Crossover Operator Research Articles

Application of wireless sensor network based on improved genetic algorithm in English blended learning

Adaptive genetic algorithm is one of the classic algorithms to solve multi-objective optimization problems. In order to make the convergence set to the Pareto boundary more uniform, this article improves how to compare multiple distances in the algorithm, introduces the concept of comparing the difference between single distances, and uses the adaptive and mutation of crossover operators and the minimum distance filtering mechanism. An improved NSGA-II adaptive algorithm is proposed. Among other things, in the machine learning system model, the environment refers to an external source of information, which provides research-related information. This study proved that learning requires systematic organization, similar to short-term memory in information processing models. Find the external environment, then use the methods of thinking analysis, integration, comparison and induction to obtain knowledge, and store the knowledge in a scientific way. With the rapid development of knowledge in modern society, on the basis of the development of English text analysis technology and network technology, the reform of China's education system has reached a new level. Reforming modern teaching methods is an important part of education reform. More and more educators are considering analyzing English texts as one of the important tools to reform teaching methods and improve teaching effects. Among other things, based on the B/S model development technology, the teaching system is researched using production theories and methods, and the teaching system is actually being tested, developed and applied. By comparing the effects of computer network teaching methods and traditional teaching methods, the feasibility and effectiveness of the network teaching system are determined.

Crossover Boosted Grey Wolf Optimizer‐based framework for leader election and resource allocation in Intrusion Detection Systems for MANETs

AbstractMobile Ad hoc Networks (MANETs) is a self‐organizing networks without having a fixed infrastructure for making them susceptible to security threats. Intrusion Detection Systems (IDS) promotes security in MANETs by identifying malicious activities. Leader election is a fundamental aspect of IDS deployment, impacting resource allocation and system efficiency. This article presents a novel approach, the Crossover Boosted Grey Wolf Optimizer (CBGWO), for leader election and resource allocation in MANET‐based IDS. The proposed CBGWO algorithm integrates the Grey Wolf Optimizer (GWO) with innovative crossover operators that have an ability to enhance the capabilities of exploration and exploitation in the optimization process. The leader election problem is solved through applying multi‐objective optimization by considering energy consumption, reputation, and communication overhead. Objective functions are defined to maximize energy efficiency while maintaining a balanced distribution of leadership roles. Extensive simulations are conducted, varying network densities and the percentage of selfish nodes. Results demonstrate the effectiveness of the CBGWO‐based model in balancing energy consumption, prolonging network lifespan, and enhancing intrusion detection by comparing different state‐of‐the‐art models such as PCA‐FELM, CTAA‐MPSO, FLS‐RE, LEACH, DCAIDS, WOA‐GA, and VOELA. The proposed model achieved an energy consumption of 4.31 J, network lifetime of 560.482 ms, and average intrusion detection latency of 0.12 s, respectively. The proposed model outperforms than existing random and connectivity‐based leader election methods that is evaluated by taking main consideration of energy efficiency and network survivability. This research contributes to the field by introducing a robust algorithm for leader election in MANET‐based IDS, addressing challenges posed by network dynamics and resource constraints. The CBGWO‐based approach showcases its potential to achieve effective leader election and efficient resource allocation, thereby enhancing the security and sustainability of MANETs.

Solving the multi-objective path planning problem for mobile robot using an improved NSGA-II algorithm

Path planning is one of the most concerned problems in mobile robotics. Given the NP-hard nature of path planning problems, the non-dominated sorting genetic algorithm (NSGA-II), as one of the outstanding evolutionary algorithms with robust optimization capabilities, is a good candidate to deal with them. In this study, an improved NSGA-II (INSGA-II) is proposed to solve the multi-objective path planning problems, in which path length, path safety and path smoothness are optimized simultaneously. First, a grid-based environment model is created, and each feasible solution is encoded as a discrete ordered point set. Second, a hybrid initialization strategy is employed to enrich the diversity of initial population. Third, besides the usual selection, crossover and mutation operators, problem-specific evolutionary operators are developed. In addition, an adaptive variable neighborhood local search strategy is designed to increase the exploitation ability of the algorithm. Moreover, a hybrid global search strategy is applied to further improve the algorithm’s the exploration ability. Finally, the proposed INSGA-II is compared with six state-of-the-art algorithms on 16 instances of four representative maps. Simulation results indicate that INSGA-II can solve the multi-objective path planning problems with high efficiency.

