A method for calculating a helicopter damage probability

Abstract

The application provides a helicopter damage probability calculation method, and relates to the technical field of physical measurement testing.The application fundamentally solves the prediction inaccuracy problem caused by the mode dependence of damage accumulation and the inherent uncertainty of a model in the prior art under continuous impact simulation by constructing a multi-dimensional, self-feedback and self-cognition calculation system with uncertainty, can greatly improve the accuracy of structural chain failure risk prediction, realizes accurate evaluation of "threat-dependent structural vulnerability", and enhances the safety and reliability of the simulation system in complex and changeable environments.

Description

Technical Field

[0001] This invention relates to the field of physical measurement and testing technology, specifically a method for calculating the probability of helicopter damage. Background Technology

[0002] For rotorcraft like helicopters, operational survivability is a key indicator of their core effectiveness. With increasing environmental complexity and escalating adversarial intensity, the threat patterns faced by aircraft have evolved from traditional single-point, high-energy attacks to high-density, multi-point, and time-saturated composite attacks. Therefore, the technology for assessing the structural integrity of aircraft urgently needs a strategic transformation from traditional static, post-incident damage detection to dynamic, real-time damage process prediction and survivability assessment to support next-generation protective design, tactical decision-making, and intelligent mission planning.

[0003] Existing technologies mainly fall into two categories: post-incident detection and pre-incident prediction. For example, the damage non-destructive testing device disclosed in Chinese patent CN118671097A focuses on the accurate diagnosis of static damage that has already occurred, but it cannot predict the dynamic failure process of a structure under continuous impact. This predictive challenge is precisely what pre-incident simulation technology needs to solve, but this technological approach still faces the following core challenges: The simplification of damage state characterization: Existing predictive models often use a single macroscopic scalar (such as energy absorption value) to quantify damage, which cannot distinguish physically distinct microscopic damage patterns (such as matrix cracking versus fiber breakage). Therefore, the models struggle to accurately predict the differentiated resistance of the "properties" of preceding damage to subsequent threats.

[0004] One-way barrier between computational scales: In simulations, the information flow from macroscopic impact parameters to microscopic damage analysis is usually unidirectional, lacking a real-time feedback loop from microscopic damage diagnosis results to the macroscopic computational model. This simplifies the simulation to a linear superposition of independent events, failing to capture the crucial "damage mode-dependent performance evolution" process.

[0005] The conflict between efficiency and reliability: High-fidelity physical simulations are computationally expensive and difficult to apply in real time. Meanwhile, the surrogate models used for acceleration are typically deterministic, lacking the ability to quantify the uncertainty of their own predictions. This can lead to a significant underestimation of failure risk when facing unknown operating conditions due to the inability to assess prediction confidence.

[0006] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a method for calculating the probability of helicopter damage, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A method for calculating helicopter damage probability, comprising the following steps: S1: Initialize a dynamic damage state map characterizing the three-dimensional structural model of the helicopter. The dynamic damage state map discretizes the three-dimensional structural model into multiple structural units and associates an initial material property state parameter with each structural unit. S2: When a preset impact event interacts with the target structural unit in the three-dimensional structural model, the macroscopic impact parameters of the interaction are calculated. S3: Based on the macroscopic impact parameters, call the preset material microscopic damage mechanism model to quantitatively evaluate the degree of internal damage to the target structural unit caused by this interaction; S4: Based on the internal damage degree of the quantitative assessment, calculate and generate a material performance degradation coefficient characterizing the change in the elastic resistance of the target structural unit, and use the material performance degradation coefficient to update the material performance state parameters of the corresponding target structural unit in the dynamic damage state map; S5: When subsequent impact events interact with structural units in the three-dimensional structural model, damage probability is calculated based on the updated material property state parameters in the dynamic damage state map.

[0009] Compared with the prior art, the beneficial effects of the present invention are as follows: In view of the limitation of single damage characterization, the present invention innovatively introduces the concept of multimodal damage state vector. By decoupling and quantifying different physical damage modes such as fiber, matrix, and interlayer, the present invention constructs a closed-loop calculation framework for macro-micro bidirectional information interaction, so that the specific mode diagnosed by micro-damage can be used as a key dynamic parameter to provide real-time feedback and correct the calculation model of subsequent macro-penetration probability, thereby enabling the simulation to accurately predict the evolution process of elastic performance dependent on the damage mode. Furthermore, to address the dilemma between efficiency and reliability, this invention introduces an uncertainty-aware surrogate model. This model not only rapidly outputs the macroscopic performance degradation coefficient but also simultaneously outputs the prediction confidence level of this inference. This prediction confidence level is actively applied as a dynamic risk adjustment factor, automatically and intelligently adjusting the final damage probability when the model's prediction becomes uncertain. This mechanism fundamentally enhances the robustness and safety of the entire simulation system in the face of unknown and extreme conditions. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the overall application process of the present invention; Figure 2 This is a schematic diagram of the execution logic of step S1 of the present invention; Figure 3 This is a schematic diagram illustrating the execution logic of steps S2 and S3 of the present invention; Figure 4 This is a schematic diagram illustrating the execution logic of steps S4 and S5 of the present invention. Detailed Implementation

[0011] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0012] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0013] Example 1: Please see Figures 1 to 4 The present invention provides a technical solution: A method for calculating helicopter damage probability, comprising the following steps: S1: Initialize a dynamic damage state map characterizing the three-dimensional structural model of the helicopter. The dynamic damage state map discretizes the three-dimensional structural model into multiple structural units and associates an initial material property state parameter with each structural unit. S2: When a preset impact event interacts with a target structural unit in a three-dimensional structural model, the macroscopic impact parameters of that interaction are calculated. S3: Based on macroscopic impact parameters, a preset material microscopic damage mechanism model is invoked to quantitatively assess the degree of internal damage to the target structural unit caused by this interaction. S4: Based on the quantitative assessment of the degree of internal damage, calculate and generate a material performance degradation coefficient that characterizes the change in the elastic resistance of the target structural unit, and use this material performance degradation coefficient to update the material performance state parameters of the corresponding target structural unit in the dynamic damage state map; S5: When subsequent impact events interact with structural units in the three-dimensional structural model, damage probability is calculated based on the updated material property state parameters in the dynamic damage state map.

[0014] It should be further explained that, Figure 1In the middle section, "Macroscopic Impact and Microscopic Damage Assessment" summarizes steps S2 and S3, reflecting the starting point of the linkage between macroscopic and microscopic aspects; "Dynamic Damage Map Update and Feedback" summarizes step S4, emphasizing the dynamic nature and closed-loop characteristics of the system. "Probability prediction based on uncertainty perception" summarizes step S5 and its subsequent core features, highlighting the final intelligent risk assessment. Further explanation: The material performance state parameters are defined as multimodal damage state vectors, including exponential components that characterize the integrity of fibers, matrix, and interlayer bonds, respectively. The exponential components are calculated and determined by the normalized brittle fracture energy ratio, ductile deformation energy ratio, and shear wave attenuation factor, respectively. The equivalent ballistic performance degradation coefficient is obtained by weighted geometric mean fusion of the exponential components of the multimodal damage state vector. The weights used in the weighted geometric mean fusion are dynamically and adaptively determined based on the threat characteristic parameters of the current impact event.

[0015] The core technical feature of this embodiment lies in associating a multimodal damage state vector with each discretized structural unit in the dynamic damage state map. Each dimension of this multimodal damage state vector corresponds to a physically distinguishable microscopic damage mode derived from materials science, specifically including fiber integrity, matrix integrity, and interlaminar bond integrity. By decoupling different damage modes, this embodiment can capture and quantify the differentiated "internal injury" characteristics caused by different types of impacts (including high-speed armor-piercing projectiles and low-speed fragments). When this multimodal damage state vector is used for subsequent calculations, it can dynamically and selectively assess the remaining protective capability of the structure based on the type of impending threat, thereby achieving accurate prediction of "threat-dependent structural vulnerability".

