Explosion shock wave leading edge distribution prediction method based on particle swarm optimization

Abstract

The invention discloses an explosive shock wave leading edge distribution prediction method based on particle swarm optimization. The method comprises the following steps: S1, forming a preprocessed initial explosive shock wave data set; s2, establishing an explosion shock wave front nonlinear prediction model based on the initial explosion shock wave data set; s3, performing global search on the explosion shock wave front nonlinear prediction model by using an improved bitter fish optimization algorithm, and determining an optimal solution region; s4, in the optimal solution region, performing fine tuning on the explosive shock wave front nonlinear prediction model parameters by adopting a hierarchical particle swarm optimization algorithm to obtain an updated explosive shock wave front nonlinear prediction model; and S5, carrying out leading edge evolution prediction on the initial explosion shock wave data set formed in the step S1 by using the updated explosion shock wave leading edge nonlinear prediction model, and outputting a final prediction result. According to the method, the model parameters are recursively updated, so that the prediction process can adjust the response to the environment variables in real time, and the identification capability of the shock wave attenuation trend is improved.

Description

Technical Field

[0001] The present invention relates to the field of blasting technology, and in particular to a method for predicting explosion shock wave front distribution based on particle swarm optimization. Background Art

[0002] The propagation characteristics of explosion shock waves have important research value in the fields of national defense security, industrial blasting, and disaster prevention. The propagation trajectory, peak pressure, attenuation rate, and impact range of the shock wave directly affect the degree of damage to the environment and targets. Therefore, accurately predicting the evolution of the explosion shock wave front is of great significance for optimizing blasting design, improving protective measures, and improving explosion simulation technology.

[0003] At present, the prediction of the front distribution of explosion shock waves mainly relies on three methods: theoretical models, experimental measurements and numerical simulations. Among them, theoretical models are usually based on classical gas dynamics equations or semi-empirical formulas, and use simplified assumptions to quantitatively describe the propagation of shock waves. However, there are great limitations in practical applications: on the one hand, theoretical models rely on a large number of idealized assumptions, such as homogeneous media, no energy loss, and fixed boundary conditions, which make the models difficult to adapt to the influence of complex environmental variables; on the other hand, traditional theoretical models find it difficult to accurately describe the nonlinear evolution characteristics of explosion shock waves in different media, especially for strong nonlinear propagation, environmental variable coupling and high-dimensional shock wave characteristics.

[0004] Experimental measurement is another important method for shock wave front analysis. High-speed photography, laser interferometers, or pressure sensors are usually used to capture the shock wave propagation process. However, there are two major problems with this method. On the one hand, experimental measurement requires highly precise equipment, which is costly. In addition, it is limited by the resolution and response time of the measuring equipment, making it difficult to obtain shock wave propagation data in the entire time and space range. On the other hand, experimental measurements can usually only be carried out in specific environments and under fixed boundary conditions, making it difficult to achieve universal predictions under multiple environmental variables. In addition, experimental errors and nonlinear sensor response factors can also lead to uncertainty in the measurement results, further affecting the prediction accuracy of the shock wave front distribution.

[0005] In recent years, numerical simulation methods have gradually become an important tool for studying the propagation characteristics of explosion shock waves. Computational methods based on fluid dynamics, such as solving the Euler equations, finite element analysis, finite volume method and computational fluid dynamics simulation, can perform high-precision numerical calculations of shock waves under different environmental conditions. However, numerical simulation methods still have some key defects: First, existing numerical simulation calculations usually rely on high computing resources, have high computing costs, and are difficult to achieve real-time predictions; second, numerical simulations have limited processing capabilities for complex media and multi-physical field coupling effects. In the case of inhomogeneous media and variable boundary conditions, the stability and reliability of the calculation results are difficult to guarantee; in addition, traditional numerical simulation methods usually require manual parameter setting and lack an adaptive optimization mechanism, which leads to a high degree of subjectivity in parameter selection of the model, affecting the prediction accuracy and generalization ability.

[0006] In summary, the existing explosion shock wave front distribution prediction technology has significant limitations in accuracy, computational efficiency and environmental adaptability, and cannot fully meet the needs of engineering applications. Therefore, there is an urgent need for an efficient prediction method that can combine experimental data, numerical simulation and intelligent optimization algorithm to improve the accuracy of shock wave front distribution prediction and enhance the adaptability of the model in complex environments. Summary of the Invention

[0007] One purpose of the present invention is to propose a method for predicting the front distribution of explosion shock waves based on particle swarm optimization. The present invention recursively updates model parameters so that the prediction process can adjust its response to environmental variables in real time, thereby improving the ability to identify the shock wave attenuation trend.

[0008] A method for predicting the front distribution of an explosion shock wave based on particle swarm optimization according to an embodiment of the present invention includes the following steps:

[0009] S1. Collecting explosion shock wave experimental data and explosion shock wave simulation data, preprocessing the explosion shock wave experimental data to form an initial explosion shock wave data set after preprocessing;

[0010] S2. Establish a nonlinear prediction model for the explosion shock wave front based on the initial explosion shock wave dataset;

[0011] S3. Using an improved bitterfish optimization algorithm, a global search is performed on the nonlinear prediction model of the explosion shock wave front to determine the optimal solution area;

[0012] S4. Within the optimal solution region, fine-tune the parameters of the nonlinear prediction model for the blast shock wave front using a hierarchical particle swarm optimization algorithm to obtain an updated nonlinear prediction model for the blast shock wave front;

[0013] S5. Use the updated nonlinear prediction model of the explosion shock wave front to predict the front evolution of the initial explosion shock wave data set formed in step S1, and output the final prediction results, which include the explosion shock wave front propagation trajectory, peak pressure, attenuation rate and impact range.

