Method for rapidly generating composite explosive formula aiming at multi-scene requirements

Abstract

The invention provides a composite explosive formula rapid generation method for multi-scene requirements, and relates to the technical field of explosive formula generation, and the method comprises the steps: mapping formula requirements into a target function and a constraint function; solving the target function and the constraint function through a non-dominated sorting genetic algorithm to obtain the proportion of the main components of the composite explosive; predicting the composite explosive functional aid ratio through the trained multiple machine learning models, the composite explosive main component ratio and the composite explosive performance index to obtain the predicted composite explosive functional aid ratio; and combining the proportion of the main components of the composite explosive with the predicted proportion of the functional additives of the composite explosive to obtain the composite explosive formula. According to the method, the function of generating different explosive formulas according to different scenes is realized, the formula generation efficiency is improved, and the time and economic cost is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of explosive formula generation, and in particular to a method for quickly generating a mixed explosive formula for multiple scenario requirements. Background Art

[0002] As a special material, energetic materials are widely used in defense, aerospace, blasting and other fields. Especially in the military field, energetic materials are the energy source for weapon system explosions to produce destruction and killing effects, the basis for weapons and equipment to achieve efficient damage, and the key factor in determining the power of weapon systems. However, the commonly used single-element explosives have a contradictory relationship between energy and safety and are not suitable for adapting to various complex application scenarios. Usually, researchers design orthogonal experiments and then conduct experiments to collect experimental data, and then adjust the explosive formula ratio by analyzing the experimental data.

[0003] However, traditional methods such as orthogonal experiments rely on researchers to continuously adjust components based on experience for trial and error, which makes the design inefficient, and existing software lacks an effective method for quickly designing formulations according to application requirements. Therefore, it is very necessary to design a method for quickly generating mixed explosive formulations for multiple scenarios. Summary of the invention

[0004] The purpose of the present invention is to provide a method for quickly generating a mixed explosive formula for multiple scene requirements, so as to realize the function of generating different explosive formulas according to different scenes, improve the formula generation efficiency and reduce time and economic costs.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for quickly generating a mixed explosive formula for multiple scenario requirements includes the following steps:

[0007] Mapping recipe requirements into objective functions and constraint functions;

[0008] The objective function and constraint function are solved by non-dominated sorting genetic algorithm to obtain the main component ratio of mixed explosives.

[0009] The functional additive ratio of the mixed explosive is predicted by using multiple trained machine learning models, the main component ratio of the mixed explosive and the performance index of the mixed explosive to obtain the predicted functional additive ratio of the mixed explosive;

[0010] The mixed explosive formula is obtained by combining the main component ratio of the mixed explosive with the predicted functional additive ratio of the mixed explosive.

[0011] Optionally, the formula requirement is expressed as:

[0012]

[0013] Among them, p n is the ratio of the nth main component of the mixed explosive; f m is the mth recipe performance evaluation function, t m is the mth performance index value, m is the optimized performance number required by the mixed explosive formula, is the minimum value that the i-th component can account for, is the maximum proportion that the i-th component can account for, f 1 :+ indicates that the performance optimization direction is maximization, f m :-Indicates that the optimization direction of the mth performance indicator is minimization.

[0014] Optionally, the objective function is expressed as:

[0015]

[0016] Among them, x n-1 is the ratio of the n-1th main component of the mixed explosive;

[0017] The expression of the constraint function is:

[0018]

[0019] in, is the minimum proportion of the i-th principal component of the mixed explosive, is the maximum proportion of the i-th principal component of the mixed explosive, x i is the proportion of the i-th principal component of the mixed explosive, It is the minimum proportion of the nth principal component of the mixed explosive.

[0020] Optionally, the specific steps of the non-dominated sorting genetic algorithm include:

[0021] Use Hua Luogeng's optimal point set method to initialize the population and obtain the initial population;

[0022] Calculate the objective function value and constraint function violation value of the initial population, and perform non-dominated sorting stratification on the initial population;

[0023] The initial population is selected, crossed and mutated through the tournament selection mechanism to generate the offspring population;

[0024] The initial population and the offspring population are merged to obtain a new population, and the new population is subjected to non-dominated sorting stratification and crowding calculation, and a new parent population is selected according to the crowding and non-dominated sorting stratification results;

[0025] The tournament selection mechanism and population merging are repeated until the number of iterations reaches a preset maximum.