[0016] The key parameters involved in this embodiment are defined as follows: The multimodal damage state vector, whose parameter symbols are: It is used to comprehensively characterize the microscopic damage state inside the structural unit; it is calculated using the following three component parameters; the initial state of this multimodal damage state vector is that each component is 1.0, indicating that the structure is intact; Fiber integrity index, its parameter symbol is This is The first component is a dimensionless scalar defined in the interval [0,1], used to characterize the integrity of the reinforcing fibers that bear the main tensile load in the composite material; a value of 1.0 indicates that the fiber has not broken or been damaged, and a value close to 0 indicates that the fiber has broken on a large scale and completely lost its load-bearing capacity; the fiber integrity index is determined by the normalized brittle fracture energy ratio. The matrix integrity index, whose parameter symbol is... This is The second component is a dimensionless scalar defined in the interval [0,1], used to characterize the integrity of the resin matrix that encapsulates the fibers and transmits the load; a value of 1.0 indicates that the matrix structure is intact, and a value close to 0 indicates that the matrix has undergone extensive microcracks or plastic failure. This matrix integrity index is determined by calculating the normalized toughness deformation energy ratio.

[0017] Interlayer bond integrity index, its parameter symbol is This is The third component is a dimensionless scalar defined in the interval [0,1], used to characterize the interfacial bond strength between composite material layers. A value of 1.0 indicates no delamination or separation between layers, while a value close to 0 indicates severe delamination, where the structure cannot effectively transmit interfacial shear forces. This index is determined by the normalized shear wave attenuation factor.

[0018] Normalized brittle fracture energy percentage, with parameter sign as follows: This is a dimensionless scalar, and its physical meaning is the proportion of energy carried by high-frequency, high-amplitude burst signals (characterizing brittle events such as fiber fracture) in the acoustic emission signal excited by an impact event, relative to the total signal energy. The calculation is based on acoustic emission nondestructive testing technology in materials science and Fourier transform or wavelet transform theory in signal processing. The calculation logic is as follows: First, a fast Fourier transform is performed on the acoustic emission time-domain signal to obtain its spectrum. Second, based on the frequency threshold pre-calibrated in the materials acoustic experiment (in this embodiment, the frequency range above 500kHz is defined as the brittle fracture characteristic frequency band), the signal energy integral within this characteristic frequency band is calculated and denoted as the brittle event energy. Then, the total signal energy integral of the entire frequency band is calculated. The brittle event energy is divided by the total energy to obtain an initial ratio. Finally, this initial ratio is mapped using a S-type logistic function, which compresses any positive real number input to the (0,1) interval, thus obtaining the final... .

[0019] If the calculated brittle event energy is 85 units and the total energy is 100 units, with an initial ratio of 0.85, substituting 0.85 into the logistic function for normalization yields a final output value of 0.92. .

[0020] Normalized ductile deformation energy percentage, its parameter sign is: Its physical meaning is the proportion of energy carried by low-frequency, long-duration continuous signals (characterizing ductile events such as microcrack propagation and plastic flow in the matrix) in the total signal energy of acoustic emission signals. Its determination method is similar to... Similarly, the difference lies in its characteristic frequency band, which is the low-frequency range of 50kHz to 200kHz as determined by experiments; details will not be elaborated further. Normalized shear wave attenuation factor, its parameter symbol is: The physical meaning of attenuation is the degree of amplitude reduction of a shear stress wave as it penetrates the interface of a composite material layup. The calculation is based on the acoustic simulation of Beer-Lambert's law for wave propagation in a medium in solid-state physics. The calculation method is as follows: In the wave dynamics model, the amplitude of the shear wave before entering the target layup interface is measured and recorded as the incident amplitude. Next, the amplitude after penetrating the interface is measured and recorded as the transmitted amplitude. Then, according to the acoustic form of Beer-Lambert's law, the attenuation coefficient is calculated, which is proportional to the natural logarithm of the ratio of the incident amplitude to the transmitted amplitude. Finally, this attenuation coefficient is compared with the maximum attenuation coefficient, experimentally calibrated to characterize complete material delamination, and the ratio is calculated and limited to the interval [0,1] to obtain the final attenuation coefficient. .

[0021] If the incident amplitude is 10 units and the transmitted amplitude is 2 units, the calculated attenuation coefficient is 1.61. If the calibrated maximum attenuation coefficient is 2.0, then the final... Dividing 1.61 by 2.0 gives 0.805.

[0022] The equivalent ballistic performance degradation coefficient, whose parameter symbol is: Its cause The calculated, comprehensive dimensionless scalar, defined in the [0,1] interval, is used to correct the ballistic performance parameters of materials in macroscopic ballistic calculations. In this embodiment, the complete calculation process for initializing the dynamic damage state map is as follows: 1.1) The initial input consists of a set of three fundamental physical quantities for impact events, calculated by the upstream acoustic emission characteristic analysis model: normalized brittle fracture energy percentage. Normalized toughness deformation energy ratio and normalized shear wave attenuation factor .

[0023] 1.2) Processing Step One: Based on Input The fiber integrity index is determined by the following formula. : The value is equal to 1.0 minus The value of .

[0024] Input-based The matrix integrity index is determined by the following formula. : The value is equal to 1.0 minus The value of .

[0025] Input-based The interlayer adhesion integrity index is determined by the following formula. : The value is equal to 1.0 minus The value of .

[0026] 1.3) Processing Step Two: Calculate the three indices obtained in the previous step... , and The three orthogonal components of this three-dimensional vector together constitute the multimodal damage state vector of the structural unit after this impact. .

[0027] 1.4) Processing Step 3: Convert the multimodal damage state vector Using current threat characteristic parameters as input, the system calculates and outputs a comprehensive equivalent ballistic performance degradation coefficient. .

[0028] 1.5) Step 4: At the beginning of the entire simulation task, execute the initialization process: traverse all structural elements of the 3D structural model, and generate the multimodal damage state vector of each structural element. The initial vector is set to [1.0, 1.0, 1.0], and its initial "equivalent ballistic performance degradation coefficient" is calculated from it. The value is 1.0. Store this initial vector and coefficients in the corresponding positions of the dynamic damage state map.

[0029] 1.6) During the simulation, whenever a structural element is impacted, the final output of the process is the updated multimodal damage state vector of that structural element. and its corresponding equivalent ballistic performance degradation coefficient These two values ​​will be used to update the dynamic damage state map.

[0030] In this embodiment, the geometric mean is in product form, which better reflects the physical nature of the "weakest link effect" in composite material damage. That is, when the integrity of any damage mode approaches zero, the overall elastic resistance will decrease sharply or even approach zero.

[0031] The weighting coefficients are dynamically determined by the "threat type weighting mapping function"; this "threat type weighting mapping function" takes threat characteristic parameters as input; in this embodiment, the "threat characteristic parameters" include the specific kinetic energy of the fragments. and shape factor ; It should be noted that: specific kinetic energy It is a physical quantity used to characterize the degree of energy concentration of an impactor in the local area of ​​its impact point; it is defined as the ratio of the total kinetic energy of the impactor at the moment of contact with the target to its equivalent windward area acting on the target surface. The calculation of this parameter is entirely based on the kinetic energy theorem and the principle of geometric projection in classical mechanics. The prerequisite parameters required for its calculation are all directly derived from the real-time calculation results of the macroscopic ballistic tracking simulation cycle of this invention; In the simulation process, whenever a fragment is about to interact with a structural unit, its specific kinetic energy is... The calculation process is as follows, obtaining the prerequisite parameters: Remaining mass of fragments : Obtain the remaining mass of the fragment after experiencing previous penetration events from the macroscopic ballistic tracking module.