[0014] Optionally, the S1 includes the following steps:

[0015] S11. Collecting explosion shock wave experimental data and explosion shock wave simulation data. The explosion shock wave experimental data is obtained by experimental measurement equipment, including explosion energy, medium density, shock wave propagation velocity and front morphological characteristics. The explosion shock wave simulation data is obtained by numerical simulation calculation, including shock wave isentropic evolution characteristics, pressure attenuation characteristics and shock wave propagation parameters under the influence of different environmental variables;

[0016] S12. Using a data format standardization method to uniformly format the explosion shock wave experimental data, the experimental data, simulation data and real-time monitoring data are converted into an explosion shock wave experimental data matrix;

[0017] S13. Performing data denoising on the explosion shock wave experimental data matrix to remove noise components in the signal and performing normalization processing to map all characteristic variables to the interval [0, 1] to obtain a normalized explosion shock wave experimental data matrix;

[0018] S14. Perform outlier detection on the normalized explosion shock wave experimental data matrix, identify abnormal data points, and use the nearest neighbor interpolation method to complete the explosion shock wave experimental data matrix to obtain the interpolated explosion shock wave experimental data matrix. The completed explosion shock wave experimental data matrix D final As the initial explosion shock wave dataset.

[0019] Optionally, S2 includes the following steps:

[0020] S21. Based on the initial explosion shock wave data set D final , a nonlinear prediction model for the explosion shock wave front is constructed. The nonlinear prediction model for the explosion shock wave front is based on the physical characteristics of the explosion shock wave, the multi-physical field coupling effect, the time-varying characteristics and the influence of environmental variables, and establishes a mapping relationship between the input parameter set X and the explosion shock wave front evolution result set Y:

[0021] Y=f(X,Θ);

[0022] Where X={x1,x2,...,x N} represents the input parameter set, x i is the input characteristic variable of the explosion shock wave, including the explosion energy E b, medium density ρ, shock wave propagation velocity v w and frontier morphological characteristics S f ,Y={y1,y2,...,y M} represents the spatiotemporal evolution of the explosion shock wave front, including the propagation trajectory, peak pressure, attenuation rate, and impact range of the explosion shock wave front. Θ is the parameter set of the nonlinear prediction model of the explosion shock wave front.

[0023] S22. Based on the initial explosion shock wave data set D final , the least squares regression method is used to make preliminary parameter estimation of the nonlinear prediction model of the explosion shock wave front, the initial model parameters Θ0 are set, and the initial estimated values ​​of the initial model parameters are calculated:

[0024]

[0025] Among them, Θ0 is the initial model parameter, which makes the nonlinear prediction model of the explosion shock wave front initially fit the evolution trend of the explosion shock wave front;

[0026] S23. Modify the nonlinear prediction model of the explosion shock wave front and construct the environmental variable matrix Z to describe the impact of the environment on the evolution of the explosion shock wave front:

[0027] Z={z1,z2,...,z K}

[0028] Among them, Z represents the environmental variable matrix, z k Represents environmental variables, including temperature, humidity, atmospheric pressure and medium characteristics, and combines the environmental variable matrix Z to adjust the weight parameters of the nonlinear prediction model of the explosion shock wave front;

[0029] S24. Apply time series constraints to the nonlinear prediction model of the explosion shock wave front and set the time series state vector:

[0030] S t =g(S t-1 ,X t ,Θ t );

[0031] Among them, S t It represents the state of the explosion shock wave front at time t, S t-1 is the state at the previous moment, X t is the current input parameter, Θ t is the current model parameter;

[0032] S25. Obtain the final nonlinear prediction model of the explosion shock wave front:

[0033] Y t =f(Xt ,Z t ,S t-1 ,Θ);

[0034] Among them, S t-1 Indicates the state of the explosion shock wave front at time t-1.

[0035] Optionally, S3 includes the following steps:

[0036] S31. Define the current model parameter set Θ = {θ1, θ2, ..., θ P}, where each model parameter θ p All meet the following requirements:

[0037]

[0038] in, and Represent the minimum and maximum physical constraint values ​​of the pth parameter respectively;

[0039] S32. An adaptive feedback factor λ(t) is introduced to reflect the relative value of the prediction error of the explosion shock wave front and its time-varying gradient, so that the improved bitter fish optimization algorithm can dynamically respond to the nonlinear changes of the wave front during the explosion process:

[0040]

[0041] Among them, Y t represents the actual explosion shock wave front state at time t, κ is the error scaling constant, and η is the error change rate weight;

[0042] S33. Using multidimensional chaotic mapping to generate dynamic perturbation factor χ(t) and diffusion factor ψ(t), we improve the global search capability of the bitter fish optimization algorithm for the complex and nonlinear behavior of explosion shock waves. The chaotic perturbation factor adopts logistic mapping:

[0043] χ(t+1)=μ·χ(t)·(1-χ(t));

[0044] The diffusion factor ψ(t) adopts the improved Tent mapping:

[0045]

