Structure optimization design method and system for resisting detonation wave and fragment group combined impact

Abstract

The invention relates to the field of anti-impact structure optimization design, in particular to a detonation wave and fragment group combined impact resistant structure optimization design method and system. The method comprises the following steps: obtaining a structure design scheme of anti-detonation wave and fragment group combined impact, and carrying out first numerical analysis on the structure design scheme to obtain a sample set; constructing a classification agent model, and training the classification agent model by using the sample set; introducing a genetic algorithm, and improving a fitness evaluation mechanism of the genetic algorithm by using the classification agent model; selecting a parent for iteration of the genetic algorithm based on the improved fitness evaluation mechanism to obtain an improved genetic algorithm; and adopting a mixed point adding strategy to optimize the selection of sample points of the classification agent model and the iteration process of the improved genetic algorithm, and determining an optimal structure design scheme for resisting detonation wave and fragment group combined impact. The problems of insufficient optimization precision and high calculation cost caused by adopting a single point adding criterion are solved.

Description

Technical Field

[0001] The present invention relates to the field of optimization design of anti-impact structures, specifically a structural optimization design method and system against combined impacts of detonation waves and fragment groups. Background Art

[0002] The combined damage effect of shock waves and high-speed fragment groups generated by an explosion on a structure is significantly different from their individual effects, and has a cumulative enhancement effect under certain conditions. Therefore, the design of a new type of protective structure needs to consider the combined damage effect of detonation waves and fragment groups. At present, the design strategies of protective structures mainly include structural design based on traditional metal materials and the development of new composite protective structures using multiphase materials to improve the protective performance of the structure against combined loads. However, the above structural designs mostly rely on engineering experience and lack the support of mathematical optimization means, making it difficult to fully explore the potential of structural performance. The optimization of anti-combined impact structures involves strong non-linearity such as materials, geometry, and component contact under combined loads of shock waves and fragment groups, resulting in time-consuming structural response solutions and difficult sensitivity solutions, which poses a great challenge to the optimization design of structures.

[0003] The optimization method based on surrogate models (surrogate-based optimization, SBO) constructs a direct mapping between inputs and outputs to replace the time-consuming numerical analysis in the optimization process, and combines corresponding intelligent optimization algorithms to achieve structural optimization design, with high optimization efficiency. This method has been successfully applied to the optimization design of structural parameters under single loads of anti-shock waves and fragments / projectiles, and has achieved certain results. Commonly used surrogate models in SBO include polynomial response surfaces, Kriging models, radial basis neural networks, and support vector machines, etc. Among them, the classification surrogate model based on support vector machines does not need to predict the specific values of the structural mechanical responses under extreme loads, and has the advantage of low calculation cost for the sample set.

[0004] The accuracy of the surrogate model has a great influence on the optimization results. In the "one-step" optimization design framework, that is, directly using the samples obtained from one sampling to establish a surrogate model for optimization and solution without involving the process of adding points, the accuracy of the obtained surrogate model is often difficult to guarantee, and the optimization results have large deviations. In the process of structural optimization design, the research on the accuracy of the surrogate model is also of great significance. Therefore, it is urgent to develop an effective structural optimization design method against combined impacts. Summary of the Invention

[0005] Aiming at the defects in the prior art, the present invention provides a structural optimization design method and system against combined impacts of detonation waves and fragment groups, which solves the problems of insufficient optimization accuracy and high calculation cost faced by using a single point addition criterion.

[0006] To achieve the above object, an aspect of the present invention provides a structural optimization design method for combined impact of anti-detonation wave and fragment group, and the method includes: obtaining a structural design scheme for combined impact of anti-detonation wave and fragment group, performing a first numerical analysis on the structural design scheme to obtain a sample set; constructing a classification surrogate model, and training the classification surrogate model by using the sample set; introducing a genetic algorithm, and improving the fitness evaluation mechanism of the genetic algorithm by using the classification surrogate model; based on the improved fitness evaluation mechanism, selecting parents for iteration of the genetic algorithm to obtain an improved genetic algorithm; adopting a hybrid sampling strategy to optimize the selection of sample points of the classification surrogate model and the iteration process of the improved genetic algorithm, and determining an optimal structural design scheme for combined impact of anti-detonation wave and fragment group.

[0007] The present invention obtains a sample set through numerical analysis and trains a classification surrogate model, which can accurately simulate the response of the structure under the combined impact of detonation wave and fragment group, providing a reliable basis for the design. The fitness evaluation mechanism of the improved genetic algorithm enables the algorithm to more efficiently screen out individuals adapted to the environment and avoid falling into local optimal solutions. The hybrid sampling strategy optimizes the selection of sample points and the algorithm iteration, further explores the design space, improves the optimization quality, speeds up the convergence rate, and reduces the calculation cost.

[0008] Optionally, the improving the fitness evaluation mechanism of the genetic algorithm by using the classification surrogate model includes: using the classification surrogate model to divide individuals in the population of the genetic algorithm into feasible solutions and infeasible solutions; calculating the diversity difference between the infeasible solutions and the individuals in the population to obtain an infeasible solution diversity difference value; and improving the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value.

[0009] The present invention uses the classification surrogate model for prediction, providing data support for evaluating the population, and can quickly distinguish between feasible solutions and infeasible solutions. Calculating the diversity difference between the infeasible solutions and the population, the obtained infeasible solution diversity difference value introduces a new dimension for fitness evaluation, avoiding the algorithm from prematurely converging to local optimal solutions. Finally, improving the fitness evaluation mechanism based on this difference value makes the genetic algorithm more in line with the target requirements of the present invention. At the same time, it makes the genetic algorithm more scientific and flexible in the iteration process, enhances its global search ability, and effectively improves the search efficiency.

[0010] Optionally, the improving the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value includes: calculating an objective function of the improved genetic algorithm according to the infeasible solution diversity difference value to obtain an objective function value; sorting the individuals in the population according to the objective function value to obtain a sorting position; and calculating the fitness value of the individuals in the population according to the sorting position to obtain a population fitness value.