[0032] Fragment Remaining Velocity : Obtain the current flight speed of the fragment from the same macroscopic ballistic tracking module.

[0033] Fragment equivalent windward area This parameter characterizes the projected area of ​​a fragment along its flight velocity vector on a plane perpendicular to that vector. For fragments with a pre-defined regular shape, this area is calculated directly from its geometric dimensions and instantaneous attitude angle. For fragments with an irregular shape, this value is approximated, in its initial definition, by the diameter of an equivalent sphere or a pre-defined average cross-sectional area.

[0034] Calculate the current total kinetic energy of the fragment. The calculation relationship is as follows: The value is equal to 0.5 multiplied by the remaining mass of the fragment. Multiply by the remaining velocity of the fragment. The square of the kinetic energy; The value is equal to the total kinetic energy. Divide by the equivalent windward area of ​​the fragments .

[0035] When the fragment impacts a beam structure inside a skin, its remaining mass... The remaining velocity is 0.002 kg. With a speed of 800 m / s, its fragment equivalent frontal area is... The area is 0.00001 square meters. Its total kinetic energy is 0.5 × 0.002 × (800^2) = 640 joules. Its specific kinetic energy is 640 / 0.00001 = 64,000,000 joules / square meter, or 64 megajoules / square meter. This calculation result will be used as one of the inputs to the threat type weight mapping function.

[0036] shape factor It is a dimensionless parameter used to quantify the "sharpness" or "elongation" of an impactor's shape. Its core function is to distinguish between easily penetrating threats such as rod-shaped and needle-shaped fragments, and threats that tend to cause large-area impacts such as blocky and sheet-shaped fragments. In this invention, the value range of this parameter is normalized to the [0,1] interval. It is constructed by the ratio of the critical geometric dimensions of the fragments; shape factor. The calculation process is as follows: Determine the maximum characteristic length of the fragment : Retrieves the maximum dimension of the fragment in its geometry as defined in the threat model database. For a cylindrical fragment, this is its length; for a cuboid fragment, this is the length of its longest side.

[0037] Fragment feature width : Obtain the representative width of the fragment in the direction perpendicular to its maximum characteristic length. For a cylinder, this is its diameter; for a cuboid, this is the length of its second longest side.

[0038] Calculate the slenderness ratio The calculation relationship is as follows: slenderness ratio The value is equal to the maximum characteristic length of the fragment. Divide by the characteristic width of the fragment .

[0039] For slenderness ratio Normalization is performed to obtain the final shape factor. This embodiment employs a normalization method based on the arctangent function, which can smoothly map any positive slenderness ratio to the interval [0,1). The calculation relationship is: shape factor... The value is equal to 2 divided by pi, then multiplied by the slenderness ratio. The arctangent value. As the aspect ratio increases from 0, the output value increases rapidly from 0 and gradually saturates, approaching 1.0 infinitely.

[0040] The following is an example of an armor-piercing projectile: The threat model is a length-to-width ratio... A rod-shaped jet with a shape factor of 10. The value is 0.936. This is a high value approaching 1.0, indicating a "sharp" threat.

[0041] The following is an example of explosive fragments: The threat model is a length-to-width ratio The fragment is a near-cubic piece with a shape factor of 1.2. It is 0.558. This is an intermediate value, indicating a "blunt" threat.

[0042] The "threat type weight mapping function" internally consists of a pre-defined lookup table or a small rule-based reasoning system; specifically: If we compare kinetic energy High and shape factor To characterize armor-piercing projectiles, the fiber integrity index is assigned the highest weight of 0.7, while the matrix integrity index and interlayer bond integrity index are both assigned a weight of 0.15.

[0043] If we compare kinetic energy Low and shape factor For large blunt-tipped fragments, the matrix integrity index and interlaminar bond integrity index are assigned a high weight of 0.4, while the fiber integrity index is assigned a weight of 0.2.

[0044] This embodiment requires ensuring that the sum of all weights is always 1.0. The weighting coefficients and their values ​​("high", "small", "low", "large") are determined through regression analysis and optimization of a large amount of target penetration physics experimental data under different threats. Equivalent ballistic performance degradation coefficient The value is equal to the product of the weighted power of the fiber integrity index, the weighted power of the matrix integrity index, and the weighted power of the interlayer bond integrity index.

[0045] When the equivalent ballistic performance degradation coefficient As the value approaches 1.0, the material's elastic resistance, which characterizes the structural unit, is closer to its original design state; that is, it remains intact or suffers only minor internal damage. When the equivalent ballistic performance degradation coefficient The closer the value is to 0, the more severe the internal damage accumulation of the structural unit, and the greater the degradation of its macroscopic elastic resistance. Fiber integrity index Equivalent ballistic performance degradation coefficient of final output It exhibits a non-linear positive correlation. When other parameters remain constant, The increase will lead to a degradation coefficient of the equivalent ballistic performance. Monotonically increasing.

[0046] Matrix Integrity Index The final output "equivalent ballistic performance degradation coefficient" "It exhibits a non-linear positive correlation. When other parameters remain constant, The increase will lead to a degradation coefficient of the equivalent ballistic performance. Monotonically increasing.

[0047] Interlayer bonding integrity index The final output "equivalent ballistic performance degradation coefficient" "It exhibits a non-linear positive correlation. When other parameters remain constant, The increase will lead to a degradation coefficient of the equivalent ballistic performance. Monotonically increasing.

[0048] This invention employs a weighted geometric average fusion method, the mathematical essence of which is a weighted power product of the exponential components. According to mathematical principles, for a power function with a base in the interval [0,1], the function value monotonically increases as the base increases. Therefore, any increase in any integrity index will directly lead to the final product result being the "equivalent ballistic performance degradation coefficient". The improvement of "".

[0049] To verify the effectiveness of the "threat-adaptive multimodal damage assessment" method proposed in this invention, the following comparative experiments were designed. Two typical application threats were selected for the experiments: Scenario 1 involved high-speed, small-section armor-piercing projectile (AP) impact, with the primary damage mode being fiber shear fracture; Scenario 2 involved low-speed, large-section explosive fragment impact, with the primary damage modes being large-area matrix crushing and interlayer delamination. The evaluation results of the method of this invention ("the model of this invention") were compared with those of traditional models that only considered a single damage mode ("Comparative Model A", considering only fiber damage; "Comparative Model B", considering only delamination damage) to highlight the significant progress of this invention in terms of accuracy and comprehensiveness.

[0050]

[0051] Comparing the data from Scenario 1 (Experiment AP-1) and Scenario 2 (Experiment Frag-1), in Scenario 1, since the object is an armor-piercing projectile, the model of this invention automatically assigns the highest weight (0.70) to the fiber integrity index. Although this index has decreased to 0.30, due to the relatively good matrix and bond integrity, the final calculated "equivalent ballistic performance degradation coefficient" is still relatively high. The coefficient of performance degradation is 0.51, accurately reflecting that the structure still retains some residual strength despite severe damage. However, in scenario two, facing explosive fragments, the model prioritizes the matrix and bonding (both 0.40). In this case, although the fibers are almost intact (0.90), the severe damage to the matrix and bonding (0.40 and 0.30 respectively) ultimately results in a significantly higher "equivalent ballistic performance degradation coefficient." The calculated value is 0.48. These two scenarios output similar macroscopic degradation coefficients, but the underlying microscopic damage "profiles" are drastically different. This strongly demonstrates that the present invention can intelligently identify and focus on the most critical damage patterns based on threat type, thereby making assessments that are more consistent with physical reality.