[0046] Among them, μ is the chaotic mapping parameter, δ is the boundary parameter of the Tent mapping, χ(t) and ψ(t) are used to simulate the uncertain perturbation and diffusion trend of the explosion shock wave front in different environments, respectively, reflecting the physical phenomenon of random diffusion of energy in the explosion shock wave;

[0047] S34. Initialize the population of the improved bitter fish optimization algorithm and set the number of individuals Ns and each individual parameter vector Make each entity randomly distributed in the parameter search space to cover various physical states existing at the front of the explosion shock wave;

[0048] S35. Use the fusion differential evolution mechanism to update the position of each bitterling individual:

[0049]

[0050] in, represents the global optimal parameter vector with the best performance in the tth generation, α is the global exploration weight, β is the differential evolution weight, and are two sets of randomly selected individual parameters to provide differential information, γ is the reward factor weight, is the individual reward function, which is used to reward individuals with low errors in predicting the state of the explosion shock wave front, so that the update process optimizes the capture of changes in explosion energy transfer and wavefront attenuation:

[0051]

[0052] Among them, the fitness function for:

[0053]

[0054] Where C is the positive regulation constant;

[0055] S36. The global optimal parameter vector is updated through an information sharing mechanism. The information sharing mechanism integrates the contributions of each individual to the dynamic evolution of the explosion shock wave front during the global search process:

[0056]

[0057] Among them, the weight factor Defined as:

[0058]

[0059] S37. Set the stopping condition, when the preset maximum number of iterations T is reached max Or the fitness function changes to satisfy:

[0060]

[0061] Among them, ∈ is the convergence threshold, terminating the optimization iteration;

[0062] S38. Output the optimal parameter vector Determine the optimal solution region of the nonlinear prediction model of the explosion shock wave front.

[0063] Optionally, the S4 specifically includes: the S4 includes the following steps:

[0064] S41. In the global optimal parameter vector Θ * On this basis, the search range Θ of the local optimization phase of the hierarchical particle swarm optimization algorithm is set local ,in It is used to describe the parameter fine-tuning range of the nonlinear prediction model of the explosion shock wave front in the local area, and capture the local energy transfer, subtle waveform changes and attenuation characteristics of the explosion shock wave front;

[0065] S42. Initialize the hierarchical particle swarm optimization algorithm population and set the number of particle swarm individuals N p And the parameter vector of each particle Make each particle in Θ local Random distribution within

[0066] S43. A multi-level search structure is used to divide the particle swarm into L levels, each level corresponding to a different search granularity, to optimize the global local features and subtle local features within the local area of ​​the explosion shock wave front respectively;

[0067] S44. Update the particle positions within each level l, using adaptive inertia weight adjustment and global-local search fusion strategy:

[0068]

[0069] in, represents the parameter vector of the jth particle in the lth layer of the tth generation, represents the local optimal parameter vector in the lth layer, Represents the overall optimal parameter vector in the current local search, w l (t) is the adaptive inertia weight of the lth layer, which is used to balance the search speed and accuracy, c l is the learning factor of the lth layer, r l is a random variable that follows a uniform distribution [0,1];

[0070] S45. Introduce an adaptive neighborhood search mechanism to perform neighborhood perturbations on each particle within the current level and update the neighborhood parameter vector as follows:

[0071]

[0072] Among them, δ l represents the neighborhood search step size of the lth layer, ξ is a random perturbation taken from the standard normal distribution, which is used to capture the small fluctuations of the explosion shock wave front in the local area and refine the description of the local nonlinear evolution law of the explosion shock wave front nonlinear prediction model;

[0073] S46. Repeat steps S44 to S45 until the local optimization convergence condition is met, that is, when the local optimal parameter vector update in any level satisfies:

[0074]

[0075] Among them, ∈ local is the local search convergence threshold, the iteration is terminated, and the updated explosion shock wave front nonlinear prediction model parameter vector Θ is output updated The parameter vector of the nonlinear prediction model of the explosion shock wave front reflects the optimization result after fine-tuning the energy transfer, waveform attenuation and propagation characteristics of the explosion shock wave front in the local area.

[0076] Optionally, the step S4 further includes implementing adaptive inertia weight adjustment on each level l, setting its weight w l The update of (t) is:

[0077]

[0078] Among them, w max,l With w min,l Represent the initial and final inertia weights of the lth layer, T local is the maximum number of iterations in the local optimization phase.

[0079] The beneficial effects of the present invention are:

[0080] (1) In the process of optimizing the parameters of the nonlinear prediction model of the shock wave front, the present invention adopts an improved bitter fish optimization algorithm, and introduces an adaptive feedback factor and a multidimensional chaotic perturbation mechanism to enable the optimization algorithm to dynamically adjust the search step size and direction, thereby enhancing the global exploration capability of the complex nonlinear shock wave evolution process. By introducing the adaptive feedback factor, the optimization process can respond to the changing trend of the shock wave front in real time, and the global search perturbation is introduced in combination with the Logistic chaotic map and the Tent map, thereby enhancing the adaptability to the shock wave propagation parameters.

[0081] (2) The present invention proposes a hierarchical particle swarm optimization method, which divides the particle swarm into different levels by setting a multi-layer search structure in the global optimal solution area, and adopts adaptive inertia weight and neighborhood search mechanism in each level to achieve fine optimization of the local area. In the local optimization process, random perturbation parameters are introduced to enable the particle swarm to flexibly adjust the search accuracy and direction between different levels, thereby improving the fitting accuracy of the shock wave front parameters.

