A rapid damage assessment method for typical engineering targets based on image recognition technology

Abstract

This invention belongs to the field of damage assessment and relates to a rapid damage assessment method for typical engineering targets based on image recognition technology. It establishes damage assessment criteria for different categories of image damage features: Damage images and blast damage data are obtained, damage levels are classified, and a P-I curve is plotted. Damage levels and critical damage states are defined. Circular interpolation is used to obtain the critical damage condition, critical progressive impulse, and critical progressive overpressure, and these are substituted into the classic P-I curve expression for fitting to obtain the critical damage curve expression. Further simulation is performed to obtain the critical damage area, and a corresponding relationship between the critical damage area and damage level is established for the image damage feature damage assessment criterion for the engineering target category. The damage image is captured, the damaged area is extracted, and the damaged area is compared with the image damage feature damage assessment criterion to rapidly determine the damage level. By establishing a mapping relationship between image damage features and structural mechanical responses, the present invention achieves rapid damage assessment.

Description

Technical Field

[0001] The present invention belongs to the technical field of weapon damage assessment, and in particular relates to a method for rapid damage assessment of typical engineering targets based on image recognition technology. Background Art

[0002] As a key component of modern military decision-making, weapon damage assessment (WDA) provides operational commanders with decision-making support through scientific, quantitative, and dynamic analysis. This technology not only helps commanders understand the actual effectiveness of combat operations in real time but also provides an engineering basis for optimizing and adjusting subsequent strike strategies. At the technical implementation level, the pressure-impulse curve method (PI curve method), with its precise characterization of structural explosion resistance, has become one of the most valuable quantitative criteria for engineering applications in the field of blast damage assessment.

[0003] In the practice of military strike and disaster assessment, the current PI curve method has exposed significant technical bottlenecks for critical infrastructure composed of typical concrete components such as beams, slabs, and columns. Although the predictive model constructed based on mechanical parameters such as deflection-span ratio, structural rotation angle, and concrete spalling thickness has demonstrated excellent guidance in pre-war strike plan planning, its reliance on precise parameter measurement has limited its practical application in post-war damage assessment scenarios. Especially in complex battlefield or post-disaster environments, the acquisition of key parameters faces severe challenges due to two core contradictions: 1) the contradiction between the security and timeliness of sensor network deployment; and 2) the contradiction between the feasibility and accuracy of manual inspection of hazardous areas. Traditional assessment methods can no longer meet the dual requirements of modern warfare and emergency rescue for immediate situational awareness and reliable assessment results.

[0004] It is worth anticipating that with the breakthrough development of cutting-edge technologies such as computer vision, drone swarms, and intelligent analysis of multi-source heterogeneous data, real-time panoramic imaging and multidimensional data collection in battlefields and post-disaster environments have become possible. Against this backdrop, the question of how to deeply mine the damage characteristics of damaged targets' image data, construct damage assessment methods based on these characteristics, and thus achieve non-contact rapid damage assessment has become a pressing technological challenge in the fields of military engineering and emergency management. This research direction not only involves a paradigm shift in assessment methods but also holds significant strategic value for improving battlefield response speed and reducing the risk of casualties. Summary of the Invention

[0005] The purpose of the present invention is to address the above-mentioned shortcomings in the prior art and to propose a rapid damage assessment method for typical engineering targets based on image recognition technology to achieve rapid and efficient damage assessment.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A typical engineering target rapid damage assessment method based on image recognition technology includes the following steps:

[0008] According to the typical engineering target category, the corresponding image destruction feature damage assessment criterion is established. The specific method is as follows: the damage picture and anti-blast damage data of the engineering target are obtained, the damage level is divided and the PI curve is drawn, and different damage levels and damage critical states are defined according to the degree and characteristics of explosion damage; the fixed overpressure scatter points, fixed impulse scatter points, limit overpressure scatter points and limit impulse scatter points are respectively subjected to the cyclic interpolation method to obtain the critical damage condition, critical progressive impulse and critical progressive overpressure, and are substituted into the classic PI curve expression for fitting to obtain the expression of the damage critical curve; according to the expression of the damage critical curve, the new damage critical condition is simulated to obtain damage pictures and anti-blast damage data, and the damage pictures of each new damage critical condition are identified by the image destruction feature recognition program to obtain the damage area; the damage areas of the same type of damage critical conditions are averaged to obtain the damage critical area; the image destruction feature damage assessment criterion under the engineering target category is established, and the image destruction feature damage assessment criterion is the corresponding relationship between the damage critical area and the damage level;