[0052] The model of this invention (Experiment AP-1) is compared with comparative model A. Comparative model A only considers fiber damage, and its evaluation result, "equivalent ballistic performance degradation coefficient," is... The original value of 0.30 severely underestimated the residual performance of the structure because it completely ignored the residual support capacity provided by the matrix and bonding interface. The result of the model in this invention (0.51) is a 70% improvement in comparison, providing commanders with more accurate "availability" information and avoiding premature equipment abandonment due to overly pessimistic assessments.

[0053] Similarly, the model of this invention (Experimental Frag-1) is compared with comparative model B. Comparative model B only considers layering, and its evaluation result, "equivalent ballistic performance degradation coefficient," is... "The same value of 0.30 also severely underestimates performance. The result of the model of this invention (0.48) is a 60% improvement in comparison. This set of comparisons clearly shows that the present invention overcomes the one-sidedness of traditional single-dimensional evaluation methods through multimodal fusion, and its evaluation accuracy and comprehensiveness have both made significant improvements by orders of magnitude."

[0054] The comparison between experiments AP-1 and AP-2, and between Frag-1 and Frag-2, demonstrates the model's stability and sensitivity. Under similar threats, when the input damage index undergoes a reasonable small deterioration, the output "equivalent ballistic performance degradation coefficient" shows... "It also decreased smoothly and accordingly, proving the robustness and feasibility of the algorithm design."

[0055] Further explanation: The steps for calculating macroscopic impact parameters include: calculating the theoretical breakdown probability and using the Monte Carlo method to determine the macroscopic interaction result; The theoretical breakdown probability is calculated based on the specific kinetic energy of the fragment and the dynamic limit penetration specific kinetic energy of the structural unit.

[0056] Further explanation: The material micro-damage mechanism model is a stress wave propagation and acoustic emission characteristic analysis model; The microscopic quantitative assessment step outputs the degree of brittle damage dominance characterizing the damage mode; Based on the dominance of brittle damage, the damage mode correction factor is calculated; When calculating the dynamic limit penetration ratio kinetic energy, the damage mode correction factor obtained from the previous interactive calculation is used for correction.

[0057] The following specific implementations will be carried out on the above content: The core technical feature of this embodiment lies in constructing a closed-loop computational framework that enables bidirectional information interaction between the probabilistic penetration determination of macroscopic ballistics and the damage mode diagnosis of microscopic materials science. Specifically, macroscopic impact parameters are used to drive microscopic damage simulation, and the specific damage modes diagnosed by the microscopic damage simulation are fed back in real time and used to dynamically correct the subsequent calculation model of macroscopic penetration probability. This allows the simulation system to characterize how the internal damage modes (including ductile matrix cracking or brittle fiber fracture) of a material after experiencing an impact decisively change its resistance to different types of impacts. This allows the simulation to leap from a simple "energy dissipation" experiment to an accurate prediction of the "damage mode-dependent evolution of ballistic performance." The key parameters involved in this embodiment are defined as follows: The specific kinetic energy of the current fragment, its parameter sign is: The calculation method is the same as the "specific kinetic energy" in step S1. "The same applies, so I will not elaborate further." The dynamic limit penetration ratio kinetic energy of a structural unit, with parameter sign as follows: The physical meaning is the minimum specific kinetic energy required to theoretically penetrate the current target structural unit with a 50% probability; this parameter changes dynamically according to the current damage state of the structural unit.

[0058] Damage mode correction factor, whose parameter symbol is: Defined in The range is used to dynamically adjust the dynamic limit penetration ratio kinetic energy based on the microscopic damage pattern caused by the previous impact. The damage mode correction factor is determined by the calculation of brittle damage dominance.

[0059] The dominance of brittle damage, with parameter symbol as follows: Defined in the [0,1] interval, this parameter quantifies the severity of brittle damage relative to ductile damage in the internal damage caused by a single impact. A value of 1 indicates that the damage is entirely dominated by the brittle mode, while a value of 0 indicates that it is entirely dominated by the ductile mode. In this embodiment, brittle damage includes fiber fracture, and ductile damage includes matrix cracking. The calculation of this parameter is based on the statistical characteristic analysis of acoustic emission signals, specifically the kurtosis and skewness analysis of the signal. The calculation logic is as follows: the fourth-order central moment of the simulated acoustic emission time-domain signal is calculated to obtain the kurtosis value, which is sensitive to the sharp impulses of the "brittle event characteristics" in the signal. Next, the calculated kurtosis value is processed through a pre-calibrated normalization function. This function is established based on a large amount of material impact experimental data and can map kurtosis values ​​of different ranges to the [0,1] interval, thereby obtaining... High kurtosis values ​​correspond to high value.

[0060] If the calculated kurtosis value of the signal is 8.5, and the calibrated normalization function maps kurtosis values ​​greater than 7.0 to close to 1.0, then the final... It is 0.95.

[0061] Theoretical breakdown probability, its parameter sign is: Defined in the interval [0,1], this parameter represents the theoretical probability that a fragment will successfully penetrate a target structural unit under the current conditions. The calculation of this parameter is based on the probability penetration theory in ballistics, specifically employing a probability model based on a normal or Weibull distribution. The theoretical penetration probability is calculated based on the standard normal distribution cumulative function. The input variable of the standard normal distribution cumulative function is the specific kinetic energy of the current fragment. Dynamic limit penetration ratio kinetic energy of structural unit The difference is then divided by the standard deviation parameter, which characterizes the randomness of the penetration process. This function outputs a probability value in the interval [0,1], which is the theoretical breakdown probability. .

[0062] If we compare kinetic energy The dynamic limiting penetration specific kinetic energy is 1.2 MJ / m². The standard deviation is 0.1 MJ / m², with a value of 1.0 MJ / m². The calculated standardized variable is 2.0. Referring to the standard normal distribution table, the corresponding value is 0.977. It is 0.977.

[0063] The parameter symbol for determining the macroscopic interaction result is: This is a Boolean or enumerated variable whose value is "breakdown" or "non-breakdown", determined by Monte Carlo random sampling. Impact energy transfer coefficient, its parameter symbol is Defined in the interval (0,1), this parameter represents the proportion of kinetic energy of a fragment transferred to the target structural unit during a single impact interaction, thus generating internal stress waves. The calculation of this parameter is based on collision dynamics and the law of conservation of energy. The calculation of the impact energy transfer coefficient is related to the determination of macroscopic interaction results. Directly related. Specifically, it involves calculating the energy transfer under ideal conditions using an empirical model based on elastoplastic collision theory. In this embodiment, if the macroscopic interaction result is determined... If the impact is determined to be "non-breakdown", the impact energy transfer coefficient is set to a higher offline calibration value of 0.8. If the impact is determined to be "breakdown", the impact energy transfer coefficient is set to a lower offline calibration value of 0.3, in which case most of the energy is carried away in the form of residual kinetic energy of the fragments. In this embodiment, the complete calculation process for macro-micro coupled impact damage assessment is as follows: 2.1) The initial inputs are: the kinematic parameters of the current fragment related to "mass, velocity, and shape", the geometric and material properties of the target structural unit with which it will interact, and the damage mode correction factor updated by the preceding impact, which is read from the dynamic damage state map. .