Economic optimization of fresh logistics pick-up routing problems with time windows based on gray prediction

The rapid development of modern cold chain logistics technology has greatly expanded the sales market of agricultural products in rural areas. However, due to the uncertainty of agricultural product harvesting, relying on the experience values provided by farmers for vehicle scheduling can easily lead to low utilization of vehicle capacity during the pickup process and generate more transportation cost. Therefore, this article adopts a non-linear improved grey prediction method based on data transformation to estimate the pickup demand of fresh agricultural products, and then establishes a mathematical model that considers the fixed vehicle usage cost, the damage cost caused by non-linear fresh fruit and vegetable transportation damage and decay rate, the cooling cost generated by refrigerated transportation, and the time window penalty cost. In order to solve the model, a hybrid simulated annealing algorithm integrating genetic operators was designed to solve this problem. This hybrid algorithm combines local search strategies such as the selection operator without repeated strings and the crossover operator that preserves the best substring to improve the algorithm’s solving performance. Numerical experiments were conducted through a set of benchmark examples, and the results showed that the proposed algorithm can adapt to problem instances of different scales. In 50 customer examples, the difference between the algorithm and the standard value in this paper is 2.30%, which is 7.29% higher than C&S. Finally, the effectiveness of the grey prediction freight path optimization model was verified through a practical case simulation analysis, achieving a logistics cost savings of 9.73%.

Analysing Equilibrium States for Population Diversity

Population diversity is crucial in evolutionary algorithms as it helps with global exploration and facilitates the use of crossover. Despite many runtime analyses showing advantages of population diversity, we have no clear picture of how diversity evolves over time. We study how the population diversity of (μ+1)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$(\\mu +1)$$\\end{document} algorithms, measured by the sum of pairwise Hamming distances, evolves in a fitness-neutral environment. We give an exact formula for the drift of population diversity and show that it is driven towards an equilibrium state. Moreover, we bound the expected time for getting close to the equilibrium state. We find that these dynamics, including the location of the equilibrium, are unaffected by surprisingly many algorithmic choices. All unbiased mutation operators with the same expected number of bit flips have the same effect on the expected diversity. Many crossover operators have no effect at all, including all binary unbiased, respectful operators. We review crossover operators from the literature and identify crossovers that are neutral towards the evolution of diversity and crossovers that are not.

Omnidirectional AGV Path Planning Based on Improved Genetic Algorithm

To address the issues with traditional genetic algorithm (GA) path planning, which often results in redundant path nodes and local optima, we propose an Improved Genetic Algorithm that incorporates an ant colony algorithm (ACO). Firstly, a new population initialization method is proposed. This method adopts a higher-quality random point generation strategy to generate random points centrally near the start and end of connecting lines. It combines the improved ACO algorithm to connect these random points quickly, thus greatly improving the quality of the initial population. Secondly, path smoothness constraints are proposed in the adaptive function. These constraints reduce the large-angle turns and non-essential turns, improving the smoothness of the generated path. The algorithm integrates the roulette and tournament methods in the selection stage to enhance the searching ability and prevent premature convergence. Additionally, the crossover stage introduces the edit distance and a two-layer crossover operation based on it to avoid ineffective crossover and improve convergence speed. In the mutation stage, we propose a new mutation method and introduce a three-stage mutation operation based on the idea of simulated annealing. This makes the mutation operation more effective and efficient. The three-stage mutation operation ensures that the mutated paths also have high weights, increases the diversity of the population, and avoids local optimality. Additionally, we added a deletion operation to eliminate redundant nodes in the paths and shorten them. The simulation software and experimental platform of ROS (Robot Operating System) demonstrate that the improved algorithm has better path search quality and faster convergence speed. This effectively prevents the algorithm from maturing prematurely and proves its effectiveness in solving the path planning problem of AGV (automated guided vehicle).