[0009] Take pictures of the damage to the engineering target after it is destroyed;

[0010] Input the damage picture into the image damage feature extraction program to extract the damage area;

[0011] Determine damage assessment criteria for image destruction features corresponding to the photographed engineering target according to the typical engineering target category;

[0012] The extracted damage area is compared with the corresponding image destruction feature damage assessment criteria to quickly determine the damage level of the photographed engineering target, thus achieving rapid damage assessment of the engineering target.

[0013] Preferably, the cyclic interpolation method comprises the following steps:

[0014] S1-1. Find two explosion conditions with different damage levels and closest to the critical damage level, and record them as blasting condition a and blasting condition b respectively;

[0015] S1-2. Take the numerical average of the overpressure P and impulse I of blasting condition a and blasting condition b to obtain a new blasting condition, which is recorded as blasting condition n;

[0016] S1-3. Perform numerical simulation or experimental research on blasting condition n to obtain damage images and anti-blast damage data. Classify the damage level of blasting condition n based on the anti-blast damage data and plot the damage on a PI curve.

[0017] S1-4. If the damage level of blasting condition n is the same as that of blasting condition a, blasting condition a is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if the damage level of blasting condition n is the same as that of blasting condition b, blasting condition b is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if blasting condition n falls on the critical damage condition, then the blasting condition n at this time is the critical damage condition.

[0018] Preferably, the blasting conditions include: selecting a fixed overpressure and different impulses that increase in sequence as blasting conditions; selecting a fixed impulse and different overpressures that increase in sequence as blasting conditions; selecting an extreme overpressure and different impulses that increase in sequence as blasting conditions; selecting an extreme impulse and different overpressures that increase in sequence as blasting conditions.

[0019] Preferably, the explosion damage resistance data include: deflection-span ratio, structural rotation angle and spalling thickness.

[0020] Preferably, the damage levels are divided into slight damage, moderate damage, severe damage and complete damage.

[0021] Preferably, the critical state of destruction includes moderate destruction criticality, severe destruction criticality and complete destruction criticality. The critical state between mild destruction and moderate destruction is defined as moderate destruction criticality, the critical state between moderate destruction and severe destruction is defined as severe destruction criticality, and the critical state between severe destruction and complete destruction is defined as complete destruction criticality.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] The innovative breakthrough of this invention lies in the construction of a damage assessment framework based on "image damage characteristics and mechanical response." By applying sophisticated image recognition technology, damage characteristics are identified and extracted from images of typical engineering targets. The extracted results are compared with preset damage assessment standards. By establishing a mapping relationship between image damage characteristics and structural mechanical response, rapid assessment of the damage level of engineering targets is achieved. This method overcomes the physical limitations of data acquisition in complex battlefield environments, enabling rapid and secure non-contact assessments. This method can improve military decision-making efficiency and reduce personnel risks. It provides key technical support for the formation of a closed-loop intelligent combat chain of "reconnaissance-assessment-strike." Patent protection is urgently needed to promote technological transformation and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a typical PI curve diagram in the prior art. Figure 1 (a) is a schematic diagram of the PI curve. Figure 1 (b) is the PI curve classification diagram;

[0025] Figure 2 A flowchart of establishing a damage assessment criterion for image destruction features according to an embodiment of the present invention;

[0026] Figure 3 Flowchart of a rapid damage assessment method based on image recognition according to an embodiment of the present invention;

[0027] Figure 4 This is a damage picture of the experimental results of the cross beam-slab composite structure according to an embodiment of the present invention;