[0064] 2.2) Processing Step 1: Calculate the specific kinetic energy of the current fragment. Calculate the dynamic limit of penetration specific kinetic energy. The logical steps are as follows: The baseline equivalent target thickness is calculated based on the original material properties and geometric dimensions of the target structural unit in an intact state. A preset material equivalent model is invoked to calculate the baseline equivalent target thickness, with the parameter notation as follows: This value represents the overall ballistic resistance of the structural unit in its initial state. From the dynamic damage state map, read the damage mode correction factor calculated and updated by the preceding impact event. Using the damage mode correction factor The baseline equivalent target thickness is corrected to obtain a dynamic equivalent target thickness that reflects the current damage state; its parameter symbol is . The specific correction method is as follows: Dynamic equivalent target thickness Equal to the reference equivalent target thickness Multiply by damage mode correction factor .

[0065] Based on dynamic equivalent target thickness Recalculate the dynamic limit penetration ratio kinetic energy ; The dynamically equivalent target thickness, which has been weakened by damage, calculated in the previous step. As the core input parameter, it is substituted into a preset ballistic physics model (including an analytical calculation function based on the Recht-Ipson model) that characterizes the interaction between the fragment and the target plate. The dynamic limit penetration kinetic energy of this structural unit under the current damage state against a specific fragment is then recalculated and obtained. The parameter symbol is... .

[0066] Will and Substitute the data into the probability penetration model to calculate the theoretical breakdown probability. Generate a uniformly distributed random number in the interval [0,1].

[0067] Logical Judgment 1: Compare "uniformly distributed random numbers" with the theoretical breakdown probability. If the "uniformly distributed random number" is less than or equal to the theoretical breakdown probability. Then the macroscopic interaction result is determined. If it is "breakdown", then it is "non-breakdown"; otherwise, it is "non-breakdown".

[0068] 2.3) Processing Step Two: Judgment Based on Macro Interaction Results The judgment result determines the impact energy transfer coefficient of this interaction. The effective impact energy transferred to the structural unit is calculated, and its value is equal to the current fragment kinetic energy multiplied by... .

[0069] 2.4) Step 3: Using the effective impact energy calculated in the previous step as the energy source, initialize and run the stress wave propagation and acoustic emission characteristic analysis model. Extract and calculate the brittle damage dominance from the virtual acoustic emission signal output by the model. .

[0070] 2.5) Step 4: Calculate the brittle damage dominance... Substituting a pre-defined mapping function, the new damage mode correction factor is calculated. This function, calibrated through offline experiments, describes the impact of different damage modes on future ballistic performance.

[0071] Logical Judgment 2: If the macroscopic interaction result is determined If the result is "penetration", the remaining velocity and mass of the fragment are calculated, and the process is redirected to the next interactive structural unit. If the result is "non-penetration", the penetration test for that fragment is terminated.

[0072] 2.6) The final output is: the macro-level interaction result determination for this interaction. And an updated damage mode correction factor prepared for the next interaction with this structural unit. .

[0073] The core innovation of this embodiment lies in the damage mode correction factor. Dynamic generation and application, Damage mode correction factor With brittle damage dominance Mapping function relationship between It was calibrated through a series of precise offline experiments and simulations; the calibration process is as follows: Two rounds of impact tests were conducted on the same batch of material samples. The first round used impacts with different energies to create different degrees of brittle damage dominance. The pre-damaged samples were then subjected to a second round of standard ballistic limit testing to measure their dynamic limit penetration kinetic energy. Finally, by fitting the "brittle damage dominance of pre-damage", the following was determined. "Value" and "Dynamic Limit Penetration Specific Kinetic Energy Measured in the Second Round" The mapping function is obtained by finding the data points between the "rate of change".

[0074] Further implementation details of the above content are provided below: When damage mode correction factor When the value is greater than 1.0, it indicates that the structural unit has a beneficial effect of work hardening or stress redistribution due to the preceding impact, and its equivalent strength against subsequent impacts is enhanced. When damage mode correction factor When the value is equal to 1.0, it indicates that its ballistic performance remains unchanged; When damage mode correction factor The closer the value is to 0, the more severe the internal damage, mainly brittle fracture, caused by the preceding impact on the structural unit, the more deteriorated its elastic resistance and the lower its resistance to subsequent impacts. When the theoretical breakdown probability The closer it is to 1, the higher the probability that the fragment will successfully penetrate the target structural unit under the current conditions; When the theoretical breakdown probability The closer it is to 0, the lower the probability that the fragment will penetrate the target structural unit.

[0075] Fragility damage dominance Damage mode correction factor The influence exhibits a predefined, non-monotonic, nonlinear relationship; this design aligns with the physical reality of composite material damage mechanics. Initially, the damage is dominated by matrix microcracks, resulting in ductile damage (brittle damage dominance). (A lower value) By releasing localized stress concentration, the material exhibits stronger macroscopic resistance; at this point, the damage mode correction factor... Greater than 1. When brittle damage, which characterizes large-scale fiber fracture, becomes dominant, the material's load-bearing capacity drops precipitously, leading to a significant decrease in the damage mode correction factor. The damage decreases dramatically. This nonlinear design accurately maps the entire physical process of a material from damage tolerance to catastrophic failure.

[0076] Damage mode correction factor Dynamic limit penetration specific kinetic energy The effect is a direct positive correlation, damage mode correction factor. It is a direct quantitative correction to the current elastic performance state of the material. Therefore, when the damage mode correction factor... When the energy is increased, it means the material is more "robust," and its dynamic limit penetration ratio is higher than its kinetic energy. It should also be improved accordingly.

[0077] With other parameters remaining constant, the dynamic limit of penetration ratio kinetic energy Theoretical breakdown probability The effect is negatively correlated; it indicates that for a fixed energy impact, the probability of successfully overcoming it is lower.

[0078] To verify the effectiveness of the "macro-micro coupling" mechanism of this invention, the following comparative experiment was designed. The experimental object was the same composite material structural unit, characterized by its exposure to two consecutive impacts. The "comparative scheme" adopted existing technology, with a constant damage correction factor of 1.0, and did not consider feedback from micro-damage modes. The "inventive scheme" employed a coupling feedback mechanism. By comparing the evolution differences in the theoretical breakdown probabilities calculated by the two schemes after two impacts, the significant progress of this invention in predicting cumulative damage effects was demonstrated.

[0079]

[0080] Comparing the data from Scenario 1 (ductile pre-damage): the first low-energy impact (0.80 MJ / m²) produced a lower brittle damage dominance of 0.25 in the present invention's scheme, which is interpreted as ductile damage dominance. Based on the nonlinear mapping relationship, the calculated damage mode correction factor is 1.15, greater than 1.0, characterizing the work hardening effect of the material. This results in the dynamic limiting penetration kinetic energy (1.15 MJ / m²) calculated by the present invention for the second impact being higher than that of the comparative scheme (1.00 MJ / m²), ultimately leading to a significantly lower theoretical breakdown probability (0.36) for the second impact compared to the comparative scheme's 0.69. This data strongly demonstrates that the present invention can capture and quantify the temporary performance enhancement effect brought about by a specific damage mode (ductile damage), which is completely impossible to obtain by the existing simple linear damage accumulation model, proving the physical authenticity and advancement of the present invention.