[0082] (3) In the nonlinear prediction model of the shock wave front, the present invention combines the time series constraint mechanism and sets the time series state vector, so that the model can be dynamically updated using historical state information, and construct a shock wave propagation prediction model that is more in line with physical laws. By recursively updating the model parameters, the prediction process can adjust the response to environmental variables in real time, thereby improving the ability to identify the shock wave attenuation trend. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0084] Figure 1 This is a flow chart of a method for predicting explosion shock wave front distribution based on particle swarm optimization proposed by the present invention. DETAILED DESCRIPTION

[0085] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0086] refer to Figure 1 , a method for predicting the front distribution of explosion shock wave based on particle swarm optimization, comprising the following steps:

[0087] S1. Collecting explosion shock wave experimental data and explosion shock wave simulation data, preprocessing the explosion shock wave experimental data to form an initial explosion shock wave data set after preprocessing;

[0088] S2. Establish a nonlinear prediction model for the explosion shock wave front based on the initial explosion shock wave dataset;

[0089] S3. Use the improved bitterfish optimization algorithm to perform a global search on the nonlinear prediction model of the explosion shock wave front to determine the optimal solution region;

[0090] S4. Within the optimal solution region, fine-tune the parameters of the nonlinear prediction model for the blast shock wave front using a hierarchical particle swarm optimization algorithm to obtain an updated nonlinear prediction model for the blast shock wave front;

[0091] S5. Use the updated nonlinear prediction model of the explosion shock wave front to predict the front evolution of the initial explosion shock wave data set formed in step S1, and output the final prediction results, which include the explosion shock wave front propagation trajectory, peak pressure, attenuation rate and impact range.

[0092] In this embodiment, S1 includes the following steps:

[0093] S11. Collect experimental data and simulation data of explosion shock waves. The experimental data are obtained by experimental measurement equipment, including explosion energy, medium density, shock wave propagation velocity, and front morphological characteristics. The simulation data are obtained by numerical simulation calculations, including shock wave isentropic evolution characteristics, pressure attenuation characteristics, and shock wave propagation parameters under the influence of different environmental variables.

[0094] S12. Use a data format standardization method to uniformly format the explosion shock wave experimental data, and convert the experimental data, simulation data, and real-time monitoring data into an explosion shock wave experimental data matrix;

[0095] S13. Performing data denoising on the explosion shock wave experimental data matrix to remove noise components in the signal and performing normalization processing to map all characteristic variables to the interval [0, 1] to obtain a normalized explosion shock wave experimental data matrix;

[0096] S14. Perform outlier detection on the normalized explosion shock wave experimental data matrix, identify abnormal data points, and use the nearest neighbor interpolation method to complete the explosion shock wave experimental data matrix to obtain the interpolated explosion shock wave experimental data matrix. The completed explosion shock wave experimental data matrix D final As the initial explosion shock wave dataset.

[0097] In this embodiment, S2 includes the following steps:

[0098] S21. Based on the initial explosion shock wave data set D final , a nonlinear prediction model for the explosion shock wave front is constructed. The nonlinear prediction model for the explosion shock wave front is based on the physical characteristics of the explosion shock wave, the multi-physical field coupling effect, the time-varying characteristics and the influence of environmental variables. The mapping relationship between the input parameter set X and the explosion shock wave front evolution result set Y is established:

[0099] Y=f(X,Θ);

[0100] Where X={x1,x2,...,x N} represents the input parameter set, x i is the input characteristic variable of the explosion shock wave, including the explosion energy E b , medium density ρ, shock wave propagation velocity v w and frontier morphological characteristics S f ,Y={y1,y2,...,y M} represents the spatiotemporal evolution of the explosion shock wave front, including the propagation trajectory, peak pressure, attenuation rate, and impact range of the explosion shock wave front. Θ is the parameter set of the nonlinear prediction model of the explosion shock wave front.

[0101] S22. Based on the initial explosion shock wave data set D final , the least squares regression method is used to make preliminary parameter estimation of the nonlinear prediction model of the explosion shock wave front, the initial model parameters Θ0 are set, and the initial estimated values ​​of the initial model parameters are calculated:

[0102]

[0103] Among them, Θ0 is the initial model parameter, which makes the nonlinear prediction model of the explosion shock wave front initially fit the evolution trend of the explosion shock wave front;

[0104] S23. Modify the nonlinear prediction model of the explosion shock wave front and construct the environmental variable matrix Z to describe the impact of the environment on the evolution of the explosion shock wave front:

[0105] Z={z1,z2,...,z K}

[0106] Among them, Z represents the environmental variable matrix, z k Represents environmental variables, including temperature, humidity, atmospheric pressure and medium characteristics, and combines the environmental variable matrix Z to adjust the weight parameters of the nonlinear prediction model of the explosion shock wave front;

[0107] S24. Apply time series constraints to the nonlinear prediction model of the explosion shock wave front and set the time series state vector:

[0108] S t =g(S t-1 ,X t ,Θ t );

[0109] Among them, S t It represents the state of the explosion shock wave front at time t, S t-1 is the state at the previous moment, X t is the current input parameter, Θ t is the current model parameter;

[0110] S25. Obtain the final nonlinear prediction model of the explosion shock wave front:

[0111] Y t =f(X t ,Z t ,S t-1 ,Θ);

[0112] Among them, S t-1 Indicates the state of the explosion shock wave front at time t-1.

