrane at pilot scale

In this work, the influence of initial solution pH on membrane performance was carried out for arsenic (III) removal using spiral wound polyethersulfone (HFN300) Nanofiltration membrane at pilot scale. The rejection percentage was found to increase from 49 % to 96 % while increasing initial solution pH from 4 to 10. Speciation of arsenic, solute-membrane affinity, electric-migration effect along with convection and diffusion were found to be dominant mechanisms for arsenic removal. Donnan Steric Pore Model was selected to estimate the membrane parameters including pore radius (0.313 nm), membrane thickness (2.17 μm), and membrane charge density (−3.56 mol/m3).Excellent agreements were found between the experimental and simulated values for all the studied pH range. Selected model works satisfactorily within 10 % of error for both rejection percentage and permeate flow. Economic feasibility study was carried out for the rural population (India) and total annualized cost was found to be USD 0.90/m3 that seems both reasonable and affordable for arsenic free drinking water. It can be concluded that annualized production cost was dependent on membrane lifespan, electricity tariff, per capita, population size, arsenic feed concentration, etc. The result obtained in the study suggests the feasibility of using the NF process in removing arsenic from contaminated water. Overall, the study provides valuable insights into the mechanisms governing arsenic (III) removal through nanofiltration process using spiral wound polyethersulfone nanofiltration membrane at pilot scale, demonstrating the effectiveness of pH optimization and membrane modelling in improving removal efficiency, which can contribute to the development of cost-effective solutions for safe drinking water in arsenic-affected rural areas.

Machine learning-based rainfall forecasting in real-time optimal operation of urban drainage systems

This paper presents a rainfall forecast-based real-time optimal operation model of urban drainage systems (UDSs) investigating the role of rainfall forecasts in predictive real-time operation of UDSs. Two rainfall forecast models, namely long short-term memory (LSTM) and optimized-by-PSO (particle swarm optimization) support vector regression, are linked to an integrated simulation–optimization model. For any given operation policies, the rainfall-runoff stormwater management model (SWMM) simulates the urban system’s operation under flooding conditions in which the operation policies are optimized by the metaheuristic harmony search (HS) optimizer. The objective function of the HS algorithm is minimizing flood volumes at control points. The presented approach is therefore a dynamic predictive real-time operation (PRTOP) model that continuously updates rainfall forecasts and operational decisions during flood events. The performance of the developed forecast-based integrated SWMM-HS simulation–optimization modeling framework is tested through its application in a real urban system case study located in Tehran, Iran. Results demonstrate that a notable reduction in flood volume (11.8%) is achieved by using the LSTM-driven rainfall forecasts in the proposed PRTOP model compared to a reactive real-time operation (RTOP) model not utilizing rainfall forecasts.

Hybrid CNC–MXene Nanolubricant for Tribological Application: Characterization, Prediction, and Optimization of Thermophysical Properties Evaluation

In the present work, hybrid Cellulose Nanocrystal–MXene (CNC–MXene) nanolubricants were prepared via a two-step method and investigated as potential heat-transfer hybrid nanofluids for the first time. CNC–MXene nanolubricants were synthesized via a two-step method by varying the weight percentage of CNC–MXene nanoparticles (ranging from 0.01 to 0.05 wt%) and characterized using Fourier-Transform Infrared Spectroscopy and TGA (Thermogravimetric Analysis). Response surface methodology (RSM) was used in conjunction with the miscellaneous design model to identify prediction models for the thermophysical properties of the hybrid CNC–MXene nanolubricant. Minitab 18 statistical analysis software and Response Surface Methodology (RSM) based on Central Composite Design (CCD) were utilized to generate an empirical mathematical model investigating the effect of concentration and temperature. The analysis of variance (ANOVA) results indicated significant contributions from the type of nanolubricant (p < 0.001) and the quadratic effect of temperature (p < 0.001), highlighting non-linear interactions that affect viscosity and thermal conductivity. The findings showed that the predicted values closely matched the experimental results, with a percentage of absolute error below 9%, confirming the reliability of the optimization models. Additionally, the models could predict more than 85% of the nanolubricant output variations, indicating high model accuracy. The optimization analysis identified optimal conditions for maximizing both dynamic viscosity and thermal conductivity. The predicted optimal values (17.0685 for dynamic viscosity and 0.3317 for thermal conductivity) were achieved at 30 °C and a 0.01% concentration, with a composite desirability of 1. The findings of the percentage of absolute error (POAE) reveal that the model can precisely predict the optimum experimental parameters. This study contributes to the growing field of advanced nanolubricants by providing insights into the synergistic effects of CNC and MXene in enhancing thermophysical properties. The developed models and optimization techniques offer valuable tools for tailoring nanolubricant formulations to specific tribological applications, potentially leading to improved efficiency and durability in various industrial settings.