[0081] Comparing the data from Scenario 2 (brittle pre-damage): the first high-energy impact (1.50 MJ / m²) resulted in a high brittle damage dominance of 0.85 in the present invention's scheme, interpreted as severe fiber fracture. This led to the damage mode correction factor being calculated as 0.30, much less than 1.0. Therefore, when faced with a second impact (1.10 MJ / m²) of the same energy as in Scenario 1, the dynamic limit penetration ratio kinetic energy of the present invention's scheme plummeted to 0.30 MJ / m², making the theoretical breakdown probability approach 1.00. In contrast, the comparative scheme, failing to detect this catastrophic internal damage, still calculated a breakdown probability of 0.69, the same as in Scenario 1. This data demonstrates the core innovation of the present invention—the ability to provide early warning of catastrophic failure. This invention, through feedback of microscopic damage patterns, can accurately predict the drastic structural changes caused by the accumulation of brittle damage. Its prediction results are highly consistent with physical reality, while existing technologies seriously underestimate this risk, with their prediction results having an error of (1.00-0.69) / 0.69≈45%.

[0082] Comparing the second impact scenarios in the two scenarios, the impact energies were exactly the same (1.10 MJ / m²), but due to the different damage patterns characterized by the pre-damage history, the solution of this invention gave drastically different breakdown probability predictions. The comparative solution, lacking awareness of the damage history, gave the exact same incorrect prediction of 0.69.

[0083] Further explanation: The material property degradation coefficient is calculated by calling a pre-trained proxy model; The surrogate model not only outputs the material property degradation coefficient, but also simultaneously outputs the prediction confidence level, which characterizes the reliability of the output; Calculate the risk adjustment factor based on the prediction confidence level; The theoretical breakdown probability calculated based on the material property degradation coefficient is corrected using a risk adjustment factor to obtain the corrected damage probability. "Correction" is achieved by multiplying the theoretical breakdown probability by a risk adjustment factor in the probability space.

[0084] The following are specific implementation instructions for the above content: The core technical feature of this embodiment lies in constructing and applying an uncertainty-aware surrogate model trained through deep physical simulation. This surrogate model can not only quickly infer the macroscopic material property degradation coefficient from microscopic damage characteristics, but also simultaneously output the prediction confidence level of this inference. This prediction confidence level information is used as a dynamic risk adjustment factor to intelligently adjust the results in subsequent damage probability calculations. When the surrogate model has a low confidence level in its predicted material property degradation coefficient, the system automatically introduces a conservative penalty term to moderately amplify the calculated theoretical breakdown probability. This mechanism enhances the robustness and safety of the entire simulation system when facing extreme or unknown conditions.

[0085] The key parameters involved in this embodiment are defined as follows: Internal damage feature vector, whose parameter symbol is: This is the output of the upstream material microscopic damage mechanism model, with each dimension corresponding to different quantified microscopic damage indices, specifically including the defined fiber integrity index. Matrix integrity index and interlayer adhesion integrity index .

[0086] The equivalent ballistic performance degradation coefficient, whose parameter symbol is: ; is a dimensionless scalar predicted by the surrogate model and defined in the interval [0,1], used to directly correct the macroscopic ballistic resistance index; its physical meaning is the ratio of the residual ballistic resistance performance of the structural unit to the original performance.

[0087] Prediction confidence, its parameter symbol is It is a dimensionless scalar defined in the interval [0,1], synchronously output by the surrogate model, used to quantify the surrogate model's predictions. The reliability of the value. A value of 1 indicates high confidence, representing a high degree of agreement between the input features and the training data; a value close to 0 indicates low confidence, representing the encounter of an "outside-the-domain" impairment pattern that the model has never seen before. This parameter is obtained based on the machine learning model, namely a Bayesian Neural Network (BNN) or Gaussian Process Regression (GPR) model. These models are essentially probabilistic models, and their output includes not only a point estimate but also the probability distribution of that estimate, including variance or standard deviation.

[0088] This embodiment uses a Gaussian process regression model as an example, when the internal damage feature vector is input... At that time, the model outputs a value about The Gaussian distribution is defined by two parameters: mean and variance. The mean is directly used as... The point estimate. Variance reflects the uncertainty of the model at that input point. Prediction confidence. The variance is calculated using the inverse transform function. Specifically, the output variance is first normalized (divided by the maximum baseline variance determined through cross-validation) to obtain the normalized uncertainty. Then, the prediction confidence is calculated. The value is equal to 1.0 minus the value of the normalized uncertainty.

[0089] If the model output The predicted mean is 0.6, and the variance is 0.01. The maximum baseline variance is 0.25, so the normalized uncertainty is 0.01 divided by 0.25, which is 0.04. Therefore, the prediction confidence level is... Subtracting 0.04 from 1.0 gives 0.96.

[0090] Risk adjustment factor, its parameter symbol is Defined in the interval [1.0, +∞), it is used to adjust the theoretical breakdown probability according to the prediction confidence level; the calculation of this risk adjustment factor is a design function based on risk management theory, which aims to transform the uncertainty of the model into a safety redundancy at the system level; Risk adjustment factor Based on prediction confidence The monotonically decreasing function is calculated; the specific implementation method is, Equal to the baseline risk coefficient (1.0 minus the "prediction confidence level") ") to the power of. When the prediction confidence When the value is 1.0, (1.0 minus the prediction confidence) ) is 0, This is the baseline coefficient raised to the power of 0, i.e., 1.0. When the prediction confidence level... When it approaches 0, (1.0 minus the prediction confidence) It approaches 1.0. The risk factor approaches its maximum value, i.e., the baseline risk coefficient. In this embodiment, the baseline risk coefficient is set to 1.5. If the prediction confidence level The value is 0.96, and the benchmark risk coefficient is 1.5. It equals 1.5 raised to the power of (1.0 minus 0.96), which is 1.5 raised to the power of 0.04, and the result is approximately 1.016.

[0091] The corrected damage probability, whose parameter symbol is: It is after "risk adjustment factor" The adjusted damage probability value is defined in the interval [0,1]. In this embodiment, the complete calculation process is as follows: 3.1) The initial input is the internal damage feature vector from the upstream material micro-damage mechanism model. And the uncorrected theoretical penetration probability from the macroscopic ballistics calculation module. .

[0092] 3.2) Processing Step 1: Transfer the internal damage feature vector As input, a pre-trained uncertainty-aware surrogate model based on Gaussian process regression is fed in; from the output of the surrogate model, two parameters are obtained simultaneously: the equivalent ballistic performance degradation coefficient. The predicted mean and variance are calculated. The predicted mean is directly used as the equivalent ballistic performance degradation coefficient. The final value; based on the predicted variance, the corresponding prediction confidence level is calculated. .

[0093] 3.3) Processing Step Two: The equivalent ballistic performance degradation coefficient obtained in the previous step is... The value is written into the material performance state parameter field of the corresponding target structural unit in the dynamic damage state map, so that all subsequent calculation modules that need this parameter can call it.

[0094] 3.4) Processing Step Three: Calculate the obtained prediction confidence level. Substituting the values ​​into the calculation function for the risk adjustment factor, we obtain the risk adjustment factor. .

[0095] 3.5) Step 4: When calculating the damage probability of subsequent impact events, first call the updated equivalent ballistic performance degradation coefficient in the dynamic damage state map. The value is used to correct the equivalent target thickness, which is included in the ballistic performance index of the target structural unit; the corrected equivalent target thickness is equal to the original thickness multiplied by the "equivalent ballistic performance degradation coefficient". ".

[0096] Recalculate the theoretical penetration probability using the revised ballistic resistance index. Theoretical breakdown probability The calculation results are used for final risk adjustment; the corrected damage probability The calculation method is as follows: Calculate The probability, that is Divide by (1.0 minus) Then, multiply that probability by the risk adjustment factor. This yields the risk-adjusted odds. Finally, converting the risk-adjusted odds back to probability form, i.e., dividing the adjusted odds by (1.0 plus the adjusted odds), gives the result. This multiplicative adjustment in the probability space ensures that the final probability value remains within the [0,1] interval.

