Genetic Algorithm Search Research Articles

A hybrid genetic algorithm with type-aware chromosomes for Traveling Salesman Problems with Drone

There are emerging transportation problems known as the Traveling Salesman Problem with Drone (TSPD) and the Flying Sidekick Traveling Salesman Problem (FSTSP) that involve using a drone in conjunction with a truck for package delivery. This study presents a hybrid genetic algorithm for solving TSPD and FSTSP by incorporating local search and dynamic programming. Similar algorithms exist in the literature. Our algorithm, however, considers more sophisticated chromosomes and less computationally complex dynamic programming to enable broader exploration by the genetic algorithm and efficient exploitation through dynamic programming and local search. The key contribution of this paper is the discovery of how decision-making processes for solving TSPD and FSTSP should be divided among the layers of genetic algorithm, dynamic programming, and local search. In particular, our genetic algorithm generates the truck and the drone sequences separately and encodes them in a type-aware chromosome, wherein each customer is assigned to either the truck or the drone. We apply local search to each chromosome, which is decoded by dynamic programming for fitness evaluation. Our new algorithm is shown to outperform existing algorithms on most benchmark instances in both quality and time. Our algorithms found the new best solutions for 538 TSPD instances out of 920 and 74 FSTSP instances out of 132.

Reservoir characterisation using hybrid optimisation of genetic algorithm and pattern search to estimate porosity and impedance volume from post-stack seismic data: A case study

Reservoir characterisation using hybrid optimisation of genetic algorithm and pattern search to estimate porosity and impedance volume from post-stack seismic data: A case study

Waiting Strategy for the Dynamic Meal Delivery Routing Problem with Time-Sensitive Customers Using a Hybrid Adaptive Genetic Algorithm and Adaptive Large Neighborhood Search Algorithm

In this paper, we study the dynamic meal delivery routing problem (MDRP) with time-sensitive customers. The multi-objective MDRP optimization model is developed to maximize customer satisfaction and minimize delay penalty cost and riding cost. To solve the dynamic MDRP, a novel waiting strategy is proposed to divide the dynamic problem into a series of static subproblems. This waiting strategy utilizes the decision threshold to determine rerouting points based on the number of dynamic meal orders. Meanwhile, time-sensitive priority is introduced to accelerate assignment and routing decisions for orders from customers with high time sensitivity. For each static subproblem, a hybrid AGA–ALNS algorithm that incorporates the adaptive genetic algorithm and adaptive large neighborhood search is developed to improve both the global and local search capabilities of the genetic algorithm. We validate the performance of the proposed waiting strategy and the AGA–ALNS algorithm through numerical instances. In addition, managerial insights are obtained from sensitivity analysis experiments.

Q-learning guided algorithms for bi-criteria minimization of total flow time and makespan in no-wait permutation flowshops

Combining Deep Reinforcement Learning and meta-heuristic techniques represents a new research direction for enhancing the search capabilities of meta-heuristic methods in the context of production scheduling. Q-learning is a prominent reinforcement learning in which its utilization aims to direct the selection of actions, thus preventing the necessity for a random exploration in the iterative process of the metaheuristics. In this study, we provide Q-learning guided algorithms for the Bi-Criteria No-Wait Flowshop Scheduling Problem (NWFSP). The problem is treated as a bi-criteria combinatorial optimization problem where total flow time and makespan are optimized simultaneously. Firstly, a deterministic mixed-integer linear programming (MILP) model is provided. Then, Q-learning guided algorithms are developed: Bi-Criteria Iterated Greedy Algorithm with Q-Learning (BC-IGQL). Bi-Criteria Block Insertion Heuristic Algorithm with Q-Learning (BC-BIHQL). Moreover, the performance of the proposed Q-learning guided algorithms is compared over a collection of Bi-Criteria Genetic Local Search Algorithms (BC-GLS), Bi-Criteria Iterated Greedy Algorithm (BC-IG), Bi-Criteria Iterated Greedy Algorithm with a Local Search (BC-IGALL) and Bi-Criteria Variable Block Insertion Heuristic Algorithm (BC-VBIH). The complete computational experiment, performed on 480 problem instances of Vallada et al. (2015), which is known as the VRF benchmark set, indicates that the BC-BIHQL and the BC-IGQL algorithms outperform the BC-GLS, BC-IG, BC-IGALL, and BC-VBIH algorithms in comparative performance metrics. More specifically, the proposed BC-BIHQL and BC-IGQL algorithms can yield more non-dominated bi-criteria solutions with the most substantial competitiveness than the remaining algorithms. At the same time, both are competitive with each other on the benchmark problems. Moreover, the BC-IGQL algorithm dominates almost 97% and 99% of the solutions reached by the BC-IG, BC-IGALL, and BC-VBIH algorithms in small and large datasets. Similarly, The BC-BIHQL algorithm dominates almost 98% and 99% of the solutions reached by the BC-IG, BC-IGALL, and BC-VBIH algorithms in small and large datasets, respectively. This means that, among all the features that have been compared, the Q-learning-guided algorithms demonstrate the highest level of competitiveness. The outcomes of this study encourage us to discover many more bi-criteria NWFSPs to reveal the trade-off between other conflicting objectives, such as makespan & the number of early jobs, to overcome various industries' problems.