Optimal resilience-based restoration plan for disrupted interdependent infrastructure networks under uncertainty

The functioning of vital services in a society depends heavily on the proper and continuous operation of critical infrastructure networks, including water, transportation, power, natural gas, and telecommunication. Such networks are essential for maintaining the economy, security, and overall quality of life. However, since these critical networks are interdependent; hence, they are susceptible to various types of disruptions caused by natural disasters, failures, or malicious activities, leading to varying degrees of performance impacts that could directly affect people's daily lives. Thus, a disruption to one network can quickly spread to others, resulting in significant impacts on daily life. Therefore, restoring such disrupted networks in a timely manner is crucial to ensure a prompt recovery and minimize the impact on daily life. In this work, we consider the problem of restoring a set of interdependent infrastructure networks following a disruptive event. We study two essential factors that can accelerate the recovery process: (i) the restoration facilities location, which are considered as dispatching points for restoration crews, and (ii) the routes of the restoration crews within the interdependent networks. Accordingly, we formulate two optimization models using mixed integer programming. The first model, restoration facility location model, finds the optimal location for restoration facilities within interdependent infrastructure networks considering uncertainty in disruptions. The second model, restoration crew routing model, determine the optimal routes for restoration crews within the interdependent infrastructure networks considering: (i) travelled distance, (ii) restoration time required for disrupted network components, and (iii) different importance of network components. The two optimization models are solved through generated interdependent power and water networks to demonstrate their effectiveness.

Large-scale consensus in incomplete social network with non-cooperative behaviors and dimension reduction

Obtaining a consensus solution is a formidable challenge for large-scale decision-making (LSDM) in social network (SN). The reaching of large-scale consensus in SN is hindered by serious difficulties, including complex decision information, incomplete social relations, a multitude of decision-makers (DMs), and non-cooperative behaviors. This paper introduces a novel three-stage consensus framework that systematically addresses these challenges by data preprocessing, dimension reduction, and optimization modeling. Firstly, the cloud model is applied to convert the probabilistic linguistic information into numerical information, facilitating computational analysis. Meanwhile, an improved t-norm trust propagation method that incorporates the impact of opinion similarity is developed, ensuring the completeness of SN. Secondly, an improved Louvain algorithm is designed to divide large group into cohesive subgroups, enhancing the manageability of LSDM. On this basis, a three-stage consensus optimization that considers non-cooperative behaviors is proposed, which boasts threefold benefits: (i) Assures the synchronous achievement of local and global consensus. (ii) Implements self-adaptive management mechanism of non-cooperative behaviors. (iii) Provides acceptable adjusted opinions for subgroups and DMs. Finally, detailed numerical experiments and comparative analyses are given to demonstrate the effectiveness of the proposed method.