[0026] Optionally, the calculation formula of Hua Luogeng's optimal point set method is:

[0027]

[0028] Among them, n is the dimension of the current space dimension, m is the population size, is the set of good points, r ji is the value of the jth dimension of the ith point, mod is the remainder function, a j is the lower limit of the dimension, b j is the upper limit of the dimension, The value of the jth dimension of the ith individual, Take the value of the j-th dimension of the i-th data point in the good point set.

[0029] Optionally, the constraint function violation value is calculated as:

[0030]

[0031] Among them, g j (x) is the inequality constraint, h k (x) is an equality constraint, <g j (x)> if g j (x)≤0, then g j (x)=0, otherwise g j (x) = |g j (x)|, CV(x) is the constraint function violation value, and x is the independent variable.

[0032] Optionally, the specific steps of non-dominated sorting include:

[0033] Store the best individuals in the population into a set, and use the set as the first non-dominated layer;

[0034] The number of dominated individuals dominated by the first non-dominated layer is reduced by 1, and the reduced number of dominated individuals is used as the second non-dominated layer;

[0035] Optionally, the calculation formula for congestion is:

[0036]

[0037] Where d is the congestion degree, is the maximum value of the i-th objective function among all solution sets, is the minimum value of the i-th objective function among all solution sets, X j_neighbor1 and X j_neighbor2 are the two individuals closest to the objective function value, n is the total number of performances, f i (X j_neighbor1 ) and f i (X j_neighbor2 ) are Xj_neighbor1 and X j_neighbor2 The corresponding objective function.

[0038] Optionally, the trained multiple machine learning models include: a linear regression model, a random forest regression model, a k-nearest neighbor regression model, and a multi-layer perceptron model;

[0039] The calculation formula of the linear regression model is: in, is the model output, is the eigenvector, x n is the nth mixed explosive formula feature, is the weight vector of the influence of different characteristics of mixed explosive formula in the prediction of functional additive ratio, w n is the influence weight of the nth mixed explosive formula feature in the prediction of functional additive ratio, and b is the bias term;

[0040] The calculation formula of the random forest regression model is: in, is the model output, X is the recipe data to be predicted, and f i (X) is the output of the i-th decision tree, and m is the number of decision trees;

[0041] The calculation formula of the k nearest neighbor regression model is: in, is the model output, y i is the label value of the i-th real recipe data, and k is the number of sample values ​​preset by the model;

[0042] The calculation formula of the multi-layer perceptron model is: in, is the weight parameter set of the jth neuron in the ith hidden layer, w in is the weight parameter of the nth neuron in the ith hidden layer, Q is the node output, is the output of the i-th hidden layer, b in is the bias term of the nth neuron in the ith hidden layer, X in is the output of the nth neuron in the ith hidden layer, and σ is the nonlinear transformation operation of the activation function.

[0043] A mixed explosive formula rapid generation system for multiple scenario requirements, including: an interaction module, a preprocessing module and a model module;

[0044] The interactive module is used to set recipe requirements;

[0045] The preprocessing module is used to map the formula requirements into the objective function and the constraint function and solve the objective function and the constraint function through the non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive;

[0046] The model module is used to predict the functional additive ratio of mixed explosives through multiple trained machine learning models, the main component ratio of mixed explosives and the performance indicators of mixed explosives, so as to obtain the predicted functional additive ratio of mixed explosives.