[0028] Figure 5 This is a damage picture of the simulation result of the cross beam-slab composite structure according to an embodiment of the present invention;

[0029] Figure 6 The fixed overpressure scatter plot and the fixed impulse scatter plot of the simulation results of the cross beam-slab composite structure according to the embodiment of the present invention are shown;

[0030] Figure 7 The ultimate overpressure scatter plot and the ultimate impulse scatter plot of the simulation results of the cross beam-slab composite structure according to the embodiment of the present invention are shown;

[0031] Figure 8 is a time history curve of the failure area of ​​the cross beam-slab composite structure according to an embodiment of the present invention;

[0032] Figure 9 is a time history curve of the deflection-span ratio of the cross beam-slab composite structure according to an embodiment of the present invention;

[0033] Figure 10 This is a damage assessment criterion diagram for image destruction characteristics of a cross beam-slab composite structure according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0035] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0036] Typical PI curves are as follows: Figure 1 As shown in Figure 2, its core mechanism lies in constructing a correlation response model between the overpressure value and impulse value of the explosion load. Figure 1The PI curve shown in (a) divides the coordinate plane into two characteristic regions through the critical damage curve: when the explosion parameter falls to the lower left of the curve, the degree of damage to the component is lower than the preset damage threshold; when the explosion parameter is to the upper right of the curve, it means that the component structure has suffered substantial damage exceeding the expected value. Figure 1 (b) The corresponding PI curve divides the area into multiple regions, corresponding to different degrees of damage to the component. Each region corresponds to a degree of damage, namely mild damage, moderate damage, severe damage, and complete damage. The overpressure and impulse corresponding to the blast load fall in a certain region. By determining the region where the explosion parameters are located, the damage level can be accurately calibrated. This graded assessment mechanism significantly improves the engineering applicability of damage prediction.

[0037] The core of the rapid damage assessment method for typical engineering targets provided by the embodiments of the present invention is to establish damage assessment criteria based on image damage features. The overall concept is to conduct experiments or numerical simulations for different explosion conditions to obtain damage images and damage levels under these explosion conditions, thereby establishing a correspondence between the damage images and damage levels. However, due to the high cost of explosion experiments and the significant time and computing power required for explosion simulations, it is difficult to simulate all blasting conditions in the entire PI diagram. Therefore, the embodiments of the present invention adopt a computational strategy that "selects some blasting conditions, obtains critical damage conditions through cyclic interpolation, fits these critical damage conditions to obtain a curve expression for the critical damage conditions, and uses this expression to find more critical damage conditions."

[0038] like Figure 2 As shown in the figure, the specific steps to establish the image destruction feature damage assessment criterion are as follows:

[0039] S1. Determine the type of engineering target to be studied. Typical engineering target categories, such as beams, slabs, columns, and walls, are commonly classified in blasting science.

[0040] For example, the engineering target category of the embodiment of the present invention is a cross beam-slab composite structure.

[0041] S2. Select different blasting conditions to conduct explosion-resistant experiments or simulations on the engineering target, obtain damage images and explosion-resistant damage data under the blasting conditions, classify the blasting conditions into damage levels based on the explosion-resistant damage data and plot them on the PI diagram; the critical state between different damage levels is called the "critical damage state", and the blasting condition corresponding to the critical damage state is called the critical damage condition.

[0042] The “different blasting conditions” include the following four types: ① Selecting a fixed overpressure and different impulses that increase in sequence as blasting conditions; ② Selecting a fixed impulse and different overpressures that increase in sequence as blasting conditions; ③ Selecting an extreme overpressure and different impulses that increase in sequence as blasting conditions; ④ Selecting an extreme impulse and different overpressures that increase in sequence as blasting conditions. The extreme overpressure refers to an extremely high overpressure value. Under this overpressure condition, the impulse that destroys the critical condition tends to an impulse constant, which is called the “critical progressive impulse I”. sn "The critical condition of failure is drawn on the PI diagram as follows Figure 1 As shown in (a), when the overpressure increases infinitely, the impulse I gradually approaches a constant value; when the impulse increases infinitely, the overpressure P gradually approaches a constant value. The limit overpressure / limit impulse refers to taking an extremely large value of P or I.