Statistical pathways to low-carbon cities: Analyzing renewable integration, energy-efficient design, and job creation

To address the existing gap in understanding the intricate dynamics of resource management in Cloud Fog Computing (CFC)-driven Smart Grids (SG), this study introduces an innovative adaptive framework integrating Genetic Optimization Algorithm (GOA) and Cuckoo Search Optimization Algorithm (CSOA), aiming to optimize response times and enhance overall efficiency. In the domain of contemporary energy grids, the implementation of intelligent grids (IG) has ushered in a new era of dependable, streamlined, eco-friendly, and economically viable electric services. These grids constitute a fundamental facet of present-day energy resources. Moreover, the advent of cloud and mist computing (CMC) provides on-demand access to computational resources. Overcoming associated challenges, CMC not only delivers advantages like cost-effectiveness, energy conservation, scalability, flexibility, and adaptability but also presents an innovative model for resource management in IG. This study introduces a CMC-driven strategy for managing resources in IG by digitally replicating key grid components. The proposed framework aims to furnish diverse computational services for resource management within IG through a hierarchical arrangement of CMCs. Additionally, this investigation scrutinizes and simulates two optimization approaches: the cuckoo search and grasshopper optimization algorithms. These algorithms operate on an adaptive machine learning mechanism designed to balance the load, and the study meticulously compares their efficacy. These algorithms operate on an adaptive machine learning mechanism designed to balance the load, and the study meticulously compares their efficacy in the context of Renewable Integration, Energy-Efficient Design, and Job Creation.

Transferable multi-objective factory layout planning using simulation-based deep reinforcement learning

Factory layout planning aims at finding an optimized layout configuration under consideration of varying influences such as the material flow characteristics. Manual layout planning can be characterized as a complex decision-making process due to a large number of possible placement options. Automated planning approaches aim at reducing the manual planning effort by generating optimized layout variants in the early stages of layout planning. Recent developments have introduced deep Reinforcement Learning (RL) based planning approaches that allow to optimize a layout under consideration of a single optimization criterion. However, within layout planning, multiple partially conflicting planning objectives have to be considered. Such multiple objectives are not considered by existing RL-based approaches. This paper addresses this research gap by presenting a novel deep RL-based layout planning approach that allows consideration of multiple objectives for optimization. Furthermore, existing RL-based planning approaches only consider analytically formulated objectives such as the transportation distance. Consequently, dynamic influences in the material flow are neglected which can result in higher operational costs of the future factory. To address this issue, a discrete event simulation module is developed that allows simulating manufacturing and material flow processes simultaneously for any layout configuration generated by the RL approach. Consequently, the presented approach considers material flow simulation results for multi-objective optimization. To investigate the capabilities of RL-based factory layout planning, different RL architectures are compared based on a simplified application scenario. Throughput time, media supply, and material flow clarity are considered as optimization objectives. The best performing architecture is then applied to an exemplary application scenario and compared with the results obtained by a combined version of the genetic algorithm and tabu search, the non-dominated sorting genetic algorithm, and the optimal solution. Finally, two industrial planning scenarios, one focusing on brownfield and one on greenfield planning, are considered. The results show that the performance of RL compared to meta-heuristics depends on the considered computation time. With time the results generated by the RL approach exceed the quality of the best conventional solution by up to 11%. Finally, the potential of applying transfer learning is investigated for three different application scenarios. It is observed that RL can learn generalized patterns for factory layout planning, which allows to significantly reduce the required training time and can lead to improved solution quality. Thus, the use of pre-trained RL models shows a substantial performance potential for automated factory layout planning which cannot be achieved with conventional automated planning approaches.