[0047] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: the method for quickly generating mixed explosive formulas for multi-scenario requirements provided by the present invention comprises: mapping formula requirements into objective functions and constraint functions; solving the objective functions and constraint functions by non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive; predicting the functional auxiliary agent ratio of the mixed explosive by trained multiple machine learning models, the main component ratio of the mixed explosive and the performance index of the mixed explosive to obtain the predicted functional auxiliary agent ratio of the mixed explosive; combining the main component ratio of the mixed explosive with the predicted functional auxiliary agent ratio of the mixed explosive to obtain the mixed explosive formula. The method realizes the function of generating different explosive formulas according to different scenarios, improves the formula generation efficiency and reduces time and economic costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 A flow chart of a method for quickly generating a mixed explosive formula for multiple scenario requirements according to an embodiment of the present invention;

[0050] Figure 2 is a preprocessing flow chart of an embodiment of the present invention;

[0051] Figure 3 A flow chart of a non-dominated sorting genetic algorithm according to an embodiment of the present invention;

[0052] Figure 4 4 is a structural diagram of a multilayer perceptron model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] like Figure 1 As shown, an embodiment of the present invention provides a method for quickly generating a mixed explosive formula for multiple scene requirements, comprising the following steps:

[0056] Step 100: Map the recipe requirements into objective functions and constraint functions; the specific steps are as follows Figure 2 As shown, including:

[0057] Step 101: unify the independent variables of the formula representing the performance index of the main explosive; the specific implementation process is: first, define the main component of the formula of the mixed explosive as (p 1 ,p 2 ,...,p n ), where p i represents the ratio of the i-th principal component, and p i The following conditions must be met: in, The minimum proportion of the i-th principal component, The maximum proportion of the i-th principal component, n is the number of principal components. Then, the principal component (p 1 ,p 2 ,...,p n ) is unified as the independent variable (x 1 ,x 2 ,...,x n-1 ).

[0058] Specifically, the formula requirements include the main explosive performance index, performance optimization direction, performance calculation formula and component variables. The formula requirements are expressed as:

[0059]

[0060]

[0061] Among them, p n is the ratio of the nth main component of the mixed explosive; f m is the mth recipe performance evaluation function, t mis the mth performance index value, m is the optimized performance number required by the mixed explosive formula, is the minimum value that the i-th component can account for, is the maximum proportion that the i-th component can account for, f 1 :+ indicates that the performance optimization direction is maximization, f m :-Indicates that the optimization direction of the mth performance indicator is minimization.

[0062] Step 102: Combining the performance calculation formula and the performance optimization direction, and unifying the optimization direction as minimization to construct an objective function;

[0063] Specifically, the expression of the objective function is:

[0064]

[0065] Among them, x n-1 is the ratio of the n-1th main component of the mixed explosive.

[0066] Step 103: preprocess the objective function to obtain a constraint function; the specific steps are: combine the objective function according to the performance index, and unify the direction of the inequality to construct the constraint function.

[0067] Specifically, the unified inequality direction is to split the minimization form objective function into forms less than or equal to 0.

[0068] More specifically, the constraint function is expressed as:

[0069]

[0070] in, is the minimum proportion of the i-th principal component of the mixed explosive, x i is the proportion of the i-th principal component of the mixed explosive, It is the minimum proportion of the nth principal component of the mixed explosive.

[0071] Step 200: solving the objective function and the constraint function by using a non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive;

[0072] Specifically, the specific steps of the non-dominated sorting genetic algorithm are as follows: Figure 3 As shown, including:

[0073] Step 201: Initialize the population using Hua Luogeng's optimal point set method to obtain an initial population;

[0074] Furthermore, when the population is initialized, a random initialization method is generally used. However, the randomly generated population is unevenly distributed in the entire solution space. Sometimes, it is too concentrated in some areas, and too discrete in some areas, which makes the utilization rate of the entire space low. In order to better find the global optimal solution set, this embodiment adopts the Hua Luogeng optimal point set method to sample from the n-dimensional space of the independent variable distribution to obtain an initial population with uniform distribution; the calculation formula of the Hua Luogeng optimal point set method is:

[0075]

[0076]

[0077] Among them, n is the dimension of the current space dimension, m is the population size, is the set of good points, r ji is the value of the jth dimension of the ith point, mod is the remainder function, a j is the lower limit of the dimension, b j is the upper limit of the dimension, The value of the jth dimension of the ith individual, Take the value of the j-th dimension of the i-th data point in the good point set.