[0043] The limit impulse refers to an extremely high impulse value. Under this impulse condition, the overpressure that destroys the critical condition tends to the overpressure constant value. The overpressure constant value is called the "critical progressive overpressure P". sn In this embodiment, the fixed overpressure is 0.22 MPa, the fixed impulse is 11 MPa×ms, the ultimate overpressure is 0.3 MPa, and the ultimate impulse is 60 MPa×ms. The critical conditions for failure are moderate, severe, and complete. Therefore, the critical progressive impulse and critical progressive overpressure are also divided into moderate, severe, and complete.

[0044] Blasting conditions are typically described by overpressure P and impulse I. Overpressure P refers to the difference between the pressure on the shock wave front generated by the explosion and the ambient pressure. Impulse I is the integral of the shock wave pressure over time, representing the total effect of the shock wave on the object. For explosion-resistant experiments and simulations, overpressure P and impulse I can be set by adjusting factors such as the explosive charge and blasting distance.

[0045] The simulation refers to establishing an engineering target model, setting blasting conditions and performing numerical simulation of the target blasting process in blasting simulation software (such as Ansys-dyna, Abauqs, Autodyn), which is a commonly used method in this field.

[0046] The damage picture can be obtained by photographing the destruction surface of the experimental engineering target, or by directly exporting the destruction surface image of the engineering target of the simulation model. Figure 4 、 Figure 5 The damage pictures of the experimental and simulation results of the cross beam-slab composite structure are shown respectively.

[0047] The blast damage data includes the deflection-to-span ratio, structural rotation angle, and spalling thickness. These parameters are commonly used in the blasting field and can be obtained through simple measurement and calculation based on experimental results. The specific calculation methods are well known in the art and will not be detailed here.

[0048] In addition, damage images and explosion-resistant damage data can also be obtained from public papers, patents and other materials.

[0049] In this embodiment, the damage levels are divided into four levels: light damage, moderate damage, severe damage and complete damage. The specific meanings of the definitions of the damage levels are as follows: ① Light damage: A small amount of cracks or local deformations appear on the surface of the engineering target, but it does not affect the overall bearing capacity and safety of the structure, and only requires simple repairs to continue using; ② Moderate damage: Obvious cracks, deformations or local damage appear on the surface of the engineering target, but the overall structure still has good bearing capacity and safety, and can be repaired and continued to be used in a short time; ③ Severe damage: Part of the structure of the engineering target has obvious damage, large deformations or a large number of cracks, and loses its bearing capacity. Part of the structure has a certain degree of bending damage, but the structure still has a certain integrity and bearing capacity, and there is a possibility of repair. ④ Complete damage: The engineering target structure completely loses its bearing capacity, the overall structure completely fails, and it is difficult to continue using, and it completely loses its repair value.

[0050] An explanation of the damage levels is given here. However, the actual judgment of which level a particular explosion result belongs to is not based on manual observation to see which description the engineering target meets. Instead, the damage level can be rigorously determined by calculating the blast damage data (deflection-span ratio, structural rotation angle, and spalling thickness) based on the explosion results and comparing this data with the existing damage level comparison table.

[0051] The critical damage state is specifically defined as follows: ① "Moderate damage criticality" is defined as the critical state between mild damage and moderate damage; ② "Severe damage criticality" is defined as the critical state between moderate damage and severe damage; and ③ "Complete damage criticality" is defined as the critical state between severe damage and complete damage. Correspondingly, the critical damage conditions are also divided into moderate damage criticality, severe damage criticality, and complete damage criticality.