[0011] The present invention calculates the objective function through the infeasible solution diversity difference value, making the objective function more in line with the actual problem, enhancing the exploration ability of the solution space. Ranking the population individuals based on the objective function value can intuitively distinguish the advantages and disadvantages of individuals, providing a clear guidance for subsequent operations. Calculating the fitness value according to the ranking position can avoid the problems of excessive or too small selection pressure caused by directly using the objective function value, and reasonably balance the population diversity and the convergence speed. This not only prevents the algorithm from converging prematurely, but also speeds up the process of approaching the optimal solution, enabling the algorithm to find high-quality solutions more efficiently in complex optimization problems.

[0012] Optionally, the improved fitness evaluation mechanism is used to iteratively select parents for the genetic algorithm to obtain an improved genetic algorithm, including: in the iterative process of the genetic algorithm, according to the population fitness value, using the roulette wheel selection method to calculate the probability of an individual in the population being a parent; determining the parents in the iterative process of the genetic algorithm according to the probability to obtain the improved genetic algorithm.

[0013] The present invention uses the roulette wheel selection method to make the probability of an individual in the population being selected proportional to its fitness value, thereby calculating the probability of becoming a parent. This method has randomness, can avoid the algorithm from falling into a local optimal solution prematurely, enhance the global optimization ability of the genetic algorithm, and find the optimal solution faster.

[0014] Optionally, the hybrid sampling strategy is adopted to optimize the selection of sample points of the classification surrogate model and the iterative process of the improved genetic algorithm, and determine the optimal structural design scheme for the combined impact of anti-detonation wave and fragment group, including: obtaining the decision boundary according to the classification surrogate model; updating the sample set according to the decision boundary and the hybrid sampling strategy; optimizing and training the classification surrogate model with the updated sample set; using the optimized and trained classification surrogate model and the improved genetic algorithm for optimization and solution to determine the optimal structural design scheme for the combined impact of anti-detonation wave and fragment group.

[0015] The present invention obtains the decision boundary through the classification surrogate model, which can clearly define the effective area of the solution space. Based on this boundary and the hybrid sampling strategy, optimized sample points are generated, which can accurately supplement high-quality sample points, improve the sample quality of the classification surrogate model, and be used to optimize and train the classification surrogate model, thereby further improving the prediction accuracy of the classification surrogate model. Combining with the improved genetic algorithm for optimization and solution can efficiently search for the global optimal solution under the guidance of a more accurate model. Through cyclic iterative optimization, the optimal structural design scheme for the combined impact of anti-detonation wave and fragment group can be quickly and accurately determined finally, effectively enhancing the structural protection ability.

[0016] Optionally, the updating of the sample set according to the decision boundary and the hybrid sampling strategy includes: obtaining a structurally optimized design set and a first structurally optimized design according to the improved genetic algorithm; obtaining one or more second structurally optimized designs by calculating the distance from the structurally optimized design set to the decision boundary; performing a second numerical analysis on the first structurally optimized design and the second structurally optimized designs, and updating the sample set according to the results of the second numerical analysis.

[0017] The present invention improves the data quality of the sample set by adding new high-quality data sample points.

[0018] Optionally, the use of the optimized and trained classification surrogate model and the improved genetic algorithm for optimization solution to determine the optimal structural design scheme against the combined impact of detonation waves and fragment groups includes: using the optimized and trained classification surrogate model and the improved genetic algorithm for optimization solution to obtain a local structural optimization design; obtaining a local structural optimal design according to the sample points of the classification surrogate model; calculating the maximum value of the relative deviation between the local structural optimization design and the local structural optimal design to obtain the maximum relative deviation value; setting a threshold, and determining whether the maximum relative deviation value is less than the threshold. If so, determining the local structural optimal design as the optimal structural design scheme against the combined impact of detonation waves and fragment groups.

[0019] The present invention uses the optimized and trained classification surrogate model and the improved genetic algorithm for solution, which can efficiently explore the design space and obtain local structural optimization designs. Obtaining the local structural optimal design through sample points provides a reliable reference for scheme evaluation. Calculating the relative deviation between the two and comparing it with the threshold can effectively judge the convergence degree of the design scheme. If the relative deviation is less than the threshold, determining the local structural optimal design as the final scheme not only ensures the accuracy and stability of the scheme but also avoids excessive calculation, improving the efficiency of determining the optimal structural design scheme against the combined impact of detonation waves and fragment groups.

[0020] Optionally, the structurally optimized design set includes multiple structurally optimized designs. The obtaining of one or more second structurally optimized designs by calculating the distance from the structurally optimized design set to the decision boundary includes: calculating the distance from each of the structurally optimized designs in the structurally optimized design set to the decision boundary; selecting one or more of the structurally optimized designs from the structurally optimized design set according to the distance to form candidate sample points; screening the candidate sample points by calculating the difference between the candidate sample points and the sample set to obtain one or more second structurally optimized designs.

[0021] The present invention calculates the distance from each of the structural optimization designs in the centralized optimization design of the calculation structure to the decision boundary, and selects sample points according to the distance, which can focus on the key areas near the decision boundary, mine design points with potential value, and these points often contain key information that determines the quality of the design scheme. Secondly, by calculating the difference between the candidate sample points and the sample set and removing the points with small differences, sample redundancy can be effectively avoided and the diversity of the samples can be increased.

[0022] Optionally, the objective function value satisfies the following formula: , where is the objective function processed based on the constraint handling strategy, is an individual in the population of the genetic algorithm, is the individual in the population 's structural quality, is a feasible solution, is the maximum structural quality of the individuals in the population, is the infeasible solution diversity difference value.

[0023] The fitness value calculation formula of the present invention incorporates infeasible solutions into the calculation of the entire fitness value, fully considering the potential effective information of the infeasible solution diversity, and improving the applicability of the present invention.

[0024] Another aspect of the present invention also provides a structural optimization design system for resisting combined shock of detonation waves and fragment groups. The system includes a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. Among them, the memory is used to store computer programs, and the computer programs include program instructions. The processor is configured to call the program instructions to execute a structural optimization design method for resisting combined shock of detonation waves and fragment groups described in the previous aspect of the present invention.