Multi-objective coyote optimization algorithm based on hybrid elite framework and Meta-Lamarckian learning strategy for optimal power flow problem

A multi-objective coyote optimization algorithm based on hybrid elite framework and Meta-Lamarckian learning strategy (MOCOA-ML) was proposed to solve the optimal power flow (OPF) problem. MOCOA-ML adds external archives with grid mechanism on the basis of elite non-dominated sorting. It can guarantee the diversity of the population while obtaining the Pareto solution set. When selecting elite coyotes, there is a greater probability to select the elite in sparse areas, which is conducive to the development of sparse areas. In addition, combined with Meta-Lamarckian learning strategy, based on four crossover operators (horizontal crossover operator, longitudinal crossover operator, elite crossover operator and direct crossover operator), the local search method is adaptively selected for optimization, and its convergence performance is improved. First, the simulation is carried out in 20 test functions, and compared with MODA, MOPSO, MOJAYA, NSGA-II, MOEA/D, MOAOS and MOTEO. The experimental results showed that MOCOA-ML achieved the best inverted generational distance value and the best hypervolume value in 11 and 13 test functions, respectively. Then, MOCOA-ML is used to solve the optimal power flow problem. Taking the fuel cost, power loss and total emissions as objective functions, the tests of two-objective and three-objective bechmark problems are carried out on IEEE 30-bus system and IEEE 57-bus system. The results are compared with MOPSO, MOGWO and MSSA algorithms. The experimental results of OPF demonstrate that MOCOA-ML can find competitive solutions and ranks first in six cases. It also shows that the proposed method has obtained a satisfactory uniform Pareto front.

Dynamic Adaptive Resource Allocation for Edge Computing in Big Data Analytics Using GBDT, DQN, and GA Algorithms

Edge computing in big data refers to processing and analysing data closer to its source, reducing latency and bandwidth usage. It leverages devices at the network edge to perform computations, making real-time analytics feasible. This distributed approach improves efficiency and enables faster decision-making, critical for applications like IoT, autonomous vehicles, and healthcare. The research proposes an innovative approach that harnesses three machine learning algorithms Gradient Boosting Decision Trees (GBDT), Deep Q-Network (DQN), and Genetic Algorithm (GA) to enable dynamic adaptive resource allocation within edge computing environments tailored for big data analytics. GBDT enhances classification accuracy by sequentially refining predictions through decision trees, accommodating heterogeneous data types and yielding high prediction accuracy crucial for dynamic edge environments. The GA evaluates resource allocation strategies represented as chromosomes within a population, selecting promising solutions as parents for the next generation and generating diverse offspring through crossover and mutation operations to discover optimal solutions. DQN facilitates intelligent resource allocation by iteratively refining Q-values based on experiences gathered during interactions with the environment, utilizing a neural network to determine optimal actions for a given state, thereby enhancing performance and efficiency in edge computing environments. This integrated approach ensures flexible resource allocation and fortified capabilities for big data analytics within edge computing environments. The research underscores GBDT as the most promising algorithm for resource allocation in edge computing environments, owing to its exceptional performance in resource utilization, scalability, and accuracy. This nuanced understanding of algorithmic behaviour in dynamic settings offers invaluable insights for optimizing resource allocation strategies, thereby enhancing the efficiency and effectiveness of edge computing systems in handling big data analytics tasks.

Genetic Algorithm Based Path Planning for Seawater Depth Data Measurement in Real Scenarios

In this paper, firstly, the movement direction of the measuring ship is clarified, and according to the requirements of factors such as the range of overlap rate is controlled at 10% to 20% and the sea area is completely covered, the optimization equation system is derived, and the objective function is derived through the requirements, and the optimal solution can be derived from the two associations as the 34 measuring lines and the total length of 68 nautical miles of the shortest measuring lines. Using genetic algorithm to examine the real scene of seawater depth data measurement path planning, the length of the measurement line and coverage as an assessment index of the algorithm's subsequent optimization of the object of choice. Through the simulation of genetic mutation, crossover and selection operations, the total length of the shortest survey line is 21.5261 nautical miles, the percentage of missed sea area is 1.6%, and the overlap rate is 13.12% after the iterative solution of the genetic algorithm.

Dynamic Bayesian network structure learning based on an improved bacterial foraging optimization algorithm