[0052] The aforementioned "classification of blast conditions into damage levels based on blast damage data" is a commonly used method in the blasting field. A large amount of public information in this field provides comparison tables between blast damage data and damage levels for typical engineering targets, or discloses experimental methods for establishing comparison tables. For example, Table 2 in "Study on Damage Criteria for Reinforced Concrete Beams Under Explosive Loads," Acta Armamentarii, Vol. 37, No. 8, Aug. 2016, shows the correspondence between deflection-span ratio and damage level. The damage level can be determined by comparing the deflection-span ratio, structural rotation angle, and spalling thickness obtained in this experiment or simulation with the comparison table. Generally, a larger deflection-span ratio and a larger spalling thickness generally indicate more severe damage. For example, Table 1 below shows the damage levels corresponding to different deflection-span ratios for a cross-beam-slab composite structure. A deflection-span ratio of 0-7.6% is considered mild damage, 7.6%-20.9% is considered moderate damage, 20.9%-30.8% is considered severe damage, and 30.8% or above is considered complete damage.

[0053] Table 1: Comparison table of deflection-span ratio and damage level of cross beam-slab composite structure

[0054]

[0055] The specific steps for selecting different blasting conditions are as follows: ① Select a fixed overpressure P0 and different impulses I1, I2...I n , respectively (P0, I1), (P0, I2)... (P0, I n ) is used as the blasting condition for explosion resistance experiment or simulation, and the set of these blasting conditions is recorded as "fixed overpressure scatter points". ② Select a fixed impulse I0 and different overpressures P1, P2, P3, P4, P5, P6, P7, P8, P9, P10, P11, P12, P13, P14, P15, P16, P17, P18, P19, P20, P21, P22, P23, P24, P25 2······ P n , respectively (P1, I0), (P2, I0) ······ (P n , I0) is the blasting condition for anti-blast experiment or simulation, and the set of these blasting conditions is recorded as "fixed impulse scatter point". ③ Select a very high limit overpressure P max and different successively increasing impulses I1, I 2······ I n , respectively (P max , I1), (P max , I2) ······ (P max , I n ) is used as the blasting condition for anti-blast experiments or simulations, and the set of these blasting conditions is recorded as "limit overpressure scatter points". ④ Select a limit impulse I with a very high value max and different overpressures P1, P 2······ P n , respectively (P1, Imax )、(P2,I max ) ······ (P n , I max ) is used as the blasting condition for anti-blast experiments or simulations, and the set of these blasting conditions is recorded as "limit impulse scatter points".

[0056] Figure 6 The PI diagram shows two types of blasting conditions, calculated through simulation: fixed overpressure P0 and fixed impulse I0, and their corresponding damage levels. Each scattered point is represented by the horizontal and vertical axes PI, indicating its blasting condition, and the damage level is represented by different colors and shapes. Figure 6 The horizontally arranged row of scattered points in the figure are fixed overpressure scattered points, which are the simulation results of selecting a fixed overpressure P0 of 0.22 MPa and different impulses I as the blasting conditions. Figure 6 The vertically arranged column of scattered points in the figure represents fixed-impulse points. These are simulation results for a fixed impulse, I0, of 11 MPa×ms and varying overpressures, P, as the blasting conditions. It can be seen that, under the same overpressure, a larger impulse results in more severe damage; and, under the same impulse, a larger overpressure results in more severe damage.

[0057] Figure 7 The PI diagram shows the ultimate overpressure P obtained through simulation calculation. max and the limit impulse I max Two types of blasting conditions and corresponding damage levels. Figure 7 The horizontally arranged row of scattered points in the upper left corner is the limit overpressure scattered points, which is the limit overpressure P max The simulation results of the blasting conditions with an overpressure of 0.3 MPa and different impulses I are shown in the figure. By observing the three curves, we can see that under this overpressure, the curves of the three different critical conditions of destruction all tend to be vertical (the horizontal axis I tends to be fixed), that is, when the overpressure value is extremely high, the impulse of the critical condition of destruction tends to a constant value. Figure 7 The vertically arranged column of scattered points in the lower right corner is the limit impulse scatter point, which is the limit impulse I max The simulation results of blasting conditions with a velocity of 60 MPa×ms and different overpressures P show that when the impulse value is extremely high, the overpressure that destroys the critical condition approaches a constant value.