[0025] The structural optimization design system for resisting combined shock of detonation waves and fragment groups of the present invention is compact in structure, stable in performance, high in integration, and simple in composition, and can stably execute a structural optimization design method for resisting combined shock of detonation waves and fragment groups provided in the previous aspect of the present invention, further improving the overall applicability and practical application ability of the present invention. Description of the Drawings

[0026] Figure 1 is a flowchart of the structural optimization design method for resisting combined shock of detonation waves and fragment groups according to an embodiment of the present invention; Figure 2 is an algorithm flowchart of the structural optimization design method for resisting combined shock of detonation waves and fragment groups according to an embodiment of the present invention; Figure 3Schematic diagram of the structural optimization design system for combined impact of anti-detonation wave and fragment group in an embodiment of the present invention. Detailed implementation manners

[0027] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only for illustrative purposes and do not limit the present invention. In the following description, in order to provide a thorough understanding of the present invention, a large number of specific details are set forth. However, it is obvious to those of ordinary skill in the art that the present invention does not have to be implemented with these specific details. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0028] Throughout the specification, the reference to "an embodiment", "embodiment", "an example" or "example" means that a specific feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Thus, the phrases "in an embodiment", "in embodiments", "an example" or "example" appearing throughout the specification do not necessarily all refer to the same embodiment or example. In addition, specific features, structures, or characteristics can be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. In addition, those of ordinary skill in the art should understand that the drawings provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0029] Figure 1 Flowchart of the structural optimization design method for combined impact of anti-detonation wave and fragment group in an embodiment of the present invention. To solve the problems of insufficient optimization accuracy and high computational cost faced by a single addition criterion, as Figure 1 shown, the method includes the following steps: Step S1, obtain a structural design scheme for combined impact of anti-detonation wave and fragment group, and perform a first numerical analysis on the structural design scheme to obtain a sample set.

[0030] In this embodiment, obtaining the structural design scheme for combined impact of anti-detonation wave and fragment group specifically includes the following sub-steps: Step S101, determine the optimization design variables of the structural design scheme for combined impact of anti-detonation wave and fragment group, and set the value range of the optimization design variables.

[0031] The optimization design variables include the type of lattice sandwich (trapezoidal or concave hexagon), the number of lattice layers , the thickness of the lattice wall panel , the height , the flanging angle , the thickness of the front panel , the thickness of the back panel , the thickness of the polyurea panel , the thickness of the ceramic plate , the lamination sequence of both the polyurea plate and the ceramic plate , and the lamination sequence of the lattice structure and the ceramic polyurea . In this embodiment, the back panel being intact ( ) is used as the optimization constraint, and lightweight is used as the optimization goal to carry out the performance optimization design under the combined structure load.

[0032] In an alternative embodiment, the value range of the optimization design variables includes , , , , , , , , , and . Among them, in the formula, indicates that the structure protection side is intact, indicates that the polyurea is above the ceramic, and vice versa it is below, indicates that the lattice structure is above the ceramic / polyurea layer, and vice versa it is below; indicates that the lattice type is trapezoidal, and vice versa it is an inward concave hexagon.

[0033] Step S102, according to the optimization design variables and the value range of the optimization design variables, use the Latin hypercube sampling method to determine multiple structural design solutions against the combined impact of detonation waves and fragment groups in the design space as initial sample points.

[0034] In this implementation, according to the optimization design variables, an optimization design combination formed by the value range of the optimization design variables forms a design space. The Latin hypercube sampling method is used to determine multiple sample points in the design space as initial sample points, where each initial sample point is a structural design solution against the combined impact of detonation waves and fragment groups.

[0035] The Latin hypercube sampling method is an efficient sampling technique for generating samples of multivariate distributions. Its principle is to divide the value range of each optimization design variable into several intervals with equal probabilities, and then randomly select a sample point within each interval and combine these sample points to form a sample set. This sampling method can ensure that each interval of each optimization design variable has a sample point, and the sample points are more evenly distributed within the entire value range. Compared with random sampling, it can more effectively cover the optimization design variable space.

[0036] After obtaining the structural design solution against the combined impact of detonation waves and fragment groups through the above method, perform numerical analysis on the structural design solution to obtain the sample label of the structural design solution , forming a sample set In the process of constructing the numerical model, we first conduct geometric modeling, determine the constitutive model, discretize the material domain, set the geometric boundary conditions and contact conditions, etc. Through numerical simulation, we can obtain each structural design scheme. If the protection performance value meets the protection index, the label of the structural design solution Set to -1, otherwise 1, and finally get the sample set .

[0037] Specifically, numerical analysis is a mathematical modeling and calculation method used to predict and evaluate the performance of structural designs under specific conditions. It starts with geometric modeling, which involves creating a three-dimensional model of the structure using computer-aided design (CAD) software. Geometric modeling needs to accurately reflect the size, shape, and material distribution of the design to ensure the accuracy of subsequent analysis.

[0038] After completing the geometric modeling, the next step is to determine the constitutive model, that is, to select a mathematical model that is suitable for describing the behavior of the structure. This may include elastic, plastic, viscoelastic or other more complex material behavior models. The choice of constitutive model is crucial to the accuracy of the simulation results because it determines how the material responds when subjected to force.

[0039] Subsequently, the material domain is discretized, that is, the continuous material domain is divided into a finite number of small units, which can be the smallest computational unit that the structural analysis software can handle. The quality and method of discretization directly affect the accuracy and computational efficiency of numerical simulation.

[0040] After the discretization is completed, it is necessary to set the geometric boundary conditions and contact conditions. The geometric boundary conditions define the fixed and free boundaries of the structure, while the contact conditions describe the interaction between the structure and other objects or materials. These conditions are crucial to simulate the performance of the structure in the actual working environment.