[0097] 3.6) The final output is the corrected damage probability. This probability value will be used for subsequent Monte Carlo sampling decisions or as the final risk assessment result.

[0098] The core of this embodiment is the "sequential risk fusion based on surrogate model confidence" mechanism. Its innovation lies in... Step 1: Fusion: Using a surrogate model, the internal damage feature vectors are... The equivalent ballistic performance degradation coefficient, which is integrated into a macroscopic performance index, is used to characterize the overall performance. The confidence level of its uncertainty prediction .

[0099] The second step of fusion: The equivalent ballistic performance degradation coefficient from the first step of fusion is... and its prediction confidence , with an uncorrected theoretical breakdown probability To integrate.

[0100] The specific steps for constructing the proxy model are as follows: The first step is to construct the training dataset and identify all the key input variables driving the high-fidelity physical simulation, including: Impact energy: range [100J, 5000J]; Impact angle: range [0°, 60°]; Fragment mass: range [2g, 50g]. Fragment shape factor: normalized range [0, 1]. The sampling strategy employs the Latin-Hypercube-Sampling (LHS) method to generate a set of input parameter combinations in the aforementioned multidimensional parameter space.

[0101] The second step is to use commercial explicit dynamic finite element analysis software (LS-DYNA or Abaqus / Explicit) to establish a simulation model that includes a detailed geometric model of the target structural element and accurate material constitutive relations.

[0102] For each combination of input parameters generated by LHS, perform the following "double-click" simulation: The first impact is used to induce damage: a first impact simulation is performed using the current parameter combination. After the simulation, the required internal damage feature vector is extracted from the results. The post-processing program calculates the fiber fracture element ratio, matrix equivalent plastic strain cloud map, etc., and quantifies them into a fiber integrity index. Matrix integrity index and interlayer adhesion integrity index .

[0103] The second impact simulation focuses on detection performance: based on the structural model damaged by the first impact, a second standardized "detective" impact simulation is performed. The purpose of this simulation is to measure the "limiting penetration velocity" of the damaged structure. Through a series of detection simulations, the velocity at which the structure can just penetrate is found, and from this, the true, damaged, equivalent ballistic performance degradation coefficient is calculated. .

[0104] The input "internal damage feature vector" obtained from each "double-click" simulation "and output equivalent ballistic performance degradation coefficient" Store it as a data pair.

[0105] The third step is to check the simulation results and remove simulation data points that cause calculation divergence or unreasonable results due to problems such as mesh distortion. Specifically, this involves checking all input "internal damage feature vectors". "and output label "Equivalent ballistic performance degradation coefficient" "Min-Max Scaling is performed to scale all values ​​to the [0,1] interval to improve the stability and convergence speed of subsequent model training."

[0106] The fourth step is to select a Gaussian-Process-Regression (GPR) model. A kernel function is chosen for the GPR model; in this embodiment, the Matérn-5 / 2-ARD kernel with automatic correlation determination is selected.

[0107] The fifth step is to conduct the following training process: Dataset partitioning: The generated total training dataset is randomly divided into a training set (1600 datasets) and a test set (400 datasets) at a ratio of 80 / 20; the ratio of training set to test set is 6:4. The training set data is fed into the configured GPR model. The training algorithm automatically optimizes all hyperparameters of the kernel function, including length scale and signal variance, by maximizing the log-marginal-likelihood function.

[0108] The sixth step is to validate using a test set that was never used in the training. This involves transforming the internal damage feature vector... Input the trained model to obtain the predicted "equivalent ballistic performance degradation coefficient". The predicted "equivalent ballistic performance degradation coefficient" is calculated. "and the actual "equivalent ballistic performance degradation coefficient" The coefficient of determination (R²) and mean absolute error (MAE) between the two values ​​are used to determine the relationship between the two values. The goal is to make R² greater than 0.98.

[0109] For each point in the test set, the model outputs not only the predicted mean but also the predicted standard deviation. Based on this, a 95% confidence interval is calculated. The number of true values ​​in the test set that fall within their corresponding 95% confidence interval is then counted. If this proportion equals 95%, it indicates that the model's estimation of uncertainty is calibrated.

[0110] During the validation process, the maximum value of all prediction variances is recorded and set as the value used to calculate the prediction confidence level. The "maximum baseline variance" is determined. After model validation, all optimized parameters and structure are serialized and saved as a callable file.

[0111] When the damage probability is corrected The closer the value is to 1, the higher the probability that a subsequent impact event will successfully penetrate the target structural unit. This not only reflects the strength of the impact energy, but more importantly, it may incorporate the significant degradation of structural performance caused by historical damage, as well as risk compensation for the uncertainty in the surrogate model's prediction of the current complex damage state. (Damage probability approaching 1 after correction) The value is the direct quantitative basis for triggering the highest level of early warning for the risk of structural chain failure.

[0112] When the damage probability is corrected The closer the value is to 0, the lower the probability that a subsequent impact event will successfully penetrate the target structural unit; this is due to insufficient impact energy, or that the structural unit has only minor historical damage and the surrogate model has high confidence in its performance degradation assessment. (The corrected damage probability is close to 0.) The value provides reliable positive evidence for assessing the survivability and remaining lifespan of a structure under continuous impact.

[0113] The probability of damage after correction affects the final output. The core input parameters include the equivalent ballistic performance degradation coefficient. and prediction confidence .

[0114] When other parameters remain constant, the equivalent ballistic performance degradation coefficient The reduction will lead to a decrease in the corrected damage probability. Monotonically increasing, indicating a negative correlation; equivalent ballistic performance degradation coefficient It is directly used to correct the equivalent target thickness of the structural unit. Equivalent ballistic performance degradation coefficient. The smaller the value, the more severe the historical damage, and the lower the corrected ballistic resistance index. In the laws of ballistic physics, a decrease in the ballistic resistance index directly leads to a decrease in the theoretical penetration probability. The increase. Due to the corrected damage probability. Based on theoretical breakdown probability The adjusted coefficient is therefore the equivalent ballistic performance degradation factor. The reduction ultimately leads to the corrected damage probability. The increase.

[0115] When other parameters remain constant, the prediction confidence level The reduction will lead to a decrease in the corrected damage probability. Monotonically increasing, indicating a negative correlation. Prediction confidence level. Used to calculate risk adjustment factor Prediction confidence level The smaller the value, the higher the uncertainty of the proxy model regarding its own prediction results. In algorithm design, the risk adjustment factor... With prediction confidence Inversely proportional, meaning that lower confidence levels will produce a larger risk moderating factor. Value. This risk adjustment factor. This was subsequently used to amplify the "probability" of the theoretical breakdown probability. Therefore, the prediction confidence was reduced. The value will be adjusted by the increased risk moderating factor. This value ultimately leads to an increased probability of damage after correction. This negative correlation design reflects the engineering safety principle that "a more conservative risk assessment should be adopted when the model is uncertain." The following series of computational experiments were designed to characterize the damage probability assessment process under different historical damage states and model uncertainties. The experimental data were obtained from multiple rounds of impact tests on composite laminates, and were intended to compare the method of this invention with the benchmark method that only considers performance degradation but does not have confidence assessment.

[0116]

[0117] Comparing the data from Scenario 1 and Scenario 2, it can be seen that due to severe historical brittle damage, the equivalent ballistic performance degradation coefficient of Scenario 2 (0.25) is much lower than that of Scenario 1 (0.82), causing its theoretical penetration probability to soar from 0.21 to 0.78. This strongly demonstrates the basic feasibility of correctly transmitting historical damage and influencing subsequent damage assessment in the S4-S5 linkage mechanism of this invention.