[0058] S3. Use the cyclic interpolation method to obtain the critical conditions of failure for the fixed overpressure scattered points, fixed impulse scattered points, ultimate overpressure scattered points and ultimate impulse scattered points respectively; the impulse of the critical conditions of failure under ultimate overpressure is recorded as the critical asymptotic impulse I sn The overpressure that destroys the critical condition under the limit impulse is recorded as the critical progressive overpressure P sn .

[0059] Since the critical damage conditions include "moderate / severe / complete", the "obtaining the critical damage conditions" means: executing each of the above three critical damage conditions once.

[0060] The cyclic interpolation method comprises the following steps:

[0061] S1-1. Among the specified type of scatter points, find two explosion conditions with two different damage levels that are closest to the critical damage threshold. These conditions are labeled "blasting condition a" and "blasting condition b," respectively. The scatter points are selected from the group consisting of fixed overpressure scatter points, fixed impulse scatter points, extreme overpressure scatter points, and extreme impulse scatter points.

[0062] S1-2. Take the numerical average of P and I of the above two blasting conditions to obtain a new blasting condition, which is recorded as blasting condition n.

[0063] S1-3. Perform numerical simulation on blasting condition n, obtain damage images and blast damage data, classify the damage level of the blasting condition based on the blast damage data, and plot the damage level on the PI diagram. Blasting condition n must be located midway between blasting conditions a and b.

[0064] S1-4. If the destruction level of blasting condition n is the same as that of blasting condition a, blasting condition a is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if the destruction level of blasting condition n is the same as that of blasting condition b, blasting condition b is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if blasting condition n happens to fall on the critical destruction condition, the execution is terminated, and the blasting condition n at this time is the critical destruction condition.

[0065] For example, for a fixed overpressure scatter point:

[0066] S2-1. Find two explosion conditions with different damage levels and closest to the critical damage level, respectively denoted as (P0, I a ) and (P0, I b ).

[0067] Taking the search for the critical condition of moderate damage as an example, find two adjacent blasting conditions in the PI diagram, where blasting condition a (P0, I a ) belongs to light damage and blasting condition b (P0, I b ) is moderate damage, then (P0, I a ) and (P0, I b ) are the two blasting conditions closest to the moderate destruction critical point among the fixed overpressure scattered points.

[0068] S2-2, take the average of the values ​​of the above two blasting conditions to obtain a new blasting condition n (P0, I new ).

[0069] From (P0, I a ), (P0, I b ), the overpressure of the two is the same. After taking the average of the values, the overpressure value of blasting condition n is still P0, and the impulse value is I new =(I a +I b ) / 2.

[0070] S2-3, blasting conditions n (P0, I new ) Conduct numerical simulation or experimental research to obtain damage pictures and anti-blast damage data, and classify the damage level of the blasting conditions according to the anti-blast damage data and draw them on the PI diagram.

[0071] S2-4, if the blasting condition n (P0, I new ) damage level and blasting condition a (P0, I a ) are the same, then replace the blasting condition a with (P0, I new ), re-execute operations S2-2 to S2-4; if the blasting condition n (P0, I new ) damage level and blasting condition b (P0, I b ) are the same, then replace the blasting condition b with (P0, I new ), re-execute operations S2-2 to S2-4; loop until the newly obtained blasting condition n falls exactly into the critical state of destruction, then the blasting condition n at this time is the moderate critical condition of destruction.

[0072] The search for critical conditions for severe damage and complete damage is similar to the above steps. Finally, three critical conditions for damage are obtained under a fixed overpressure P0.

[0073] The cyclic interpolation method is also performed on the fixed impulse scatter points, the limit overpressure scatter points and the limit impulse scatter points to find the three critical failure conditions for each type of scatter points.

[0074] Interpolation essentially involves finding the midpoint between two points. If the interpolated point falls to the left, interpolation is performed between the new point and the original point to the right. If the new point falls to the right, interpolation is performed between the new point and the point to the left, repeating until the new point falls exactly on the critical line. For fixed overpressure scattered points, interpolation is performed between the two points above and below, finding the midpoint, and repeating the process.