[0041] After completing the above steps, run the model through numerical simulation to obtain each structural design solution The protection performance values of the structure are as follows. These performance values include but are not limited to key indicators such as stress, strain, deformation, and energy absorption. Based on the comparison between these performance values and the preset protection indicators, labels are assigned to each structural design scheme. If the performance value meets the protection index, the label is set to -1, indicating that the design is feasible; if not, the label is 1, indicating that it is not feasible. Finally, all structural design schemes and their corresponding labels are combined to form a sample set. This sample set includes a comprehensive evaluation of the initial design solution and provides the necessary data support for the subsequent optimization algorithm.

[0042] In this way, numerical analysis not only verifies the feasibility of the design scheme, but also provides a basis for the fitness evaluation mechanism of the genetic algorithm, thus guiding the optimization process to develop in the direction of meeting the design requirements. This process is an indispensable part of the structural optimization design, which ensures the effectiveness and reliability of the design scheme in practical applications.

[0043] Step S2: Construct a classification surrogate model and train the classification surrogate model using the sample set.

[0044] In this embodiment, a classification surrogate model is constructed based on the support vector machine, and the sample set is used to train the classification surrogate model. The support vector machine divides the design space into two regions by finding a decision boundary, where the design points that meet the structural protection performance index are on one side, and the design points that do not meet the protection performance index are on the other side.

[0045] The support vector machine for constructing the classification surrogate model has many remarkable advantages. First of all, it divides different categories by finding the maximum margin hyperplane. This optimization objective based on the geometric margin enables the model to have strong generalization ability during classification and can learn a relatively robust decision boundary on a limited number of training samples. Secondly, the support vector machine uses the kernel trick, which can map the linearly inseparable data in the low-dimensional space to the high-dimensional space, thus realizing complex non-linear classification and greatly expanding its application scope. In addition, the complexity of the classification surrogate model mainly depends on the number of support vectors rather than the dimension of the data, which makes it perform well in dealing with high-dimensional data and has high computational efficiency. Finally, the support vector machine has a solid theoretical basis. Based on the VC dimension and the principle of structural risk minimization of statistical learning theory, it can effectively avoid overfitting even in the case of small samples, ensuring the stability and reliability of the model.

[0046] Step S3: Introduce the genetic algorithm and improve the fitness evaluation mechanism of the genetic algorithm using the classification surrogate model.

[0047] In this implementation, the genetic algorithm is an optimization algorithm that simulates the biological evolution process. Based on natural selection and genetic mechanisms, it gradually evolves the optimal solution from the initial population through operations such as selection, crossover, and mutation. The core idea of the genetic algorithm is "survival of the fittest", that is, individuals with high fitness have a greater probability of being selected to participate in reproduction, so as to pass on the excellent genes to the next generation.

[0048] Among them, improving the fitness evaluation mechanism of the genetic algorithm using the classification surrogate model specifically includes the following sub-steps: Step S301: Use the classification surrogate model to divide the individuals in the population of the genetic algorithm into feasible solutions and infeasible solutions.

[0049] In this embodiment, the classification agent model can divide the population of the genetic algorithm into feasible solutions and infeasible solutions. The population of the genetic algorithm comes from the sample set. In an alternative embodiment, the population of the genetic algorithm can also come from new sample points.

[0050] Step S302: Calculate the diversity difference between the infeasible solutions and the individuals in the population to obtain the infeasible solution diversity difference value.

[0051] For constrained optimization problems, it is usually necessary to transform them into unconstrained optimization problems to be applicable to genetic algorithm solutions. This criterion requires that feasible solutions are always better than any infeasible solutions, and among the feasible solutions, the individual solutions with smaller objective function values are dominant, and among the infeasible solutions, the individual solutions with lower degrees of constraint violation are dominant. Therefore, the unconstrained optimization problem obtained according to the traditional processing method is: , where, is the objective function of the transformed unconstrained optimization problem, is the structural quality of the individual in the population, is the feasible solution; is the maximum structural quality of the individuals in the population, is the predicted value of the classification agent model for the structural performance.

[0052] However, as can be seen from the above formula, the individual in the infeasible solution is always +1, resulting in the inability to determine the dominant individual solution. For this reason, the present invention proposes to adopt a strategy of using individual diversity instead of default degree evaluation to solve the problem that the dominant individual of the infeasible solution cannot be determined.

[0053] Diversity difference refers to the difference between the infeasible solution and other individuals in the population. Therefore, the infeasible solution diversity difference value satisfies the following formula: , where, is the infeasible solution diversity difference value, is the population size in the genetic algorithm, is the sample point 's feature dimension, that is, the dimension of the optimization design variable, is the value of the infeasible solution in the i-th dimension, is the -th individual in the population at the i-th dimension value.

[0054] Step S303: Improve the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value.

[0055] Among them, improving the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value specifically includes the following sub-steps: Step S30301: Calculate the objective function of the improved genetic algorithm according to the infeasible solution diversity difference value to obtain the objective function value.

[0056] The objective function value satisfies the following formula: , where, is the objective function processed based on the constraint handling strategy, is an individual in the population of the genetic algorithm, is the structure quality of the individual in the population, is a feasible solution, is the maximum structure quality of the individuals in the population, is the infeasible solution diversity difference value.

[0057] Step S30302: Sort the individuals in the population according to the objective function value to obtain the sorting positions.

[0058] In this embodiment, sorting the individuals in the population according to the objective function value can clearly show the relative superiority and inferiority of each individual in the current population. Through the quantitative standard of the objective function value, all individuals are measured on the same scale, enabling the algorithm to intuitively distinguish which individuals are closer to the ideal solution of the problem and which individuals still have great room for improvement. This clear understanding of the individual performance provides a clear direction for subsequent genetic operations such as selection, crossover, and mutation.