Multi-temperature capable enhanced bidirectional long short term memory-multilayer perceptron hybrid model for lithium-ion battery SOC estimation

In this study, we propose a novel hybrid modeling framework for State of Charge (SOC) estimation across a broad temperature spectrum. First, we build a hybrid model to optimize stacked layers of stacked bidirectional long short term memory networks by introducing dropout mechanisms. At the same time, we also optimize the traditional multi-layer perceptron model to ResMLP, which is improved by introducing residual linkage, and then integrate the two optimization models. Finally, the synergistic effect and attention mechanism of genetic algorithm and particle swarm optimization are used to optimize its parameters. We then rigorously tested the model on nine datasets, including HPPC, DST and BBDST, at different temperatures of 5 °C, 15 °C and 35 °C. Using MAE, RMSE and MAXE benchmarks, our research results show that the proposed hybrid model outperforms the benchmark algorithm, achieving significantly enhanced performance and higher accuracy, and the maximum SOC estimation error is kept below 4.53 %. In addition, experimental evaluation at different temperatures shows the robustness and adaptability of the proposed algorithm.

Cost-effective and optimal pathways to selecting building microgrid components – The resilient, reliable, and flexible energy system under changing climate conditions

Amidst the changing climate conditions, electricity retailers-led demand response programs and the emergency backup power for possible grid outages are often discussed in isolation, although several underlying linkages exist, which are important for how the existing building stock evolves with the energy transition. This study navigates through the linkages while investigating the levelized cost of electricity (LCOE)-based building microgrid components and undertakes a comparative analysis of energy optimisation models with and without emergency diesel generators. It also examines the building energy system’s resilience, reliability, and flexibility by using OpenStudio and HOMER Grid to develop energy simulation and optimisation models and other probabilistic models for a chosen building archetype in Sydney, Australia. This study finds that excluding the emergency diesel generator will require a larger battery storage system, increasing LCOE from A$0.17/kWh to A$0.20/kWh across climate scenarios. However, the larger battery storage system increases the probability of surviving outages (> 4 h) by around 10 % across climate scenarios. The LCOE increases up to 45 % for outages up to 8 h across climate scenarios and up to 85 % for outages up to 12 h. Additionally, for the building archetype, the probability of grid purchase (below the current average) is 0.78 in 2030, which drops to 0.55 in 2050. The probability of grid sales (above the current average) also drops from 0.69 to 0.46. Thus, while the narratives around the building energy system’s flexibility overstate the electricity exchange between the building and the grid, this study finds that increasing the on-site production utilisation rate and larger battery throughput contributes to demand response application and flexibility.

Scheme design and assessment of hybrid pump feed system with energy management for throttleable liquid rocket engine

Currently, there is considerable emphasis on the electric pump-fed cycle for liquid engine, primarily due to its design simplicity. However, its development is hindered by the underdeveloped state of power battery technology. Drawing inspiration from hybrid power technology used in electric vehicles and turbochargers, a hybrid pump feed system for throttleable engines is originally proposed as a promising solution. This system integrates the electric motor into the gas generator cycle, with several topologies evaluated. The parallel configuration featuring a mid-motor is selected for its compact structure, efficient power-splitting and energy recovery. Additionally, customized energy management strategies and optimization models are developed to effectively allocate power throughout the operational processes of liquid engines. A comparative analysis of four engine cycles is conducted under the typically variable-thrust mission. The results indicate that attributed to the conservation of turbo-gas and battery energy, the optimized hybrid pump achieves a reduction of 2.39 % compared to the turbopump and 7.15% to the electric pump in total mass. Adaptability assessment further indicates that the mass advantage of the hybrid pump system is more significant during prolonged engine burning and deep throttling. Specific working conditions are found in which the system prefers electric-motor driving or regenerating turbine energy. Although energy-recovery results in the system efficiency decrease, it serves to lower energy demand of battery pack, thus easing the burden on cell thermal management and structural design. This study provides a practical design framework for hybrid pump-fed rocket engines in future variable-thrust missions.