A hybrid genetic search and dynamic programming-based split algorithm for the multi-trip time-dependent vehicle routing problem

We design a hybrid algorithm for the multi-trip time-dependent vehicle routing problem (MT-TD-VRP). One of its components is the Time-Dependent SPlit Algorithm (TD-SPA), which is a dynamic programming-based algorithm specifically designed to handle both the multi-trip per vehicle and the time-dependent aspects of the problem. The hybrid algorithm combines the proposed TD-SPA, designed to efficiently split a giant tour into complete vehicle routes, with a genetic algorithm for generating these tours. We introduce a monotone queue optimization (MQO) technique to accelerate the TD-SPA. The effectiveness of MQO is evaluated by comparing computation times between the original split algorithm for the capacitated vehicle routing problem (CVRP) and its MQO-enhanced counterpart. Extensive numerical experiments with a real-world dataset from a Singapore food and beverage company are conducted to assess our algorithm’s performance on various MT-TD-VRP instances. The results indicate that our algorithm surpasses the performance of the commercial solver Gurobi, with an average improvement of 25.13% on the best solutions found within a prescribed duration. Our numerical simulations further reveal the algorithm’s ability to efficiently solve both the capacitated vehicle routing problem (CVRP) and the multi-trip vehicle routing problem (MTVRP), consistently producing competitive solutions. Moreover, to highlight the importance of incorporating the time-dependent (TD) factor into our model and algorithm, we demonstrate a notable enhancement in performance–averaging at 7.47% for the best solutions under TD conditions for an MTVRP dataset.

A Triple Bottom Line optimization model for assignment and routing of on-demand home services

‘On-demand home services’ is a fast-growing industry where online platforms match independent service professionals with customers seeking aid for household tasks. In this paper, we study the assignment and routing of service professionals for serving customers of an on-demand home services platform considering the Triple Bottom Line (TBL) criteria for ensuring sustainability in operations. We characterize this as the Home Services Assignment and Routing Problem with the Triple Bottom Line (HSARP-TBL) and implement a Mixed Integer Linear Programming (MILP) model for solving it. We assign service professionals to customers based on their desired time slots and also transport modes for each customer visit by a professional, considering either combinations of public transport or a personal vehicle for each professional’s tour. The objective is to minimize costs due to time window violations and uncovered customers, catering to the economic pillar of the TBL. We incorporate additional constraints related to the TBL by improving customer satisfaction based on the ratings of assigned professionals to customers, with and without subscription (economic), controlling emissions due to transportation of professionals (environmental) and ensuring equity in service allocation and net earnings between professionals (social). For tackling large instances we implement a Hybrid Genetic Search (HGS) algorithm adapting it to our problem setting. We demonstrate that the HGS outperforms the MILP model systematically for large instances in terms of solution value and computational time. Finally, we observe that for some instances, without worsening the primary economic objective, all the TBL indicators can be improved.

Bayesian Algorithm for The Construction of Logistics Node Delay Model and Its Impact on Subsequent Nodes in Supply Chain

The logistics nodes delay of supply chain has become a focus of customer complaints at the State Post Bureau of China in recent years. Nodes delay usually also brings uncertain business loss. The purpose of this paper is to explore propagation of nodes delay and make it precisely controllable. To address this, Netica software is used and a hybrid strategy of genetic algorithm (GA) and tabu search algorithm (TS) validated by datasets is utilized to optimize the Bayesian algorithm model. Bayesian algorithm model of logistics nodes delay of supply chain is constructed through parameter learning. The empirical analysis is conducted on the basis of three types of nodes delay with 2041 sets of data for cargo departures and arrivals, upstream nodes delay, and nodes delay of supply chain. The results show that the nodes delay of supply chain gradually decreases from upstream to downstream; there are strong correlations between propagation probability and transfer time of nodes delay; the effects of initial node delay are the maximum, after 3 transitions, still having 30% probability to affect the succeeding nodes. Therefore, some suggestions, such as controlling the transfer points and transfer time and strengthening the management of initial node delay, are given.

A fuzzy interval optimization approach for p-hub median problem under uncertain information.