[0078] Step 202: Calculate the objective function value and constraint function violation value of the initial population, and perform non-dominated sorting stratification on the initial population;

[0079] Specifically, the constraint function violation value is used to quantitatively describe the degree to which a solution violates the constraint condition. The calculation formula is:

[0080]

[0081] Among them, g j (x) is the assumed inequality constraint, h k (x) is the assumed equality constraint, <g j (x)> if g j (x)≤0, then g j (x)=0, otherwise g j (x) = |g j (x)|, CV(x) is the constraint function violation value, and x is the independent variable.

[0082] More specifically, the smaller the CV value, the less the solution violates the constraint, and the better the solution. At the same time, when a solution is a feasible solution, its CV value is 0, and when it is an infeasible solution, its CV value is greater than 0.

[0083] Furthermore, in order to take the degree of constraint violation into account for adaptability, a penalty term is added to the objective function of all solutions; the penalty term is calculated as follows:

[0084] p = ρ × CV (x);

[0085] Among them, p is the penalty term, and ρ is a value larger than the range of the objective function, that is, the penalty coefficient. The greater the degree to which the solution violates the constraint, the more the penalty term increases the objective function value of the solution, so that the solution is more likely to be at a disadvantage in subsequent non-dominated comparisons, making its fitness lower than other solutions and easily eliminated in iterations.

[0086] Furthermore, since equality constraints are more restrictive than inequality constraints, using a penalty coefficient that is too small will make the constraint worse, while using a penalty coefficient that is too large will cause slight constraint violations and make the penalty term much larger than the objective function value, so that the objective function degenerates into a penalty function, making the algorithm more inclined to find a solution that satisfies the equality constraints rather than a solution to the optimization problem during the iteration process.

[0087] This embodiment uses the method of removing equality processing to process equality constraints. Take the following as an example, integrate the equality constraint into the objective function and replace the equation involving x in the objective function with n Replace the part with At the same time, the equality constraint Convert to The processed solution can automatically satisfy the equality constraints.

[0088] Specifically, the specific implementation process of non-dominated sorting is as follows:

[0089] First, the non-dominated sorting algorithm assigns two parameters N to each population individual (solution): i and S i , N i represents the number of individuals (solutions) that dominate the ith individual (solution), S i represents the number of individuals (solutions) dominated by the i-th individual (solution);

[0090] Secondly, all the i = 0 are stored in the set as the first non-dominated layer, so that the individuals (solutions) in the set are the current optimal individual set (solution set); where N i =0 means that there is no other individual (solution) in the current population (solution set) that is better than this individual (solution);

[0091] Then, the S of the first non-dominated layer i Subtract 1 and satisfy S i -1=0 individuals are stored in the set as the second non-dominated layer, where S i -1 is S minus 1 i ;

[0092] Finally, the subsequent non-dominated layers are constructed in sequence according to the above non-dominated sorting algorithm, so that the solutions at the lower non-dominated layers are better than the solutions at the higher layers.

[0093] Step 203: The initial population is selected, crossed and mutated through the tournament selection mechanism to generate a descendant population; the specific steps are: individuals with better fitness are randomly selected as parents through the tournament selection operator to perform crossover and mutation to reproduce new individuals to form the descendant population.

[0094] Specifically, the fitness is represented by the fitness function, in which the fitness of the initial reproduction is represented by the level of the non-dominated layer where the individual is located, the smaller the better; the fitness of subsequent reproduction is represented by a partial order relationship; that is, the levels of the dominant layer are different, the smaller the level, the higher the fitness, and the levels are the same, the greater the crowding, the higher the fitness.

[0095] Step 204: The initial population and the offspring population are merged to obtain a new population, and the new population is subjected to non-dominated sorting stratification and crowding calculation, and a new parent population is selected according to the crowding and stratification results; the size of the initial population and the offspring population are both M;

[0096] Specifically, if the number of individuals in the current non-dominated layer is less than M, all individuals in this layer are retained in the new parent population. Otherwise, they are included in the new parent population in descending order according to the crowding degree until the size of the new parent population reaches M.