Improving Bayesian Optimization via Hierarchical Variation Modeling for Combinatorial Experiments Given Limited Runs Guided by Process Knowledge

Active learning based on Bayesian optimization (BO) is a popular black-box combinatorial search method, particularly effective for autonomous experimentation. However, existing BO methods did not consider the joint variation caused by the process degradation over time and input-dependent variation. The challenge is more significant when the affordable experimental runs are very limited. State-of-the-art approaches did not address allocating limited experimental runs that can jointly cover (1) representative inputs over large search space for identifying the best combination, (2) replicates reflecting the true input-dependent testing variation, and (3) process variations that increase over time due to process degradation. This paper proposed Empirical Bayesian Hierarchical Variation Modeling in Bayesian Optimization (EHVBO) guided by the process knowledge to maximize the exploration of potential combinations in sequential experiments given limited experimental runs. The method first mitigates the process degradation effect through generalized linear modeling of grouped variations, guided by the knowledge of the re-calibration cycle of process conditions. Then, EHVBO introduces an empirical Bayesian hierarchical model to reduce the replicates for learning the input-dependent variation, leveraging the process knowledge of the common structure shared across different testing combinations. This way can reduce the necessary replicates for each input condition. Furthermore, the paper developed a heuristics-based strategy incorporated in EHVBO to improve search efficiency by selectively refining the search space over pivotal regions and excluding less-promising regions. A case study based on real experimental data demonstrates that the proposed method outperforms testing results from various optimization models.

Population Pharmacokinetic Modeling and Simulation for Dose Optimization of GB-5001, a Long-Acting Intramuscular Injection of Donepezil, in Healthy Participants.

GB-5001 is an intramuscular (IM) formulation of donepezil under development for the treatment of Alzheimer's disease. The objective of this study was to develop a population pharmacokinetic (PK) model for donepezil in both IM and oral formulations, and to optimize the IM dosage of GB-5001 using bioequivalence (BE) simulation. A population PK model of donepezil was developed using NONMEM. It was based on plasma concentration data from a Phase 1 dose escalation study, which involved a single administration of donepezil IM formulation at doses of 70, 140, and 280mg, and the oral formulation at 10mg. The model was evaluated based on goodness-of-fit plots, conditional weighted residuals, visual predictive checks, and bootstrapping. BE simulations were conducted using a parallel design between various doses of the IM formulation and the 10-mg dose of oral formulation. The PKs of donepezil were best described by a two-compartment model, which incorporated distinct absorption compartments for the IM (dual first-order absorption and simultaneous zero-order absorption with lag time) and oral (first-order absorption with lag time) formulations. Based on the simulation results, an IM dosage range of 210-215 mg in a sample size of over 92 was estimated to achieve a success rate of approximately 80% for BE. The population PK model well explained the PKs of donepezil following administration of both the IM and oral formulations. This model could be applied for the design and dose selection of future BE trials. ClinicalTrials.gov identifier, NCT05525780.

On the prediction and optimization of the flow boiling heat transfer in mini and micro channel heat sinks

The most common methods for predicting flow boiling heat transfer in mini/micro channels-based heat sinks rely on semi-empirical correlations derived from experimental data. However, these correlations are often limited to specific testing conditions. This study proposes a novel approach using deep learning and genetic algorithms (GA) to predict and optimize refrigerants' flow boiling heat transfer coefficients (FBHTC) in mini/microchannels-based heat sinks. The dataset used in this study includes FBHTC observations from the literature for seven refrigerants (R1234yf, R1234ze, R134A, R513A, R410A, R22, and R32). The optimal input parameters identified include hydraulic diameters ranging from 1 to 7 mm, saturation temperature from 0 to 20 °C, flow qualities from 0.006 to 0.972, heat flux from 3 to 78.8 kW/m2, and mass fluxes between 100 and 1200 kg/m2s. Gradient-boost regression trees were employed to develop the deep learning and GA models for accurate estimation and optimization. Correlation analysis and feature engineering selected the most influential parameters to construct a precise and simple model. The results demonstrate that the models could estimate refrigerants' FBHTC with high accuracy, achieving an R2 of 0.988 and a mean squared error (MSE) of 0.05%. The GA-based method effectively optimized the FBHTC for each refrigerant by determining the appropriate input parameters, including the saturation temperature, heat and mass fluxes, quality, and hydraulic diameter. Additionally, a parametric analysis using explainable artificial intelligence was conducted to interpret the impact of each input parameter on the FBHTC.