[0059] Based on the sorting positions, the algorithm can be more targeted when selecting parent individuals. For example, in the selection process, it tends to select individuals with a higher ranking, that is, individuals with better objective function values, as parents. This conforms to the concept of survival of the fittest in natural selection and helps to pass on excellent genes to the next generation, thus promoting the evolution of the population towards a better direction. At the same time, the sorting also provides a basis for adaptively adjusting the parameters of the genetic algorithm. According to the sorting results, the algorithm can dynamically adjust key parameters such as the crossover probability and mutation probability. For individuals with a higher ranking, the mutation probability is appropriately reduced to protect excellent gene combinations. For individuals with a lower ranking, the mutation probability is appropriately increased to increase their possibility of exploring new solution spaces, thereby enhancing the global search ability of the algorithm.

[0060] In addition, the sorting operation also facilitates the performance evaluation of the algorithm. By observing the changes in the ranking of individuals under different numbers of iterations, you can intuitively understand the convergence trend of the algorithm. If in consecutive iterations, the top-ranked individuals gradually stabilize and the objective function value is continuously optimized, it means that the algorithm is converging effectively; conversely, if the individual ranking fluctuates greatly and the objective function value does not increase significantly, it indicates that the algorithm may be trapped in a local optimum or there is a problem in the search process, and the algorithm needs to be adjusted. This sorting-based performance evaluation method provides strong support for the optimization and improvement of the algorithm, enabling it to better cope with various complex optimization problems. Step S30303, calculating the fitness values of the individuals in the population according to the sorting positions, and obtaining the population fitness value.

[0061] In this embodiment, the fitness values of individuals in the population are calculated based on the method of grade division. The present invention can effectively avoid the problem of imbalanced selection pressure caused by excessive differences in fitness values. The traditional method of directly calculating fitness based on the objective function may cause some individuals with extremely high fitness to dominate the selection process, causing the algorithm to converge to the local optimal solution prematurely. However, when calculating fitness based on the ranking position, no matter how big the difference in the actual objective function values of individuals is, fitness is assigned according to their relative order in the population, so that different individuals have a relatively reasonable opportunity to participate in reproduction during the selection process, which helps to maintain the diversity of the population.

[0062] At the same time, this method enhances the algorithm's adaptability to complex problems. In many problems, the objective function has a complex form and there are multiple local optimal areas. By calculating fitness by sorting positions, the algorithm will not focus too much on the individuals that seem to be the best at the moment, but will be able to explore in a wider solution space. It encourages the algorithm to fully explore potential solutions in different areas in the early stages of the search, and as the iterations proceed, better individuals are gradually selected, making it more likely to find the global optimal solution.

[0063] In addition, the computational complexity of calculating fitness values based on ranking positions is relatively low. There is no need to perform complex mathematical operations on the fitness of each individual. It is only necessary to determine the ranking position of the individual in the population, and then assign the fitness value according to the preset rules. This can significantly reduce the consumption of computing resources and computing time when dealing with large-scale populations, improve the operating efficiency of the algorithm, enable the algorithm to handle large-scale complex problems more efficiently, and provide a more feasible solution for practical applications.

[0064] The population fitness value satisfies the following formula: , In the formula, is the fitness of the ith individual, is the sorting position of the i-th individual within the population. is the selection pressure.

[0065] The said selection pressure reflects the potential ability of the selection operation to improve the fitness of the population and measures the ability of the selection operator to select relatively excellent individuals from the current population.

[0066] A larger selection pressure means that the average fitness of the selected parent individuals is significantly higher than the average fitness of the population. This will enable the genes of excellent individuals in the population to spread faster in the next generation, accelerating the convergence speed of the population towards the optimal solution. However, if the selection pressure is too large, it may lead to premature convergence of the algorithm and miss the global optimal solution. Because the algorithm will overly rely on the current excellent individuals and ignore other potential solution spaces.

[0067] A smaller selection pressure indicates that the selection operation has little discrimination between individuals, and the average fitness of the parent individuals is close to the average fitness of the population. This helps to maintain the diversity of the population because more individuals with different fitness levels have the opportunity to participate in reproduction. But if the selection pressure is too small, the evolution speed of the algorithm will slow down because the advantages of excellent individuals cannot be fully utilized, and the overall fitness of the population improves slowly.

[0068] Therefore, the value of the selection pressure can be flexibly selected according to different actual situations, or it can be gradually adjusted during the calculation process.

[0069] Step S4, based on the improved fitness evaluation mechanism, select parents for the iteration of the genetic algorithm to obtain an improved genetic algorithm.

[0070] Among them, based on the improved fitness evaluation mechanism, selecting parents for the iteration of the genetic algorithm to obtain an improved genetic algorithm specifically includes the following sub-steps: Step S401, during the iteration of the genetic algorithm, according to the population fitness value, use the roulette wheel selection method to calculate the probability of an individual in the population being a parent.

[0071] The probability of an individual in the population being a parent satisfies the following formula: , where is the probability of an individual in the population being a parent, is the population size in the genetic algorithm, is the fitness of the i-th individual.

[0072] In this embodiment, during the iterative process of the genetic algorithm, the probability of an individual becoming a parent is calculated according to the roulette wheel selection method based on the population fitness value. The roulette wheel selection method closely adheres to the theory of natural selection, taking fitness as the key indicator to measure the quality of an individual. Individuals with higher fitness have a greater chance of being selected as parents during the rotation of the "roulette wheel", so as to pass on excellent genes and guide the population to develop towards a higher fitness direction, gradually optimizing the population to approach the optimal solution of the problem.

[0073] This selection method ingeniously incorporates randomness. Even individuals with poor fitness have a certain probability of participating in the reproduction process. This characteristic is of great significance for maintaining the diversity of the population and can effectively prevent the algorithm from falling into the dilemma of local optimal solutions at an early stage. In a complex search space, individuals that seem ordinary at the beginning may, in the subsequent gene combination and evolution process, show unique value and open up new paths to find the global optimal solution.

[0074] Another prominent advantage of the roulette wheel selection method lies in its simplicity of calculation. Only by calculating the proportion of an individual's fitness value in the total population fitness can the probability of the individual becoming a parent be determined. This simple calculation method greatly reduces the computational complexity of the algorithm. When facing a large-scale population or limited computing resources, it can significantly improve the operating efficiency of the algorithm and enable it to operate efficiently. For this reason, the roulette wheel selection method can easily be integrated with other operators such as crossover and mutation in the genetic algorithm to jointly promote the stable operation of the genetic algorithm and provide strong support for solving various complex problems.