Stochastic and robust optimization approaches often result in sub-optimal solutions for the uncertain p-hub median problem when continuous design parameters are discretized to form different environmental scenarios. To solve this problem, this paper proposes a triangular fuzzy number model for the Non-Strict Uncapacitated Multi-Allocation p-hub Median Problem. To enhance the quality and the speed of optimization, a novel optimization approach, combining the triangular fuzzy number evaluation index with the Genetic-Tabu Search algorithm, is proposed. During the iterations of the Genetic-Tabu Search algorithm for finding the optimal solution, the fitness of fuzzy hub schemes is calculated by considering the relative positional relationships of triangular fuzzy number membership functions. This approach directly addresses the triangular fuzzy number model and ensures the integrity of information in the p-hub problem as much as possible. It is verified by the classic Civil Aeronautics Board and several self-constructed data sets. The results indicate that, compared to the traditional Genetic Algorithm and Tabu Search algorithm, the Genetic-Tabu Search algorithm reduces average computation time by 49.05% and 40.93%, respectively. Compared to traditional random, robust, and real-number-based optimization approaches, the proposed optimization approach reduces the total cost in uncertain environments by 1.47%, 2.80%, and 8.85%, respectively.

Crystal structure prediction using neural network potential and age-fitness Pareto genetic algorithm

While crystal structure prediction (CSP) remains a longstanding challenge, we introduce ParetoCSP, a novel algorithm for CSP, which combines a multi-objective genetic algorithm (GA) with a neural network inter-atomic potential model to find energetically optimal crystal structures given chemical compositions. We enhance the updated multi-objective GA (NSGA-III) by incorporating the genotypic age as an independent optimization criterion and employ the M3GNet universal inter-atomic potential to guide the GA search. Compared to GN-OA, a state-of-the-art neural potential-based CSP algorithm, ParetoCSP demonstrated significantly better predictive capabilities, outperforming by a factor of $$ 2.562 $$ across $$ 55 $$ diverse benchmark structures, as evaluated by seven performance metrics. Trajectory analysis of the traversed structures of all algorithms shows that ParetoCSP generated more valid structures than other algorithms, which helped guide the GA to search more effectively for the optimal structures. Our implementation code is available at https://github.com/sadmanomee/ParetoCSP .

Sensor placement algorithm for faults detection in electrical secondary distribution network using dynamic programming method: focusing on dynamic change and expansion of the network configurations

Modern power grids are developing toward smartness through the use of sensors in gathering data for situation awareness, visibility, and fault detection. In most developing countries, fault detection in the electrical secondary distribution network (SDN) is very challenging due to the lack of automated systems for network monitoring. Systems for monitoring faults require sensor placement on each node, which is not economically feasible. Hence, optimal placement algorithms are required to ensure that the network is observable with few sensors possible. The existing sensor placement methods based on mathematical and heuristic approaches are efficient for transmission and primary distribution networks which are mostly static in size and layout. Such methods may not be efficient in SDN which is dynamic in size and have a relatively large number of nodes. This study proposes an enhanced dynamic programming method for sensor placement to enhance fault detection in SDN. The proposed algorithm employs the depth search concepts and the parent–children relationship between nodes to determine sensor types and locations considering the optimal cost. The proposed algorithm was compared with other methods including particle swarm optimization, genetic algorithm, and chaotic crow search algorithm using different network configurations. The results revealed that the proposed algorithm suggested the minimum number of sensors and shortest convergence time of 1.27 min. The results also revealed that, on network expansion, maintaining the location of the existing sensors is more cost-effective by 20% than reallocating the existing sensors. Furthermore, the results revealed that an average of 30% of nodes, need sensors to observe the entire network, hence cost optimization.

Modified genetic algorithm to solve worker assignment problem with time windows

In recent years, the demand for electronic products has been increasing rapidly. T mounting technology (SMT) line is one of the production areas for electronic products, directly affecting this situation. In an SMT line, multiple machines mount electronic parts to the board. The worker must complete work when the parts used in these machines are within the remaining parts available for replacement. When a worker fails to replace parts at the right time, the production line stops, and delays occur. Besides, there may be a designated worker who should be assigned to each task. In the current situation, workers’ work procedures are not optimized, so they should schedule work procedures for each worker. This problem is called Worker Assignment Problem with Time Window (WAPTW). This paper proposes a method to solve WAPTW called Genetic Algorithm with Local Restriction (GALR). GALR combines a genetic algorithm (GA) and local search with local restriction. This paper’s main contribution is introducing WAPTW as a novel real-world optimization problem in an electricity company, its mathematical formulation, and a proposed GALR to solve WAPTW. The experiment shows that the proposed method could yield the best result in real-world WAPTW compared with other methods.