[0075] Table 2 below shows the calculation results of the critical conditions for failure at fixed impulse and fixed overpressure. The values ​​in the first row are the three critical conditions for failure at a fixed overpressure of 0.22 MPa. The values ​​in the second row are the three critical conditions for failure at a fixed impulse of 11 MPa×ms. The values ​​in the third row are the three critical conditions for failure at an ultimate overpressure of 0.30 MPa; and the values ​​in the fourth row are the three critical conditions for failure at an ultimate impulse of 60 MPa×ms. The moderate critical progressive impulse is 1.80 MPa×ms, the severe critical progressive impulse is 2.10 MPa×ms, and the fully critical progressive impulse is 3.00 MPa×ms. The moderate critical progressive overpressure is 0.210 MPa, the severe critical progressive overpressure is 0.215 MPa, and the fully critical progressive overpressure is 0.219 MPa.

[0076] Table 2: Calculation results of critical conditions for failure with fixed impulse and fixed overpressure

[0077]

[0078] This method, based on the principle of "selecting scattered points closest to the critical condition and performing cyclic interpolation," can quickly identify scattered points on each critical failure condition curve with a minimal number of simulations. After finding at least two scattered points of the blasting condition for each critical failure condition and the critical progressive overpressure and critical progressive impulse of the curve, a curve fitting method can be used to derive the curve expression for the critical failure condition, significantly reducing experimental or simulation costs.

[0079] S4. Substitute the above critical conditions for destruction, critical progressive overpressure, and critical progressive impulse into the classical PI curve expression for fitting, and obtain the expression for the critical destruction curve. The classical PI curve expression is:

[0080]

[0081] Among them, P is the overpressure value, I is the impulse value, P sn is the critical progressive overpressure; I sn is the critical asymptotic impulse; α PI is the failure mode factor, which is related to the failure mode of blasting and can be directly determined according to the engineering target category, failure mode, etc. In this embodiment, α PI =0.8; parameter A PI and β PI are constants determined by fitting.

[0082] Specifically, the critical conditions of moderate damage (0.22, 16.50) and (0.23, 11.00) obtained in the previous step and the critical progressive overpressure P sn =0.210, critical progressive impulse I sn=1.80, and then put it into the classic PI curve expression for fitting, and we get the critical value of moderate damage A. PI =0.8,β PI =1.8. Similarly, the above operations are performed on the severe and complete damage critical conditions, and finally the PI curve expressions of the three damage critical conditions are obtained:

[0083]

[0084] S5. Based on the expression of the critical damage curve, select more critical damage conditions. Simulate the newly selected critical damage conditions to obtain damage images and blast damage data. Use a mature image damage feature recognition program to identify the damage images of each critical damage condition and obtain its damage area. Average the damage areas of the same critical damage condition to obtain the critical damage area.

[0085] Taking moderate damage criticality as an example, the specific steps for selecting more moderate damage criticality conditions are as follows: specify an overpressure range and a step size ΔP. Within this overpressure range, select an overpressure P at intervals of ΔP. Substitute each overpressure P into the curve expression for moderate damage criticality to calculate the corresponding overpressure I. This blasting condition (P, I) must be a moderate damage criticality condition.

[0086] The image damage feature recognition program can adopt existing open source programs, for example, image contour detection based on OpenCV can be adopted, and the damage area can be quickly calculated based on the image contour. Currently, a large number of relevant mature algorithms are available for use.

[0087] When simulating different (P, I) values ​​on the critical failure curve, the resulting damage areas will vary slightly due to numerical errors. Therefore, this step selects multiple different critical failure conditions on the critical failure curve. After simulation, the damage areas are identified and averaged. This average is used as the critical failure area, making the data more accurate and reliable. Critical failure areas include moderate failure, severe failure, and complete failure.

[0088] S6. Establish a damage assessment criterion for image destruction features under the engineering target category.