[0075] Step S402: Determine the parents in the iterative process of the genetic algorithm according to the probability to obtain the improved genetic algorithm.

[0076] In this embodiment, determining the parents in the genetic algorithm iteration according to the probability can simulate natural selection, allowing individuals with higher fitness to have a greater chance of passing on genes and promoting the population to evolve towards a better direction. At the same time, probabilistic selection gives low-fitness individuals opportunities, avoids premature convergence, maintains population diversity, and helps the algorithm explore more effectively in a complex search space, and thus find the global optimal solution.

[0077] Based on the improvement of the above genetic algorithm, the improved genetic algorithm is further obtained.

[0078] Step S5: Adopt a mixed sampling strategy to optimize the selection of sample points of the classification agent model and the iterative process of the improved genetic algorithm, and determine the optimal structural design scheme for the combined impact of anti-detonation waves and fragment groups.

[0079] Among them, the hybrid dot-adding strategy is adopted to optimize the selection of sample points of the classification proxy model and the iterative process of the improved genetic algorithm, and the specific sub-steps for determining the optimal structural design scheme against the combined impact of detonation waves and fragment swarms are as follows: Step S501: Obtain the decision boundary according to the classification proxy model.

[0080] In this embodiment, to find a decision boundary with high classification accuracy, it is necessary to make the "margin" of data points to the decision boundary as large as possible while satisfying the constraint of correct classification, that is: , wherein, in the formula, is the number of training samples, is the Lagrange multiplier of sample i, is the Lagrange multiplier of sample j, is the sample label, is the sample label, is the kernel function, used to evaluate the similarity between the input sample and the th sample in the sample set, is the penalty parameter.

[0081] Solve the above formula through an optimization method to obtain the Lagrange multiplier of each sample point. Using any one of the support vectors , the bias can be determined. The formula is as follows: , Determine and After that, the decision boundary of the support vector machine can be obtained for the classification of structural protection performance.

[0082] The decision boundary satisfies the following formula: , wherein, is the decision boundary, is the number of training samples, is the Lagrange multiplier of the sample point , is the label, is the kernel function, used to calculate the similarity between the sample and , is the bias term.

[0083] According to The value of can be used to determine the influence of sample points on the decision boundary. When it is the case, the sample points are usually far from the decision boundary and have been correctly classified, having no direct influence on the decision boundary; when it is the case, the sample points are located near the classification boundary, are support vectors, and directly affect the classification margin and the position of the decision boundary.

[0084] Step S502: Update the sample set according to the decision boundary and the hybrid sampling strategy.

[0085] In this embodiment, the core idea of the hybrid sampling strategy is to ensure that the optimization algorithm has good local exploitation and global search capabilities. The hybrid sampling criterion adds 1 sample through local sampling and m samples through global sampling in one sequence optimization loop.

[0086] Among them, updating the sample set according to the decision boundary and the hybrid sampling strategy specifically includes the following sub-steps: Step S50201: Obtain a structural optimization design set and a first structural optimization design according to the improved genetic algorithm.

[0087] In this embodiment, the structural optimization design set refers to the improved genetic algorithm optimized population obtained after the population of the improved genetic algorithm goes through steps such as constraint handling, fitness calculation, individual selection, and crossover and mutation of the improved genetic algorithm. The first structural optimization design refers to the best individual in the optimized population, and the best individual is the individual with the highest fitness.

[0088] The first structural optimization design forms the sample for local sampling. The minimum model prediction criterion is one of the earliest applied sampling methods in sequence optimization design and is widely adopted because of its simple operation and easy implementation. This method can quickly guide the optimization algorithm to converge and shows strong local exploitation ability. Therefore, this technology uses the minimum model prediction criterion for local sampling. In each loop, the current surrogate model is optimized, and the obtained optimal design point is added to the sample set, and the surrogate model is updated accordingly. The mathematical expression of this criterion is: , In the formula, "arg" represents the mapping relationship between the independent variable and the dependent variable, and the return value is the current minimum objective corresponding structural parameters.

[0089] However, in the optimization design problem of resisting combined loads, due to the large number of optimization variables and the complex surrogate relationship with the protection performance, the global classification accuracy of the surrogate model is low, resulting in the easy omission of the global optimal point. In addition, the structural response analysis for resisting combined loads has a high computational cost, and structural response analysis needs to be carried out for each optimal point. If only one optimal point is added each time, the number of iterations will increase significantly, resulting in an unbearable computational cost.

[0090] Therefore, the present technology proposes a global sampling point adding method for a support vector machine classification surrogate model, aiming to improve the accuracy of the surrogate model by sampling in sensitive areas with large prediction deviations, so as to guide the optimization to move towards the global optimum.

[0091] Step S50202, obtain one or more second structural optimization designs by calculating the distance from the structural optimization design set to the decision boundary.

[0092] Among them, obtaining one or more second structural optimization designs by calculating the distance from the structural optimization design set to the decision boundary specifically includes the following sub-steps: Step S5020201, calculate the distance from each structural optimization design in the structural optimization design set to the decision boundary.

[0093] In this embodiment, calculate the distance from each structural optimization design in the structural optimization design set to the decision boundary. The decision boundary not only determines the position of the classification boundary, but also can quantify the distance from the sample point to the boundary. The larger the distance, the higher the classification confidence. The smaller the distance, it indicates that the sample point is near the classification boundary and makes a more significant contribution to classification, that is, the support vector. Since the sample points near the decision boundary are more likely to be misclassified, this area is regarded as a sensitive area. By sampling in these areas, the prediction accuracy of the surrogate model can be effectively improved.

[0094] Step S5020202, select one or more structural optimization designs from the structural optimization design set according to the distance to form candidate sample points.

[0095] In this embodiment, sort all the sample points according to the distance, select one or more structural optimization designs closest to the decision boundary, and use these structural optimization designs as candidate sample points for standby.