With the rapid development of artificial intelligence and data science, Dynamic Bayesian Network (DBN), as an effective probabilistic graphical model, has been widely used in many engineering fields. And swarm intelligence algorithm is an optimization algorithm based on natural selection with the characteristics of distributed, self-organization and robustness. By applying the high-performance swarm intelligence algorithm to DBN structure learning, we can fully utilize the algorithm's global search capability to effectively process time-based data, improve the efficiency of network generation and the accuracy of network structure. This study proposes an improved bacterial foraging optimization algorithm (IBFO-A) to solve the problems of random step size, limited group communication, and the inability to maintain a balance between global and local searching. The IBFO-A algorithm framework comprises four layers. First, population initialization is achieved using a logistics-sine chaotic mapping strategy as the basis for global optimization. Second, the activity strategy of a colony foraging trend is constructed by combining the exploration phase of the Osprey optimization algorithm. Subsequently, the strategy of bacterial colony propagation is improved using a "genetic" approach and the Multi-point crossover operator. Finally, the elimination-dispersal activity strategy is employed to escape the local optimal solution. To solve the problem of complex DBN learning structures due to the introduction of time information, a DBN structure learning method called IBFO-D, which is based on the IBFO-A algorithm framework, is proposed. IBFO-D determines the edge direction of the structure by combining the dynamic K2 scoring function, the designed V-structure orientation rule, and the trend activity strategy. Then, according to the improved reproductive activity strategy, the concept of "survival of the fittest" is applied to the network candidate solution while maintaining species diversity. Finally, the global optimal network structure with the highest score is obtained based on the elimination-dispersal activity strategy. Multiple tests and comparison experiments were conducted on 10 sets of benchmark test functions, two non-temporal and temporal data types, and six data samples of two benchmark 2T-BN networks to evaluate and analyze the optimization performance and structure learning ability of the proposed algorithm under various data types. The experimental results demonstrated that IBFO-A exhibits good convergence, stability, and accuracy, whereas IBFO-D is an effective approach for learning DBN structures from data and has practical value for engineering applications.

Sizing and Shape Optimization of Discrete Truss Employing a Target-Oriented Krill Herd Algorithm

Abstract The krill herd (KH) algorithm is widely used for optimizing truss structures as no gradient information is necessary, and only a few parameters require adjustment. However, when the truss structure becomes discrete and complex, KH tends to fall into a local optimum. Therefore, a novel target-oriented KH (TOKH) algorithm is proposed in this study to optimize the design of discrete truss structures. Initially, a crossover operator is established between the “best krill” and “suboptimal krill” to generate a robust “cross krill” for global exploration. Additionally, an improved local mutation and crossover (ILMC) operator is introduced to fine-tune the “center of food” and candidate solutions for local exploitation. The proposed method and other optimization approaches are experimentally compared considering 15 benchmark functions. Then, the performance of the TOKH algorithm is evaluated based on four discrete truss structure optimization problems under multiple loading conditions. The obtained optimization results indicate that the proposed method presents competitive solutions in terms of accuracy, unlike other algorithms in the literature, and avoids falling into a local minimum.

GENETIC ALGORITHM-BASED OPTIMIZATION OF THE LASER-BEAM PATH IN ADDITIVE MANUFACTURING

This study presents a methodology of genetic-algorithm-based optimization of the laser-beam path for improving laser-based additive manufacturing (AM). A simple thermal model was developed to simulate the effects of laser-induced heat input on the temperature distribution within the substrate during the fabrication of one layer. The optimization approach aims to find solutions with more homogeneous temperature properties that minimize the thermal gradient on the substrate caused by laser-based AM. The laser beam, i.e., the tool-path planning, is formulated as the search for the optimal sequence of cell depositions that minimize the fitness function, which is composed of two components, i.e., the thermal fitness and process fitness. The thermal fitness is expressed as the average thermal gradient, and the process fitness regulates the suitability of the proposed tool path for the implementation of the AM process. Various tool-path generators are proposed to initialize the initial population of tool-path solutions. Genetic-algorithm-based tool-path optimization is proposed, where custom initialization, crossover and mutation operators are developed for application in laser-based AM. Simulation studies demonstrate the effectiveness of the genetic-algorithm-based optimization in finding solutions that minimize the fitness function and therefore provide both thermally and, for the AM process implementation, more suitable laser-beam-path solutions.

GIS spatial optimization for agricultural crop allocation using NSGA-II

This study focuses on the shift from traditional farming methods, reliant on farmer intuition and manual processes, to modern, automated approaches crucial for Thailand’s agricultural sustainability. Despite its vital role in the country’s economy, outdated practices lead to supply imbalances and perpetuate poverty among smallholder farmers. Using geographic information systems (GIS) and mathematical optimization, the present study aims to determine optimal agricultural crop allocation. A multi-objective optimization crop spatial allocation model leverages geospatial data, including crop, soil and climate suitability, to enhance the accuracy of our model. Additionally, we incorporate agricultural economics data, such as market price, crop yield, production cost, distances to secondary producers, production budget limitations, and minimum crop production requirements. To speedup the convergence of the algorithm, we introduce more suitable crossover and mutation operators in NSGA-II, aiming to direct the search towards the Pareto optimal solutions. We demonstrate the effectiveness of our approach in a case study of the agricultural area in Chiang Mai province, Thailand, focusing on three major industrial crops: corn, cane, and rice. Our model suggests land allocation that adheres to both the budget constraint and the minimum production requirements, while retaining only a small surplus for each crop. The successful implementation of this approach in our case study marks a significant advancement in Thai agricultural research, paving the way for long-term economic and environmental sustainability.