[0118] The comparison of data from Scenario 2 and Scenario 3 verifies the core innovation of this invention. The equivalent ballistic performance degradation coefficient of Scenario 3 (0.45) is actually better than that of Scenario 2 (0.25), therefore its theoretical penetration probability (0.55) is also lower than that of Scenario 2 (0.78). However, since Scenario 3 represents an "unknown complex damage" that the model has not fully learned, the prediction confidence level given by the surrogate model is a low value of 0.60, far lower than the 0.95 of Scenario 2. This triggers the risk adjustment mechanism of this invention, generating a risk adjustment factor as high as 1.181. Ultimately, the corrected damage probability output by this invention (0.608) is significantly increased, even higher than its theoretical penetration probability. This demonstrates the unique advantage of this invention in automatically switching to a conservative evaluation mode when facing model uncertainty, avoiding underestimation of risk due to model defects. In contrast, the benchmark method without a risk adjustment mechanism would incorrectly consider Scenario 3 to be safer than Scenario 2.

[0119] Comparing the output of this invention (0.608) in scenario three with the output of the benchmark method (0.55), the risk assessment result of this invention is improved by 10.5%. This increase is not arbitrary, but rather due to the model's low "prediction confidence". Driven by this technology, this intelligent and dynamic risk compensation capability is something that existing technologies that rely solely on deterministic prediction lack. This makes the evaluation results of this invention more robust and reliable under various operating conditions. Experimental data (repeated experiments in scenarios four, five, and six) also show the consistency of the results, proving the stability of the mechanism.

[0120] Based on expert experience and analysis of extensive computational experimental data, the following three-level risk interval classification criteria were established:

[0121] The core technical action of this invention, namely "based on microscopic damage assessment, real-time updating and application of material performance degradation coefficient", achieves the following two closely related technical effects: Effect 1 is the linkage and coupling between microscopic mechanisms and macroscopic phenomena: the microscopic damage mechanism (internal damage feature vector) at the material science level and the macroscopic penetration phenomenon (equivalent ballistic performance degradation coefficient) at the ballistic level are quantitatively coupled in real time and in both directions; the degree of microscopic "internal damage" directly "controls" the value of macroscopic ballistic resistance parameters. The second effect is the correlation between historical state and future events in predicting the system's "historical" damage state (the "equivalent ballistic performance degradation coefficient" value stored in the dynamic graph) can directly "control" and influence the simulation results for "future" impact events. This historical dependence enables this method to predict and quantify the nonlinear, emergent system behavior of "failure acceleration" caused by damage accumulation.

[0122] Effect three is high-fidelity prediction of chain failure: based on real-time updates of macroscopic ballistic parameters to microscopic damage, it can accurately predict the actual damage process in scenarios of "swarm" attacks or near-miss fragmentation. It can express scenarios where the first fragment causes large-area delamination of composite materials (a decrease in the equivalent ballistic performance degradation coefficient), leading to the second fragment penetrating critical components with extremely low energy cost; it can also automatically adopt a more conservative assessment when facing uncertain or unknown damage accumulation, thereby issuing early warnings for potential and unexpected chain failure risks.

[0123] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale; dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for calculating the probability of helicopter damage, characterized in that, The specific steps include: S1: Initialize a dynamic damage state map characterizing the three-dimensional structural model of the helicopter. The dynamic damage state map discretizes the three-dimensional structural model into multiple structural units and associates an initial material property state parameter with each structural unit. S2: When a preset impact event interacts with the target structural unit in the three-dimensional structural model, the macroscopic impact parameters of the interaction are calculated. S3: Based on the macroscopic impact parameters, a preset material microscopic damage mechanism model is invoked to quantitatively evaluate the degree of internal damage to the target structural unit caused by this interaction; the material microscopic damage mechanism model is a stress wave propagation and acoustic emission characteristic analysis model; S4: Based on the internal damage degree of the quantitative assessment, calculate and generate a material performance degradation coefficient characterizing the change in the elastic resistance of the target structural unit, and use the material performance degradation coefficient to update the material performance state parameters of the corresponding target structural unit in the dynamic damage state map; The material performance state parameters are multimodal damage state vectors, including exponential components that characterize the integrity of fibers, matrix and interlayer adhesion, respectively. The exponential components are calculated and determined by the normalized brittle fracture energy ratio, ductile deformation energy ratio, and shear wave attenuation factor, respectively. S5: When subsequent impact events interact with structural units in the three-dimensional structural model, damage probability is calculated based on the updated material property state parameters in the dynamic damage state map.

2. The helicopter damage probability calculation method according to claim 1, characterized in that: The equivalent ballistic performance degradation coefficient is obtained by weighted geometric mean fusion of each exponential component of the multimodal damage state vector. The weights used in the weighted geometric mean fusion are dynamically and adaptively determined based on the threat characteristic parameters of the current impact event.

3. The helicopter damage probability calculation method according to claim 2, characterized in that: The range of the equivalent ballistic performance degradation coefficient is defined as the interval [0,1]. When the equivalent ballistic performance degradation coefficient approaches 1.0, the material's ballistic resistance performance, which characterizes the structural unit, approaches its original design state. When the equivalent ballistic performance degradation coefficient is closer to 0, it indicates that the structural unit suffers more severe internal damage accumulation and the macroscopic ballistic performance degradation is greater.

4. The helicopter damage probability calculation method according to claim 3, characterized in that: The steps for calculating macroscopic impact parameters include: calculating the theoretical breakdown probability and using the Monte Carlo method to determine the macroscopic interaction result; The theoretical breakdown probability is calculated based on the specific kinetic energy of the fragment and the dynamic limit penetration specific kinetic energy of the structural unit.

5. The helicopter damage probability calculation method according to claim 4, characterized in that: The microscopic quantitative assessment step outputs the degree of brittle damage dominance characterizing the damage mode; Based on the brittle damage dominance, a damage mode correction factor is calculated; When calculating the dynamic limit penetration ratio kinetic energy, the damage mode correction factor obtained from the previous interactive calculation is used for correction.

6. The helicopter damage probability calculation method according to claim 5, characterized in that: When the value of the damage mode correction factor is greater than 1.0, it indicates that the structural unit has a beneficial effect of work hardening or stress redistribution due to the preceding impact, and its equivalent strength against subsequent impacts is enhanced. When the value of the damage mode correction factor is equal to 1.0, it indicates that its ballistic performance remains unchanged; When the value of the damage mode correction factor is closer to 0, it indicates that the structural unit has more severe internal damage due to the preceding impact, its ballistic performance is more deteriorated, and its resistance to subsequent impacts is lower.

7. The helicopter damage probability calculation method according to claim 6, characterized in that: When the theoretical breakdown probability is closer to 1, it indicates that under the current conditions, the probability of the fragment successfully penetrating the target structural unit is higher. The closer the theoretical breakdown probability is to 0, the lower the probability that the fragment will penetrate the target structural unit.

8. The helicopter damage probability calculation method according to claim 7, characterized in that: The degradation coefficient of material properties is calculated by calling a pre-trained proxy model; The proxy model outputs the material property degradation coefficient and simultaneously outputs the prediction confidence level, which characterizes the reliability of the output. Calculate the risk adjustment factor based on the prediction confidence level; The theoretical breakdown probability calculated based on the material property degradation coefficient is corrected using the aforementioned risk adjustment factor to obtain the corrected damage probability.

9. The helicopter damage probability calculation method according to claim 8, characterized in that: The closer the corrected damage probability is to 1, the higher the probability that the subsequent impact event will successfully penetrate the target structural unit. The closer the corrected damage probability is to 0, the lower the probability that a subsequent impact event will successfully penetrate the target structural unit.

Images