[0096] These sample points are near the decision boundary and contain rich structural feature information. The decision boundary is the key boundary for distinguishing different design performance regions. The sample points close to it can reflect the subtle changes from one performance to another, which helps to deeply understand the complex relationship between design variables and performance.

[0097] Secondly, subsequent analysis based on these sample points can effectively improve the optimization efficiency. Compared with blindly exploring in the entire design set, focusing on the vicinity of the decision boundary can more accurately locate the optimization direction and avoid wasting computing resources in a large number of ineffective regions far from the boundary.

[0098] Furthermore, the candidate sample points are helpful for discovering potential innovative designs. Since they are in the sensitive region of performance change, in-depth study of them may reveal new ideas that break through traditional designs, thus bringing more innovative and competitive solutions for structural optimization and promoting the development of the field of structural design.

[0099] Step S5020203, screen the candidate sample points by calculating the difference between the candidate sample points and the sample set, and obtain one or more second structural optimization designs.

[0100] In this embodiment, in order to avoid duplicate selection of samples, it is necessary to ensure that the candidate sample points have sufficient differences from the existing sample set. Therefore, among the selected one or more samples, samples with smaller differences need to be excluded, and the remaining samples are retained as the second structural optimization design.

[0101] The difference between the candidate sample points and the sample set satisfies the following formula: , In the formula, represents the difference between sample i and sample j. Set a threshold. When is less than the threshold, the newly selected sample is excluded.

[0102] Step S50203, perform a second numerical analysis on the first structural optimization design and the second structural optimization design, and update the sample set according to the results of the second numerical analysis.

[0103] In this embodiment, perform a numerical analysis on the first structural optimization design and the second structural optimization design to obtain labels, and add the results after the numerical analysis to the sample set to realize data update of the sample set. This method is to improve the data quality of the sample set. The updated sample set.

[0104] Step S503, optimize and train the classification proxy model by using the updated sample set.

[0105] In this embodiment, through the above optimization calculation process, more accurate structural scheme designs at the boundary between feasible and infeasible solutions and optimized structural scheme designs can be obtained. These structural scheme designs are used to update the sample set. Therefore, the sample set reflects the current optimal and most diverse individuals in the population. Using this updated sample set to retrain the classification surrogate model enables the classification surrogate model to learn the latest design space distribution and constraint conditions. This helps the classification surrogate model more accurately identify feasible solutions that meet the design requirements and infeasible solutions that need further improvement. In addition, the optimized and trained classification surrogate model can more effectively guide the selection operation of the genetic algorithm, preferentially selecting those individuals that are more likely to produce high-quality offspring. Ultimately, this process not only improves the performance of the classification surrogate model but also accelerates the convergence speed of the genetic algorithm, reducing the number of iterations required to reach the optimal solution. Through continuous iteration and optimization, a more accurate and reliable optimal structural design scheme can be obtained to meet the complex design requirements of joint impact resistance to detonation waves and fragment groups.

[0106] Step S504: Use the optimized and trained classification surrogate model and the improved genetic algorithm for optimization solution to determine the optimal structural design scheme for joint impact resistance to detonation waves and fragment groups.

[0107] Among them, using the optimized and trained classification surrogate model and the improved genetic algorithm for optimization solution to determine the optimal structural design scheme for joint impact resistance to detonation waves and fragment groups specifically includes the following sub-steps: Step S50401: Use the optimized and trained classification surrogate model and the improved genetic algorithm for optimization solution to obtain a local structural optimization design.

[0108] In this embodiment, the updated classification surrogate model improves the accuracy of classification prediction. Combining with the improved genetic algorithm for optimization solution, the improved genetic algorithm enters the iterative process. After the iteration ends, the individual with the highest fitness is used as the local structural optimization design.

[0109] Step S50402: Obtain the local structural optimal design according to the sample points of the classification surrogate model.

[0110] In this embodiment, the structural optimal design among the sample points of the classification surrogate model is the first structural optimization design determined at the end of the previous iteration of the genetic algorithm. The first structural optimization design is used as the local structural optimal design.

[0111] Step S50403: Calculate the maximum value of the relative deviation between the local structural optimization design and the local structural optimal design to obtain the maximum relative deviation value.

[0112] The maximum relative deviation value satisfies the following formula: , is the maximum relative deviation value, is the first component of the local structure optimization design, is the first component of the optimal design of the local structure, is the th component of the local structure optimization design, is the th component of the optimal design of the local structure, is the feature dimension of the sample point .

[0113] Step S50404, set a threshold, and determine whether the maximum relative deviation value is less than the threshold. If so, determine the optimal design of the local structure as the optimal structural design solution for the combined impact of anti-detonation wave and fragment group.

[0114] In this embodiment, setting the threshold is to set the convergence condition. The setting of the threshold can be flexibly selected according to different required precisions. The smaller the threshold, the higher the final optimization design precision. At the same time, the calculation cost will also increase accordingly.

[0115] When the relative deviation value is less than the threshold, it indicates that the optimized structural design solution output by the improved genetic algorithm has met the requirements. At the same time, it also indicates that the classification surrogate model has achieved an ideal effect and no further optimization training is required.

[0116] When the relative deviation value is greater than the threshold, it indicates that the optimized structural design solution output by the improved genetic algorithm does not meet the requirements, and further optimization training of the classification surrogate model is required. At this time, the algorithm enters a loop.

[0117] Specifically, as Figure 2 shown, the present invention starts the optimization calculation. Initial sample points are generated through a sampling method and input , and then the sample point responses are obtained using a numerical model . At this time, the sample set is obtained. Using the sample set to train the surrogate model of the support vector machine to obtain a classification surrogate model, and using the surrogate model to assist the genetic algorithm to solve the anti-combined load optimization problem. Then, it is judged whether the convergence condition converges. If it converges, the final optimized design can be output, and the entire optimization calculation process ends. If it does not converge, new sample points are obtained through the point addition strategy, and the corresponding protection performance is obtained through parallel calculation, forming a new sample set, and using the new sample set to continue training the classification surrogate model and entering the next iteration until the convergence condition is finally reached.