Dynamic repositioning in bike-sharing systems with uncertain demand: An improved rolling horizon framework

The bike-sharing system (BSS) serves as a vital shared mobility mechanism, providing first- and last-mile services in urban transportation. However, one ongoing challenge in the BSS is the spatiotemporal imbalance caused by fluctuating user demand. To address this issue, this paper studies a dynamic bike repositioning problem under scenario-based demand to achieve a balance in inventory levels for electronic fences. Specifically, a multi-period two-stage stochastic model is proposed to make the trade-off between operational costs and service quality. The model incorporates a vehicle no-return (VNR) strategy, which allows vehicles to remain at the last station at the end of each period, awaiting the next repositioning task. To solve the model, an improved rolling horizon framework is introduced to effectively handle the temporal dimension complexity of the multi-period problem. Then a customized hybrid heuristic algorithm based on parallel genetic algorithm and variable neighborhood search is developed to obtain high-quality solutions. To validate the effectiveness and applicability of the proposed method, small-scale and practical-scale numerical experiments are specifically conducted based on real-world BSS data from Beijing, China. The results demonstrate the proposed method (i) creates a substantial reduction in the number of unmet demands when multiple scenarios potentially occur compared to the determined model, (ii) yields an average reduction of 29.04% in operational costs compared to the scheme without the VNR strategy, (iii) overcomes the short-sightedness inherent in the conventional single-period rolling horizon approach, exhibiting closer to optimal performance, and (iv) is superior to baseline approaches in terms of solution quality and efficiency.

Hybridizing genetic algorithm and harmony search for higher-order mutation testing

Mutation testing can evaluate the quality of the test inputs, generate test data, and simulate any test coverage criterion. Genetic algorithms and harmony search have been applied to reduce the cost of generating test inputs. Although hybridizing search algorithms enhances the efficiency of searching the solution domain, there is a shortage of applying the hybrid search techniques in mutation testing. This paper merges the genetic and harmony search algorithms to effectively generate test data to kill higher-order mutants. In addition, the performance of the proposed technique will be evaluated and compared with a stand-alone genetic algorithm and a stand-alone harmony search algorithm through an empirical study using a set of benchmark programs. The experimental study shows that the proposed technique outperformed the compared algorithms, reaching a higher killing ratio, where the proposed approach kills 92.8% of higher-order mutants for all tested programs. In comparison, GA kills 88.7%, and HA kills 86.6%. Besides, the proposed algorithm overcame the compared algorithm in reaching a targeted killing ratio faster than the compared algorithms. HGA reduced the execution time for each program with a reduction ratio ranging from 58.9% to 89.8%.

Identification of fruit using a flexible tactile sensor array

This article aims to address the issue of low recognition accuracy in existing sorting robots caused by lighting, occlusion, and environmental factors. A fruit recognition method based on a flexible tactile sensor array is described. This method enables the robot to directly perceive the attributes of the objects and identify fruits using a flexible gripper, facilitating intelligent sorting. A novel flexible tactile sensor array is utilized to construct a flexible hand tactile information acquisition platform, which collects tactile time series data for the fruits. Principal component analysis is then employed for dimensionality reduction, followed by the development of an improved particle swarm optimization for the support vector machine model. Through an experimental study, the optimized model is compared with four other models, demonstrating better classification performance. The optimized model achieves an average tactile classification accuracy of up to 98.10% for the five types of fruits. A comparison between the improved optimization algorithm, genetic algorithm, and grid search algorithm reveals the superior optimization performance of the new approach. In the future, this method is expected to be implemented in industrial sorting robots for intelligent sorting on automatic production lines. Furthermore, the algorithm will be further refined to enhance the classification accuracy and efficiency.

Optimizing routes of municipal waste collection: an application algorithm

PurposeThe purpose of this study is to formulate an algorithm designed to discern the optimal routes for efficient municipal solid waste (MSW) collection.Design/methodology/approachThe research method is simulation. The proposed algorithm combines heuristics derived from the constructive genetic algorithm (CGA) and tabu search (TS). The algorithm is applied in a municipality located at Southern Brazil, with 40,000 inhabitants, circa.FindingsThe implementation achieved a remarkable 25.44% reduction in daily mileage of the vehicles, resulting in savings of 150.80 km/month and 1,809.60 km/year. Additionally, it reduced greenhouse gas emissions (including fossil CO2, CH4, N2O, total CO2e and biogenic CO2) by an average of 26.15%. Moreover, it saved 39 min of daily working time.Research limitations/implicationsFurther research should thoroughly analyze the feasibility of decision-making regarding planning, scheduling and scaling municipal services using digital technology.Practical implicationsThe municipality now has a tool to improve public management, mainly related with municipal solid waste. The municipality reduced the cost of public management of municipal solid waste, redirecting funds to other priorities, such as public health and education.Originality/valueThe study integrates MSW collection service with an online platform based on Google MapsTM. The advantages of employing geographical information systems are agility, low cost, adaptation to changes and accuracy.