[0113] In this embodiment, S3 includes the following steps:

[0114] S31. Define the current model parameter set Θ = {θ1, θ2, ..., θ P}, where each model parameter θ p All meet the following requirements:

[0115]

[0116] in, and Represent the minimum and maximum physical constraint values ​​of the pth parameter respectively;

[0117] S32. An adaptive feedback factor λ(t) is introduced to reflect the relative value of the prediction error of the explosion shock wave front and its time-varying gradient, so that the improved bitter fish optimization algorithm can dynamically respond to the nonlinear changes of the wave front during the explosion process:

[0118]

[0119] Among them, Y t represents the actual explosion shock wave front state at time t, κ is the error scaling constant, and η is the error change rate weight;

[0120] S33. Using multidimensional chaotic mapping to generate dynamic perturbation factor χ(t) and diffusion factor ψ(t), we improve the global search capability of the bitter fish optimization algorithm for the complex and nonlinear behavior of explosion shock waves. The chaotic perturbation factor adopts logistic mapping:

[0121] χ(t+1)=μ·χ(t)·(1-χ(t));

[0122] The diffusion factor ψ(t) adopts the improved Tent mapping:

[0123]

[0124] Among them, μ is the chaotic mapping parameter, δ is the boundary parameter of the Tent mapping, χ(t) and ψ(t) are used to simulate the uncertain perturbation and diffusion trend of the explosion shock wave front in different environments, respectively, reflecting the physical phenomenon of random diffusion of energy in the explosion shock wave;

[0125] S34. Initialize the population of the improved bitter fish optimization algorithm and set the number of individuals N s and each individual parameter vector Make each entity randomly distributed in the parameter search space to cover various physical states existing at the front of the explosion shock wave;

[0126] S35. Use the fusion differential evolution mechanism to update the position of each bitterling individual:

[0127]

[0128] in, represents the global optimal parameter vector with the best performance in the tth generation, α is the global exploration weight, β is the differential evolution weight, and are two sets of randomly selected individual parameters to provide differential information, γ is the reward factor weight, is the individual reward function, which is used to reward individuals with low errors in predicting the state of the explosion shock wave front, so that the update process optimizes the capture of changes in explosion energy transfer and wavefront attenuation:

[0129]

[0130] Among them, the fitness function for:

[0131]

[0132] Where C is the positive regulation constant;

[0133] S36. The global optimal parameter vector is updated through an information sharing mechanism. The information sharing mechanism integrates the contributions of each individual to the dynamic evolution of the explosion shock wave front during the global search process:

[0134]

[0135] Among them, the weight factor Defined as:

[0136]

[0137] S37. Set the stopping condition, when the preset maximum number of iterations T is reached max Or the fitness function changes to satisfy:

[0138]

[0139] Among them, ∈ is the convergence threshold, terminating the optimization iteration;

[0140] S38. Output the optimal parameter vector Determine the optimal solution region of the nonlinear prediction model of the explosion shock wave front.

[0141] In this embodiment, S4 specifically includes: S4 includes the following steps:

[0142] S41. In the global optimal parameter vector Θ * On this basis, the search range Θ of the local optimization phase of the hierarchical particle swarm optimization algorithm is set local ,in It is used to describe the parameter fine-tuning range of the nonlinear prediction model of the explosion shock wave front in the local area, and capture the local energy transfer, subtle waveform changes and attenuation characteristics of the explosion shock wave front;

[0143] S42. Initialize the hierarchical particle swarm optimization algorithm population and set the number of particle swarm individuals N p And the parameter vector of each particle Make each particle in Θ local Random distribution within

[0144] S43. A multi-level search structure is used to divide the particle swarm into L levels, each level corresponding to a different search granularity, to optimize the global local features and subtle local features within the local area of ​​the explosion shock wave front respectively;

[0145] S44. Update the particle positions within each level l, using adaptive inertia weight adjustment and global-local search fusion strategy:

[0146]

[0147] in, represents the parameter vector of the jth particle in the lth layer of the tth generation, represents the local optimal parameter vector in the lth layer, Represents the overall optimal parameter vector in the current local search, w l (t) is the adaptive inertia weight of the lth layer, which is used to balance the search speed and accuracy, c l is the learning factor of the lth layer, r l is a random variable that follows a uniform distribution [0,1];

[0148] S45. Introduce an adaptive neighborhood search mechanism to perform neighborhood perturbations on each particle within the current level and update the neighborhood parameter vector as follows:

[0149]

[0150] Among them, δ l represents the neighborhood search step size of the lth layer, ξ is a random perturbation taken from the standard normal distribution, which is used to capture the small fluctuations of the explosion shock wave front in the local area and refine the description of the local nonlinear evolution law of the explosion shock wave front nonlinear prediction model;

[0151] S46. Repeat steps S44 to S45 until the local optimization convergence condition is met, that is, when the local optimal parameter vector update in any level satisfies:

[0152]

[0153] Among them, ∈ localis the local search convergence threshold, the iteration is terminated, and the updated explosion shock wave front nonlinear prediction model parameter vector Θ is output updated The parameter vector of the nonlinear prediction model of the explosion shock wave front reflects the optimization result after fine-tuning the energy transfer, waveform attenuation and propagation characteristics of the explosion shock wave front in the local area.