[0118] As Figure 3 shown, on the other hand, the present invention also provides a structural optimization design system for combined impact of anti-detonation wave and fragment group, including: a processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected. Among them, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the relevant steps of the relevant embodiments in the structural optimization design method for combined impact of anti-detonation wave and fragment group of the present invention.

[0119] For the structural optimization design system for combined impact of anti-detonation wave and fragment group provided by the present invention, each functional component can be integrated in a processing component, or each component can exist physically alone, or two or more components can be integrated in one component. The above-mentioned integrated components can be implemented in the form of hardware or in the form of software functions, further improving the overall applicability and practical application ability of the present invention.

[0120] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered by the scope of the claims and the description of the present invention.

Claims

1. A structural optimization design method for combined impact of anti-detonation wave and fragment group, characterized in that The method includes: Obtaining a structural design scheme for the combined impact of anti-detonation waves and fragment groups, performing a first numerical analysis on the structural design scheme, and obtaining a sample set; Constructing a classification surrogate model and training the classification surrogate model using the sample set; Introducing a genetic algorithm and improving the fitness evaluation mechanism of the genetic algorithm using the classification surrogate model; Based on the improved fitness evaluation mechanism, selecting parents for the iteration of the genetic algorithm to obtain an improved genetic algorithm; Adopting a hybrid sampling strategy to optimize the selection of sample points of the classification surrogate model and the iteration process of the improved genetic algorithm, and determining the optimal structural design scheme for the combined impact of anti-detonation waves and fragment groups.

2. The structural optimization design method against combined shock of detonation wave and fragment group according to claim 1, characterized in that The improvement of the fitness evaluation mechanism of the genetic algorithm using the classification surrogate model includes: Using the classification surrogate model to divide the individuals in the population of the genetic algorithm into feasible solutions and infeasible solutions; Calculating the diversity difference between the infeasible solutions and the individuals in the population to obtain an infeasible solution diversity difference value; Improving the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value.

3. The structural optimization design method against combined impact of detonation wave and fragment group according to claim 2, characterized in that The improvement of the fitness evaluation mechanism of the genetic algorithm according to the infeasible solution diversity difference value includes: Calculating the objective function of the improved genetic algorithm according to the infeasible solution diversity difference value to obtain an objective function value; Sorting the individuals in the population according to the objective function value to obtain a sorting position; Calculating the fitness values of the individuals in the population according to the sorting position to obtain a population fitness value.

4. The structural optimization design method against combined impact of detonation wave and fragment group according to claim 3, characterized in that The selection of parents for the iteration of the genetic algorithm based on the improved fitness evaluation mechanism to obtain an improved genetic algorithm includes: During the iteration process of the genetic algorithm, calculating the probability of the individuals in the population being selected as parents using the roulette wheel selection method according to the population fitness value; Determining the parents in the iteration process of the genetic algorithm according to the probability to obtain the improved genetic algorithm.

5. The structural optimization design method for combined impact of anti-detonation wave and fragment group according to claim 1, characterized in that The adoption of a hybrid sampling strategy to optimize the selection of sample points of the classification surrogate model and the iteration process of the improved genetic algorithm, and determining the optimal structural design scheme for the combined impact of anti-detonation waves and fragment groups includes: Obtaining a decision boundary according to the classification surrogate model; Updating the sample set according to the decision boundary and the hybrid sampling strategy; Optimally training the classification surrogate model using the updated sample set; Performing an optimal solution using the optimally trained classification surrogate model and the improved genetic algorithm to determine the optimal structural design scheme for the combined impact of anti-detonation waves and fragment groups.

6. The structural optimization design method for combined impact of anti-detonation wave and fragment group according to claim 5, characterized in that The update of the sample set according to the decision boundary and the hybrid sampling strategy includes: Obtaining a structural optimization design set and a first structural optimization design according to the improved genetic algorithm; Obtaining one or more second structural optimization designs by calculating the distance from the structural optimization design set to the decision boundary; Performing a second numerical analysis on the first structural optimization design and the second structural optimization designs, and updating the sample set according to the results of the second numerical analysis.

7. The structural optimization design method for combined impact of anti-detonation wave and fragment group according to claim 5, characterized in that The optimization solution using the optimized and trained classification surrogate model and the improved genetic algorithm to determine the optimal structural design scheme against the combined impact of detonation waves and fragment groups includes: Using the optimized and trained classification surrogate model and the improved genetic algorithm to perform optimization solution to obtain the local structural optimization design; Obtaining the local structural optimal design according to the sample points of the classification surrogate model; Calculating the maximum value of the relative deviation between the local structural optimization design and the local structural optimal design to obtain the maximum relative deviation value; Setting a threshold, and judging whether the maximum relative deviation value is less than the threshold. If so, determining the local structural optimal design as the optimal structural design scheme against the combined impact of detonation waves and fragment groups.

8. The structural optimization design method against combined impact of detonation wave and fragment group according to claim 6, characterized in that The structural optimization design set includes multiple structural optimization designs. The obtaining of one or more second structural optimization designs by calculating the distance from the structural optimization design set to the decision boundary includes: Calculating the distance from each of the structural optimization designs in the structural optimization design set to the decision boundary; Selecting one or more of the structural optimization designs in the structural optimization design set according to the distance to form candidate sample points; Screening the candidate sample points by calculating the difference between the candidate sample points and the sample set to obtain one or more second structural optimization designs.

9. The structural optimization design method against combined impact of detonation wave and fragment group according to claim 3, characterized in that The objective function value satisfies the following formula: ， Among them, is the objective function processed based on the constraint handling strategy, is an individual in the population of the genetic algorithm, is the individual in the population of the structural quality, is a feasible solution, is the maximum structural quality of the individuals in the population, is the diversity difference value of the infeasible solutions.

10. Structural optimization design system for combined impact of anti-detonation wave and fragment group, characterized in that, Including: A processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the structural optimization design method against the combined impact of detonation waves and fragment groups according to any one of claims 1 to 9.

Images