[0097] Specifically, the method for calculating the crowding degree is: first calculate the function values ​​of all objective functions, then find two individuals (solutions) with similar function values ​​and calculate the difference. The crowding degree of this individual (solution) is the sum of the differences of all adjacent individuals (solutions) of the objective function. Among them, the smaller the crowding degree, the more similar individuals there are near the individual; the higher the crowding degree, in general, individuals with high crowding degrees tend to be retained because their surroundings are more spacious and more searchable. The specific calculation formula is:

[0098]

[0099] Where d is the congestion degree, is the maximum value of the i-th objective function among all solution sets, is the minimum value of the i-th objective function among all solution sets, m is the population size, X j_neighbor1 and X j_neighbor2are the objective functions corresponding to neighbor1 and neighbor2 respectively. neighbor1 and neighbor2 are the two individuals (solutions) closest to the objective function value of the individual (solution). The former is the individual (solution) closest to the individual (solution) among all the individuals whose objective function value is greater than the individual, and the latter is the individual (solution) closest to the individual (solution) among all the individuals whose objective function value is less than the individual. For the boundary points, the congestion is set to infinity. The denominator of the calculation formula is used for normalization. Because different objective functions have different value ranges, there may be large differences. If normalization and scaling are not performed, the objective function with a larger value will dominate the congestion value.

[0100] More specifically, this embodiment provides two new congestion calculation formulas for formula design in certain scenarios:

[0101]

[0102] Furthermore, when there are specific expected values ​​for each performance value of the recipe and it is desired to improve the fitness of the solution similar to the expected vector, the new crowding formula (1) is adopted. max is the maximum value of the crowding degree of all individuals (solutions) in the level where the individual (solution) is located, d is the value calculated according to the original crowding degree, and sim is the expected approximation of the solution set (the value ranges from 0 to 1).

[0103] Furthermore, the new congestion formula (1) is divided into distance approximation and cosine approximation; assuming is the expected vector, is the objective function value vector of the individual (solution).

[0104] The distance approximation formula is:

[0105]

[0106]

[0107] Among them, distance_max represents the objective function value vector of all individuals in the non-dominated layer where the individual is located With the expected vector The maximum distance, Represents the individual objective function value vector With the expected vector The distance, e 1 Indicates the expected performance value of a formula, v 1represents the objective function value of the performance obtained through individual solution, and pre represents the approximation between the objective function value vector and the expected vector, that is, the approximation between the performance value obtained through solution and the expected performance value. The reference standard is the solution vector that is farthest from the expected vector among all solutions. The closer it is to 0, that is, the closer the individual (solution) objective function value vector is to the expected vector, the greater the fitness improvement.

[0108] The calculation formula for the cosine approximation is:

[0109]

[0110] The larger the cosine approximation value, the greater the fitness improvement.

[0111] Furthermore, when one wants to improve the priority of a certain performance of the recipe in the search process, the new crowding formula (2) is used. Where f(X) is the objective function value of the individual in priority.

[0112] Step 205: Repeat steps 203 and 204 until the number of iterations reaches a preset maximum value.

[0113] Step 300: predicting the functional additive ratio of the mixed explosive by using the trained multiple machine learning models, the main component ratio of the mixed explosive and the performance index of the mixed explosive to obtain the predicted functional additive ratio of the mixed explosive;

[0114] Specifically, the trained machine learning models include: linear regression model, random forest regression model, k-nearest neighbor regression model and multi-layer perceptron model;

[0115] The linear regression model takes into account the impact of multiple input variables on the output variable, and its calculation formula is:

[0116]

[0117] in, is the model output, is the eigenvector, x n is the nth mixed explosive formula feature, is the weight vector of the influence of different characteristics of mixed explosive formula in the prediction of functional additive ratio, w n is the influence weight of the nth mixed explosive formula feature in the prediction of functional additive ratio, and b is the bias term; and b are determined by minimizing the loss function, and the loss function calculation formula is:

[0118]

[0119] Among them, y i Indicates the label value of the actual formula data (such as sensitivity, energy), is the model output for the ith input variable.