Research on the Heterogeneous Autonomous Underwater Vehicle Cluster Scheduling Problem Based on Underwater Docking Chambers

The onboard energy supply of Autonomous Underwater Vehicles (AUVs) is one of the main limiting factors for their development. The existing methods of deploying and retrieving AUVs from mother ships consume a significant amount of energy during submerging and surfacing, resulting in a small percentage of actual working time. Underwater docking chambers provide support to AUVs underwater, saving their precious energy and addressing this issue. When an AUV cluster is assigned multiple tasks, scheduling the cluster becomes essential, and task allocation and path planning are among the core problems in AUV cluster scheduling research. In this paper, based on the underwater docking chamber, an Improved Genetic Local Search Algorithm with Prior Knowledge (IGLSAPK) is proposed to simultaneously solve the task allocation and path planning problems. Under constraints such as onboard energy supply, AUV quantity, and AUV type, the algorithm groups AUVs, assigns tasks, and plans paths to accomplish tasks at different locations, aiming to achieve overall efficiency. The algorithm first generates an initial population using prior knowledge to improve its search efficiency. It then combines an improved local search algorithm to efficiently solve large-scale, complex, and highly coupled problems. The algorithm has been evaluated through simulation experiments and comparative experiments, and the results demonstrate that the proposed algorithm outperforms other algorithms in terms of speed and optimality. The algorithm presented in this paper addresses the grouping, task allocation, and path planning problems in heterogeneous AUV clusters. Its practical significance lies in its ability to handle tasks executed by a heterogeneous AUV group, making it more practical compared to previous algorithms.

Fuzzy inference system using genetic algorithm and pattern search for predicting roof fall rate in underground coal mines

One of the most dangerous safety hazard in underground coal mines is roof falls during retreat mining. Roof falls may cause life-threatening and non-fatal injuries to miners and impede mining and transportation operations. As a result, a reliable roof fall prediction model is essential to tackle such challenges. Different parameters that substantially impact roof falls are ill-defined and intangible, making this an uncertain and challenging research issue. The National Institute for Occupational Safety and Health assembled a national database of roof performance from 37 coal mines to explore the factors contributing to roof falls. Data acquired for 37 mines is limited due to several restrictions, which increased the likelihood of incompleteness. Fuzzy logic is a technique for coping with ambiguity, incompleteness, and uncertainty. Therefore, In this paper, the fuzzy inference method is presented, which employs a genetic algorithm to create fuzzy rules based on 109 records of roof fall data and pattern search to refine the membership functions of parameters. The performance of the deployed model is evaluated using statistical measures such as the Root-Mean-Square Error , Mean-Absolute-Error, and coefficient of determination (R_2). Based on these criteria, the suggested model outperforms the existing models to precisely predict roof fall rates using fewer fuzzy rules.

Bi-Level Optimization for De-Icing Position Allocation and Unmanned De-Icing Vehicle Fleet Routing Problem.

Aircraft icing due to severe cold and local factors increases the risk of flight delays and safety issues. Therefore, this study focuses on optimizing de-icing allocation and adapting to dynamic flight schedules at medium to large airports. Moreover, it aims to establish a centralized de-icing methodology employing unmanned de-icing vehicles to achieve the dual objectives of minimizing flight delay times and enhancing airport de-icing efficiency. To achieve these goals, a mixed-integer bi-level programming model is formulated, where the upper-level planning guides the allocation of de-icing positions and the lower-level planning addresses the collaborative scheduling of the multiple unmanned de-icing vehicles. In addition, a two-stage algorithm is introduced, encompassing a Mixed Variable Neighborhood Search Genetic Algorithm (MVNS-GA) as well as a Multi-Strategy Enhanced Heuristic Greedy Algorithm (MSEH-GA). Both algorithms are rigorously assessed through horizontal comparisons. This demonstrates the effectiveness and competitiveness of these algorithms. Finally, a model simulation is conducted at a major northwestern hub airport in China, providing empirical evidence of the proposed approach's efficiency. The results show that research offers a practical solution for optimizing the use of multiple unmanned de-icing vehicles in aircraft de-icing tasks at medium to large airports. Therefore, delays are mitigated, and de-icing operations are improved.