Evolutionary Multimodal Multiobjective Optimization for Traveling Salesman Problems

Multimodal multiobjective optimization problems (MMOPs) are commonly seen in real-world applications. Many evolutionary algorithms have been proposed to solve continuous MMOPs. However, little effort has been made to solve combinatorial (or discrete) MMOPs. Searching for equivalent Pareto optimal solutions in the discrete decision space is challenging. Moreover, the true Pareto optimal solutions of a combinatorial MMOP are usually difficult to know, which has limited the development of its optimizer. In this paper, we first propose a test problem generator for multimodal multiobjective traveling salesman problems (MMTSPs). It can readily generate MMTSPs with known Pareto optimal solutions. Then we propose a novel evolutionary algorithm to solve MMTSPs. In our proposed algorithm, we develop two new edge assembly crossover operators, which are specialized in searching for superior solutions to MMTSPs. Moreover, the proposed algorithm uses a new environmental selection operator to maintain a good balance between the objective space diversity and decision space diversity. We compare our algorithm with five state-of-the-art designs. Experimental results convincingly show that our algorithm is powerful in solving MMTSPs.

基于离散人工蜂群算法的农业机器人丘陵地区农田遍历路径规划

In this study, the discrete artificial bee colony (DABC) algorithm was proposed to plan the path of agricultural robots traversing multiple fields in hilly areas. Based on the basic ABC algorithm as the framework, the path coding method was adopted, and the discrete crossover operator, reverse operator, immune operator, and single/multi-step 2-opt operator were comprehensively used to help hired bees, observing bees, and scout bees to generate new food sources. Finally, the optimized field traversal order and the entrance and exit distribution of each field were obtained. The simulation results showed that compared with the traditional ABC algorithm, the average shortest path of the DABC algorithm proposed in this study was shortened by 1.59%, accompanied by the less iterations contributing to algorithm convergence and good ability to jump out of the local optimal solution. The simulation experiment was carried out using real field data and field operation parameters. The field traversal order and the entrance and exit distribution obtained by the proposed method can effectively reduce the length of the transfer path and its repeatability. This study exhibits superiority and feasibility in the field traversal path planning of agricultural robots in hilly areas, and the trajectory coordinates output by the algorithm can provide a path reference for large-area operations of agricultural machinery drivers or unmanned agricultural machineries.

Time series forecasting of electricity consumption using hybrid model of recurrent neural networks and genetic algorithms

The forceful energy efficiency to manage the demand is essential to meet development goals. Palestine has suffered from an electricity deficit, whereas the city of Tulkarm suffers from a chronic one. The dataset was collected from Tulkarm city in Palestine; this city is considered one of the cities that suffers the most from frequent power outages. It's difficult to determine the most powerful Artificial intelligence (AI) approaches that can accurately forecast electricity consumption. This paper presents a hybrid model that combines Recurrence Neural Networks (RNNs) and Genetic Algorithms (GAs) [RNN-GAs] to forecast electricity consumption and optimize demand. In the proposed model the K-means clustering technique produces specific initial population seeding and optimization crossover operators to enhance the efficiency and find the optimal solution. The results showed that the proposed Nonlinear Autoregressive with External (Exogenous) (NARX) (NARX-GAs) with the K-means clustering technique outperforms the hybrid model NARX-GAs. The NARX-GAs-K Mean Clustering recorded an RMSE value of 0.08759, which performs a good balance with the lowest RMSE, especially in long-term forecasting, and also outperforms the other hybrid forecasting models that depend on RNN-GAs. Finally, the forecasting results of the hybrid NARX-GAs-K Mean Clustering can predict accurately the energy consumption in a city, which leads to the use of the model in similar cities to forecast and manage the demand for electricity consumption.

Directional multiobjective optimization of metal complexes at the billion-system scale.