[0154] In this embodiment, S4 specifically includes implementing adaptive inertia weight adjustment for each level l, setting its weight w l The update of (t) is:

[0155]

[0156] Among them, w max,l With w min,l Represent the initial and final inertia weights of the lth layer, T local is the maximum number of iterations in the local optimization phase.

[0157] Example 1:

[0158] At 9:30 am on June 15, 2024, a military research institute conducted a high-energy explosive explosion experiment at an explosion shock wave test site in the Gobi region of northwest China. The experiment aimed to study the propagation characteristics and attenuation laws of shock waves under complex terrain conditions. The test area is a typical desert terrain with loose sand and gravel on the ground, low air humidity (about 15%), and a temperature of up to 32°C. The research team deployed 20 high-precision pressure sensors (model: PXS-5000, sampling frequency 1MHz), 6 high-speed cameras and 3 laser interferometers in the experimental site to monitor the propagation trajectory, peak pressure, attenuation rate and impact range of the explosion shock wave.

[0159] The experiment used a TNT block (mass 5kg, theoretical explosive energy 21MJ), controlled the detonation time through a remote detonator, and used an independent data acquisition system to record the shock wave propagation process. Due to the large terrain undulations in the area, the researchers expected that the shock wave propagation trajectory would be affected by the terrain, resulting in certain refraction and reflection effects, leading to nonlinear changes in the propagation path.

[0160] After the explosives were detonated, a shock wave formed within 0.5ms and propagated in all directions at a speed of 1.9km / s. The pressure sensors at the experimental site began to collect data, and the high-speed camera captured the front propagation trajectory of the shock wave.

[0161] The research team first used traditional finite element simulation methods (based on CFD computational fluid dynamics models) to predict the propagation of the shock wave front. However, under the influence of complex terrain, the simulation results showed significant errors:

[0162] The calculated shock wave peak pressure is 0.95 MPa, while the highest peak pressure measured by the actual sensor is 1.12 MPa, with an error of 15%.

[0163] The predicted shock wave propagation distance was 38.2 m, while the actual measured value was 40.5 m, with an error of 2.3 m (5.7%). The calculation time was as long as 845 s, which could not meet the requirements of fast analysis.

[0164] The research team then used the particle swarm optimization-based explosion shock wave front distribution prediction method of the present invention to analyze the experimental data. During the data preprocessing process, the system automatically eliminated four abnormal data points caused by sensor signal interference and used the improved bitter fish optimization algorithm for parameter search.

[0165] Global search optimization: An adaptive feedback factor λ(t) is used to adjust the optimization step size, dynamically adjust the search direction, and enhance adaptability to complex environmental variables. After 320 iterations, the model determined the optimal solution region, and the error converged to within ±2.5%.

[0166] Fine-tuning using hierarchical particle swarm optimization: Within the optimal solution region, the system uses a hierarchical particle swarm optimization algorithm to divide the particle swarm into three levels, optimizing global propagation characteristics, local energy transfer, and fine waveform changes respectively. After 50 rounds of iteration, the model ultimately determines the optimal parameter vector, reducing the shock wave propagation error to ±1.3%.

[0167] Final prediction results: predicted peak pressure: 1.10MPa (relative error 1.8%, 13.2% lower than the traditional method); predicted propagation distance: 40.2m (relative error 0.7%, 5% lower than the traditional method); calculation time: 540s (36% less than the traditional method).

[0168] The final error range is stabilized within ±1.3%, which is more than 50% lower than the error of traditional methods.

[0169] The research team compared the performance of the two methods in experimental scenarios, and the data is as follows:

[0170]

[0171] The method of the present invention is significantly superior to traditional methods in terms of calculation accuracy, time efficiency and error control, and provides a more accurate and efficient solution for explosion shock wave prediction.

[0172] The success of this test verifies the ability of the present invention to predict the distribution of shock wave fronts in complex environments, and also proves its value in practical engineering applications;

[0173] Research results show that the proposed method has significantly lower prediction errors than traditional methods in complex environments. It can be used to refine protective structure design, optimize battlefield blast effect assessment, and enhance safety and protection capabilities. Computation time was reduced by 36%, enabling the method to provide reliable prediction results more quickly, potentially useful for emergency response to explosion accidents and battlefield situation analysis. Through an adaptive optimization algorithm, the proposed method can adapt to varying terrain and meteorological conditions, improving the robustness of shock wave propagation predictions.

[0174] This example verifies the effectiveness of the present invention in a real explosion test environment. Compared with traditional methods, the particle swarm optimization-based explosion shock wave front distribution prediction method of the present invention reduces the error by more than 50% and improves the computational efficiency by 36%. It can accurately predict the shock wave propagation trajectory and attenuation characteristics in complex environments, providing reliable technical support for explosion effect analysis, and can be widely used in engineering blasting and military protection fields.

[0175] In the process of optimizing the parameters of the nonlinear prediction model of the shock wave front, the present invention adopts an improved bitter fish optimization algorithm and introduces an adaptive feedback factor and a multidimensional chaotic perturbation mechanism to enable the optimization algorithm to dynamically adjust the search step size and direction, thereby enhancing the global exploration capability of the complex nonlinear shock wave evolution process. By introducing the adaptive feedback factor, the optimization process can respond to the changing trend of the shock wave front in real time, and the global search perturbation is introduced in combination with the Logistic chaotic map and the Tent map, thereby enhancing the adaptability to the shock wave propagation parameters.