[0120] The random forest regression model consists of multiple decision trees, and the real recipe data is used to generate multiple different subsets through self-sampling to train multiple decision trees. The final model output uses the average value of all decision tree outputs, and its calculation formula is:

[0121]

[0122] in, is the model output, X is the recipe data to be predicted, and f i (X) is the output of the i-th decision tree, m is the number of decision trees; the loss function calculation formula of the random forest regression model is:

[0123]

[0124] R 1 (k,v)=X|X k ≤v,R 2 (k,v)=X|X k >v;

[0125]

[0126] Among them, R 1 and R 2 Indicates that the value v of feature k is divided into two intervals, X i is the data in this interval, c is the average value of the real formula data label value of all data in this interval, N is the number of data points in the divided interval, c1 is R 1 The c value of the interval, c2 is R 2 The c value of the interval.

[0127] The calculation formula of the k nearest neighbor regression model is:

[0128]

[0129] in, is the model output, y i is the label value of the i-th real recipe data, k is the number of sample values ​​preset by the model, and the k value is determined by grid optimization;

[0130] like Figure 4 As shown in the figure, the multi-layer perceptron model is a neural network structure with multiple levels, including an input layer, at least one hidden layer and an output layer; the input layer takes the main explosive ratio, binder ratio and target performance data as input, and outputs the functional additive ratio through the hidden layer. Its calculation formula is:

[0131]

[0132]

[0133] in, is the weight parameter set of the jth neuron in the ith hidden layer, w in is the weight parameter of the nth neuron in the ith hidden layer, Q is the node output, is the output of the i-th hidden layer, b in is the bias term of the nth neuron in the ith hidden layer, X in is the output of the nth neuron in the ith hidden layer, and σ is the nonlinear transformation operation of the activation function.

[0134] Step 400: combining the main component ratio of the mixed explosive with the predicted functional additive ratio of the mixed explosive to obtain a mixed explosive formula.

[0135] The embodiment of the present invention also provides a system for quickly generating mixed explosive formulas for multiple scene requirements, including: an interaction module, a preprocessing module and a model module;

[0136] The interactive module is used by users to configure the requirements of the formula to be designed, including the determination of components, the determination of the complete mixed explosive formula index and the main explosive index, the determination of the component proportion range, the selection of the performance optimization direction and the selection of the performance calculation formula;

[0137] The preprocessing module is used to map the formula requirements into the objective function and the constraint function and solve the objective function and the constraint function through the non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive;

[0138] The model module is used to predict the functional additive ratio of mixed explosives through multiple trained machine learning models, the main component ratio of mixed explosives and the performance indicators of mixed explosives, so as to obtain the predicted functional additive ratio of mixed explosives.

[0139] The beneficial effects of the present invention are as follows:

[0140] 1) The function of generating different explosive formulas according to different scenarios;

[0141] 2) Improve the efficiency of formula generation and reduce time and economic costs.

[0142] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0143] The present invention uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for quickly generating a mixed explosive formula for multiple scene requirements, characterized in that: The steps include: Mapping recipe requirements into objective functions and constraint functions; Solving the objective function and the constraint function by a non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive; Predicting the functional additive ratio of the mixed explosive by using the trained multiple machine learning models, the main component ratio of the mixed explosive and the performance index of the mixed explosive to obtain the predicted functional additive ratio of the mixed explosive; The main component ratio of the mixed explosive and the predicted functional additive ratio of the mixed explosive are combined to obtain a mixed explosive formula.

2. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 1, characterized in that: The formula requirement is expressed as: f1:+,.......,f m :-; Among them, p n is the ratio of the nth main component of the mixed explosive; f m is the mth recipe performance evaluation function, t m is the mth performance index value, m is the optimized performance number required by the mixed explosive formula, is the minimum value that the i-th component can account for, is the maximum value that the i-th component can account for, f1:+ indicates that the performance optimization direction is maximization, f m :-Indicates that the optimization direction of the mth performance indicator is minimization.

3. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 2, characterized in that: The expression of the objective function is: Among them, x n-1 is the ratio of the n-1th main component of the mixed explosive; The expression of the constraint function is: in, is the minimum proportion of the i-th principal component of the mixed explosive, is the maximum proportion of the i-th principal component of the mixed explosive, x i is the proportion of the i-th principal component of the mixed explosive, It is the minimum proportion of the nth principal component of the mixed explosive.

4. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 1, characterized in that: The specific steps of the non-dominated sorting genetic algorithm include: Use Hua Luogeng's optimal point set method to initialize the population and obtain the initial population; Calculating the objective function value and constraint function violation value of the initial population, and performing non-dominated sorting stratification on the initial population; Selecting, crossing over and mutating the initial population through a tournament selection mechanism to generate a descendant population; Merging the initial population and the offspring population to obtain a new population, performing the non-dominated sorting stratification and crowding calculation on the new population, and selecting a new parent population according to the crowding and non-dominated sorting stratification results; The tournament selection mechanism and the population merging are repeated until the number of iterations reaches a preset maximum value.

5. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 4, characterized in that: The calculation formula of Hua Luogeng's optimal point set method is: Among them, n is the dimension of the current space dimension, m is the population size, is the set of good points, r ji is the value of the jth dimension of the ith point, mod is the remainder function, a j is the lower limit of the dimension, b j is the upper limit of the dimension, The value of the jth dimension of the ith individual, Take the value of the j-th dimension of the i-th data point in the good point set.

6. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 4, characterized in that: The calculation formula of the constraint function violation value is: Among them, g j (x) is the inequality constraint, h k (x) is an equality constraint, <g j (x)> if g j (x)≤0, then g j (x)=0, otherwise g j (x) = |g j (x)|, CV(x) is the constraint function violation value, and x is the independent variable.

7. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 4, characterized in that: The specific steps of the non-dominated sorting include: The best individuals in the population are stored in a set, and the set is used as the first non-dominated layer; The number of dominated individuals dominated by the first non-dominated layer is reduced by 1, and the reduced number of dominated individuals is used as the second non-dominated layer.

8. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 4, characterized in that: The calculation formula of the congestion degree is: Where d is the congestion degree, is the maximum value of the i-th objective function among all solution sets, is the minimum value of the i-th objective function among all solution sets, X j_neighbor1 and X j_neighbor2 are the two individuals closest to the objective function value, n is the total number of performances, f i (X j_neighbor1 ) and f i (X j_neighbor2 ) are X j_neighbor1 and X j_neighbor2 The corresponding objective function value.

9. The method for rapidly generating a mixed explosive formula for multiple scene requirements according to claim 1, characterized in that: The trained multiple machine learning models include: a linear regression model, a random forest regression model, a k-nearest neighbor regression model and a multi-layer perceptron model; The calculation formula of the linear regression model is: in, is the model output, is the eigenvector, x n is the nth mixed explosive formula feature, is the weight vector of the influence of different characteristics of mixed explosive formula in the prediction of functional additive ratio, w n is the influence weight of the nth mixed explosive formula feature in the prediction of functional additive ratio, and b is the bias term; The calculation formula of the random forest regression model is: in, is the model output, X is the recipe data to be predicted, and f i (X) is the output of the i-th decision tree, and m is the number of decision trees; The calculation formula of the k nearest neighbor regression model is: in, is the model output, y i is the label value of the i-th real recipe data, and k is the number of sample values ​​preset by the model; The calculation formula of the multi-layer perceptron model is: in, is the weight parameter set of the jth neuron in the ith hidden layer, w in is the weight parameter of the nth neuron in the ith hidden layer, Q is the node output, is the output of the i-th hidden layer, b in is the bias term of the nth neuron in the ith hidden layer, X in is the output of the nth neuron in the ith hidden layer, and σ is the nonlinear transformation operation of the activation function.

10. A system for rapidly generating mixed explosive formulas for multiple scene requirements, applied to the method for rapidly generating mixed explosive formulas for multiple scene requirements as claimed in any one of claims 1 to 9, characterized in that: include: Interaction module, preprocessing module and model module; The interaction module is used to set the recipe requirement; The preprocessing module is used to map the formula requirements into an objective function and a constraint function, and solve the objective function and the constraint function through a non-dominated sorting genetic algorithm to obtain the main component ratio of the mixed explosive; The model module is used to predict the functional additive ratio of the mixed explosive through the trained multiple machine learning models, the main component ratio of the mixed explosive and the performance index of the mixed explosive, so as to obtain the predicted functional additive ratio of the mixed explosive.

Images