The discovery of transition metal complexes (TMCs) with optimal properties requires large ligand libraries and efficient multiobjective optimization algorithms. Here we provide the tmQMg-L library, containing 30k diverse and synthesizable ligands with robustly assigned charges and metal coordination modes. tmQMg-L enabled the generation of 1.37 million palladium TMCs, which were used to develop and benchmark the Pareto-Lighthouse multiobjective genetic algorithm (PL-MOGA). With fine control over aim and scope, this algorithm maximized both the polarizability and highest occupied molecular orbital-lowest unoccupied molecular orbital gap of the TMCs within selected regions of the Pareto front, without requiring prior knowledge on the objective limits. Instead of genetic operations on small ligand fragments, the PL-MOGA did whole-ligand mutation and crossover operations, which in chemical spaces containing billions of systems, yielded thousands of highly diverse TMCs in an interpretable manner.

Genetic Algorithm-Based Framework for Optimization of Laser Beam Path in Additive Manufacturing

In this study, a genetic algorithm-based laser beam (LB) path optimization method is presented to improve laser-based additive manufacturing (LBAM). To emulate the LBAM process, LB irradiation of a thin metal substrate is applied. The LB path generation is formulated as the search for the optimal sequence of LB irradiation into the cells on the substrate that minimizes the fitness function, which is composed of two components, i.e., thermal fitness and process fitness. The thermal fitness is expressed by the average thermal gradient, and a simple thermal model is developed to simulate the effects of laser-induced heat input on the temperature distribution in the substrate. The process fitness regulates the suitability of the proposed LB path for the implementation of the LBAM process. In addition to standardized tool paths (i.e., raster, spiral, etc.), novel LB path generators are proposed to define the initial population of LB path solutions. To implement a genetic algorithm-based LB path optimization, a framework is proposed, and custom initialization, crossover, and mutation operators are developed for application in LBAM. The effectiveness of the proposed approach is demonstrated through a simulation case study aiming to identify LB paths that minimize the fitness function and thus provide more suitable LB path solutions with respect to the defined fitness function. Compared with the traditional trial-and-error LB path formulations, the proposed approach provides an improved and automated method for an efficient laser beam path selection in LBAM.

A multi-objective teaching-learning-based optimizer for a cooperative task allocation problem of weeding robots and spraying drones

In recent years, both intelligent robots and drones have been widely used in agriculture. This paper studies a cooperative task allocation problem of weeding robots and spraying drones (WRSDCTA) with two objectives of minimizing the maximum completion time and minimizing the total residual herbicide. Based on the characteristics of the problem, a mathematical model is established. An effective multi-objective teaching-learning-based optimizer (MOTLBO) is proposed to solve this problem. At first, two heuristic initialization methods are designed to generate high quality initial population. And then, a multi-teacher strategy is presented when selecting teachers and students. For teachers, two heuristic search operators are designed to improve personal ability. For students, an improved OBX crossover operator is designed to learn from selected teachers. In addition, a multi-neighborhood search strategy is presented to further improve the local search capability of the algorithm. Finally, the proposed MOTLBO algorithm is compared with several state-of-the-art algorithms from close-related literature, and the results demonstrate the effectiveness and high performance of MOTLBO for solving the WRSDCTA problem.

Latent Crossover for Data-Driven Multifidelity Topology Design

Abstract Topology optimization is one of the most flexible structural optimization methodologies. However, in exchange for its high level of design freedom, typical topology optimization cannot avoid multimodality, where multiple local optima exist. This study focuses on developing a gradient-free topology optimization framework to avoid being trapped in undesirable local optima. Its core is a data-driven multifidelity topology design (MFTD) method, in which the design candidates generated by solving low-fidelity topology optimization problems are updated through a deep generative model and high-fidelity evaluation. As its key component, the deep generative model compresses the original data into a low-dimensional manifold, i.e., the latent space, and randomly arranges new design candidates over the space. Although the original framework is gradient free, its randomness may lead to convergence variability and premature convergence. Inspired by a popular crossover operation of evolutionary algorithms (EAs), this study merges the data-driven MFTD framework and proposes a new crossover operation called latent crossover. We apply the proposed method to a maximum stress minimization problem in 2D structural mechanics. The results demonstrate that the latent crossover improves convergence stability compared to the original data-driven MFTD method. Furthermore, the optimized designs exhibit performance comparable to or better than that in conventional gradient-based topology optimization using the P-norm measure.