[0176] The present invention proposes a hierarchical particle swarm optimization method. By setting a multi-layer search structure within the global optimal solution region, the particle swarm is divided into different levels. Adaptive inertia weight and neighborhood search mechanism are used in each level to achieve fine optimization of the local area. During the local optimization process, random perturbation parameters are introduced, so that the particle swarm can flexibly adjust the search accuracy and direction between different levels, thereby improving the fitting accuracy of the shock wave front parameters.

[0177] This invention incorporates a time series constraint mechanism into the nonlinear prediction model for shock wave fronts and sets a time series state vector. This allows the model to be dynamically updated using historical state information, creating a shock wave propagation prediction model that better conforms to physical laws. By recursively updating model parameters, the prediction process can adjust its response to environmental variables in real time, improving the ability to identify shock wave attenuation trends.

[0178] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for predicting the front distribution of explosion shock waves based on particle swarm optimization, characterized in that: The steps include: S1. Collecting explosion shock wave experimental data and explosion shock wave simulation data, preprocessing the explosion shock wave experimental data to form an initial explosion shock wave data set after preprocessing; S2. Establish a nonlinear prediction model for the explosion shock wave front based on the initial explosion shock wave dataset; S3. Using an improved bitterfish optimization algorithm, a global search is performed on the nonlinear prediction model of the explosion shock wave front to determine the optimal solution area; S4. Within the optimal solution region, fine-tune the parameters of the nonlinear prediction model for the blast shock wave front using a hierarchical particle swarm optimization algorithm to obtain an updated nonlinear prediction model for the blast shock wave front; S5. Use the updated nonlinear prediction model of the explosion shock wave front to predict the front evolution of the initial explosion shock wave data set formed in step S1, and output the final prediction results, which include the explosion shock wave front propagation trajectory, peak pressure, attenuation rate and impact range.

2. The method for predicting explosion shock wave front distribution based on particle swarm optimization according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Collecting explosion shock wave experimental data and explosion shock wave simulation data. The explosion shock wave experimental data is obtained by experimental measurement equipment, including explosion energy, medium density, shock wave propagation velocity and front morphological characteristics. The explosion shock wave simulation data is obtained by numerical simulation calculation, including shock wave isentropic evolution characteristics, pressure attenuation characteristics and shock wave propagation parameters under the influence of different environmental variables; S12. Using a data format standardization method to uniformly format the explosion shock wave experimental data, the experimental data, simulation data and real-time monitoring data are converted into an explosion shock wave experimental data matrix; S13. Performing data denoising on the explosion shock wave experimental data matrix to remove noise components in the signal and performing normalization processing to map all characteristic variables to the interval [0, 1] to obtain a normalized explosion shock wave experimental data matrix; S14. Perform outlier detection on the normalized explosion shock wave experimental data matrix, identify abnormal data points, and use the nearest neighbor interpolation method to complete the explosion shock wave experimental data matrix to obtain the interpolated explosion shock wave experimental data matrix. The completed explosion shock wave experimental data matrix D final As the initial explosion shock wave dataset.

3. The method for predicting explosion shock wave front distribution based on particle swarm optimization according to claim 1, characterized in that: The S2 comprises the following steps: S21. Based on the initial explosion shock wave data set D final , a nonlinear prediction model for the explosion shock wave front is constructed. The nonlinear prediction model for the explosion shock wave front is based on the physical characteristics of the explosion shock wave, the multi-physical field coupling effect, the time-varying characteristics and the influence of environmental variables, and establishes a mapping relationship between the input parameter set X and the explosion shock wave front evolution result set Y: Y=f(X,Θ); Where X={x1,x2,...,x N } represents the input parameter set, x i is the input characteristic variable of the explosion shock wave, including the explosion energy E b , medium density ρ, shock wave propagation velocity v w and frontier morphological characteristics S f ,Y={y1,y2,...,y M } represents the spatiotemporal evolution of the explosion shock wave front, including the propagation trajectory, peak pressure, attenuation rate, and impact range of the explosion shock wave front. Θ is the parameter set of the nonlinear prediction model of the explosion shock wave front. S22. Based on the initial explosion shock wave data set D final , the least squares regression method is used to make preliminary parameter estimation of the nonlinear prediction model of the explosion shock wave front, the initial model parameters Θ0 are set, and the initial estimated values ​​of the initial model parameters are calculated: Among them, Θ0 is the initial model parameter, which makes the nonlinear prediction model of the explosion shock wave front initially fit the evolution trend of the explosion shock wave front; S23. Modify the nonlinear prediction model of the explosion shock wave front and construct the environmental variable matrix Z to describe the impact of the environment on the evolution of the explosion shock wave front: Z={z1,z2,...,z K } Among them, Z represents the environmental variable matrix, z k Represents environmental variables, including temperature, humidity, atmospheric pressure and medium characteristics, and combines the environmental variable matrix Z to adjust the weight parameters of the nonlinear prediction model of the explosion shock wave front; S24. Apply time series constraints to the nonlinear prediction model of the explosion shock wave front and set the time series state vector: S t =g(S t-1 ,X t ,I t ); Among them, S t It represents the state of the explosion shock wave front at time t, S t-1 is the state at the previous moment, X t is the current input parameter, Θ t is the current model parameter; S25. Obtain the final nonlinear prediction model of the explosion shock wave front: Y t =f(X t ,Z t ,S t-1 ,Θ); Among them, S t-1 Indicates the state of the explosion shock wave front at time t-1.

4. The method for predicting explosion shock wave front distribution based on particle swarm optimization according to claim 3, characterized in that: The S3 includes the following steps: S31. Define the current model parameter set Θ = {θ1, θ2, ..., θ P }, where each model parameter θ p All meet the following requirements: in, and Represent the minimum and maximum physical constraint values ​​of the pth parameter respectively; S32. An adaptive feedback factor λ(t) is introduced to reflect the relative value of the prediction error of the explosion shock wave front and its time-varying gradient, so that the improved bitter fish optimization algorithm can dynamically respond to the nonlinear changes of the wave front during the explosion process: Among them, Y t represents the actual explosion shock wave front state at time t, κ is the error scaling constant, and η is the error change rate weight; S33. Using multidimensional chaotic mapping to generate dynamic perturbation factor χ(t) and diffusion factor ψ(t), we improve the global search capability of the bitter fish optimization algorithm for the complex and nonlinear behavior of explosion shock waves. The chaotic perturbation factor adopts logistic mapping: χ(t+1)=μ·χ(t)·(1-χ(t)); The diffusion factor ψ(t) adopts the improved Tent mapping: Among them, μ is the chaotic mapping parameter, δ is the boundary parameter of the Tent mapping, χ(t) and ψ(t) are used to simulate the uncertain perturbation and diffusion trend of the explosion shock wave front in different environments, respectively, reflecting the physical phenomenon of random diffusion of energy in the explosion shock wave; S34. Initialize the population of the improved bitter fish optimization algorithm and set the number of individuals N s and each individual parameter vector Make each entity randomly distributed in the parameter search space to cover various physical states existing at the front of the explosion shock wave; S35. Use the fusion differential evolution mechanism to update the position of each bitterling individual: in, represents the global optimal parameter vector with the best performance in the tth generation, α is the global exploration weight, β is the differential evolution weight, and are two sets of randomly selected individual parameters to provide differential information, γ is the reward factor weight, is the individual reward function, which is used to reward individuals with low errors in predicting the state of the explosion shock wave front, so that the update process optimizes the capture of changes in explosion energy transfer and wavefront attenuation: Among them, the fitness function for: Where C is the positive regulation constant; S36. The global optimal parameter vector is updated through an information sharing mechanism. The information sharing mechanism integrates the contributions of each individual to the dynamic evolution of the explosion shock wave front during the global search process: Among them, the weight factor Defined as: S37. Set the stopping condition, when the preset maximum number of iterations T is reached max Or the fitness function changes to satisfy: Among them, ∈ is the convergence threshold, terminating the optimization iteration; S38. Output the optimal parameter vector Determine the optimal solution region of the nonlinear prediction model of the explosion shock wave front.

5. The method for predicting explosion shock wave front distribution based on particle swarm optimization according to claim 1, characterized in that: The S4 specifically includes: The S4 includes the following steps: S41. In the global optimal parameter vector Θ * On this basis, the search range Θ of the local optimization phase of the hierarchical particle swarm optimization algorithm is set local ,in It is used to describe the parameter fine-tuning range of the nonlinear prediction model of the explosion shock wave front in the local area, and capture the local energy transfer, subtle waveform changes and attenuation characteristics of the explosion shock wave front; S42. Initialize the hierarchical particle swarm optimization algorithm population and set the number of particle swarm individuals N p And the parameter vector of each particle Make each particle in Θ local Internal random distribution; S43. A multi-level search structure is used to divide the particle swarm into L levels, each level corresponding to a different search granularity, so as to optimize the global local features and subtle local features in the local area of ​​the explosion shock wave front respectively; S44. Update the particle positions within each level l, using adaptive inertia weight adjustment and global-local search fusion strategy: in, represents the parameter vector of the jth particle in the lth layer of the tth generation, represents the local optimal parameter vector in the lth layer, Represents the overall optimal parameter vector in the current local search, w l (t) is the adaptive inertia weight of the lth layer, which is used to balance the search speed and accuracy, c l is the learning factor of the lth layer, r l is a random variable that follows a uniform distribution [0,1]; S45. Introduce an adaptive neighborhood search mechanism to perform neighborhood perturbations on each particle within the current level and update the neighborhood parameter vector as follows: Among them, δ l represents the neighborhood search step size of the lth layer, ξ is a random perturbation taken from the standard normal distribution, which is used to capture the small fluctuations of the explosion shock wave front in the local area and refine the description of the local nonlinear evolution law of the explosion shock wave front nonlinear prediction model; S46. Repeat steps S44 to S45 until the local optimization convergence condition is met, that is, when the local optimal parameter vector update in any level satisfies: Among them, ∈ local is the local search convergence threshold, the iteration is terminated, and the updated explosion shock wave front nonlinear prediction model parameter vector Θ is output updated The parameter vector of the nonlinear prediction model of the explosion shock wave front reflects the optimization result after fine-tuning the energy transfer, waveform attenuation and propagation characteristics of the explosion shock wave front in the local area.

6. The method for predicting explosion shock wave front distribution based on particle swarm optimization according to claim 5, characterized in that: The S4 specifically includes implementing adaptive inertia weight adjustment for each level l, setting its weight w l The update of (t) is: Among them, w max,l With w min,l Represent the initial and final inertia weights of the lth layer, T local is the maximum number of iterations in the local optimization phase.

Images