[0089] The damage assessment criterion for the image destruction feature is the correspondence between the critical area of ​​destruction and the damage level. The specific correspondence is as follows: when the damage area is lower than the critical area of ​​moderate damage, the current damage level is mild damage; when the damage area is between the critical area of ​​moderate damage and the critical area of ​​severe damage, the current damage level is moderate damage; when the damage area is between the critical area of ​​severe damage and the critical area of ​​complete damage, the current damage level is severe damage; when the damage area is higher than the critical area of ​​complete damage, the current damage level is complete damage.

[0090] Figure 8 、 Figure 9 The time history curves of the failure area and deflection-span ratio of the cross beam-slab composite structure are shown respectively. Both the failure area and the deflection-span ratio increase with time. Figure 10 This is the damage assessment criterion for the image destruction characteristics of the cross-beam-slab composite structure. The three horizontal dotted lines in the figure represent the critical area of ​​moderate damage, severe damage, and complete damage from bottom to top. As long as the damage area at any moment is known, the damage level at that moment can be quickly compared.

[0091] like Figure 3 As shown in the figure, after establishing the damage assessment criteria of image destruction features, the typical engineering target rapid damage assessment method based on image recognition technology includes the following steps:

[0092] S01. Take pictures of the damage to the engineering target after destruction.

[0093] This can be done through personnel or advanced drone photography.

[0094] S02. Input the damage image into an image damage feature recognition program to quickly obtain the damage area.

[0095] S03. Classify the photographed engineering target according to typical structural components such as beams, plates, columns, and walls, and determine the typical engineering target category corresponding to the photographed engineering target.

[0096] S04. Compare the extracted damage area with the image destruction feature damage criteria corresponding to the typical engineering target category to determine the damage level and achieve rapid damage assessment.

Claims

1. A typical engineering target rapid damage assessment method based on image recognition technology is characterized by: The following steps are involved: According to the typical engineering target categories, the corresponding image damage feature damage assessment criteria are established. The specific method is as follows: damage images and anti-blast damage data of the engineering target are obtained, damage levels are classified and PI curves are drawn, and different damage levels and critical damage states are defined according to the degree and characteristics of explosion damage; fixed overpressure scatter points, fixed impulse scatter points, extreme overpressure scatter points and extreme impulse scatter points are respectively subjected to cyclic interpolation method to obtain the critical damage conditions, critical progressive impulse and critical progressive overpressure, and substitute them into the classic PI curve expression for fitting to obtain the expression of the critical damage curve; based on the expression of the critical damage curve, new critical damage conditions are simulated to obtain damage images and anti-blast damage data, and the damage images of each new critical damage condition are identified through the image damage feature recognition program to obtain the damage area; The damage areas of the same critical damage condition are averaged to obtain the critical damage area; a damage assessment criterion based on image damage characteristics for the engineering target category is established, wherein the image damage assessment criterion is a correspondence between the critical damage area and the damage level; Take pictures of the damage to the engineering target after it is destroyed; Input the damage picture into the image damage feature extraction program to extract the damage area; Determine damage assessment criteria for image destruction features corresponding to the photographed engineering target according to the typical engineering target category; The extracted damage area is compared with the corresponding image destruction feature damage assessment criteria to quickly determine the damage level of the photographed engineering target.

2. The method for rapid damage assessment of typical engineering targets based on image recognition technology according to claim 1 is characterized in that: The cyclic interpolation method comprises the following steps: S1-1. Find two explosion conditions with different damage levels and closest to the critical damage level, and record them as blasting condition a and blasting condition b respectively; S1-2. Take the numerical average of the overpressure P and impulse I of blasting condition a and blasting condition b to obtain a new blasting condition, which is recorded as blasting condition n; S1-3. Perform numerical simulation or experimental research on blasting condition n to obtain damage images and anti-blast damage data. Classify the damage level of blasting condition n based on the anti-blast damage data and plot the damage on a PI curve. S1-4. If the damage level of blasting condition n is the same as that of blasting condition a, blasting condition a is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if the damage level of blasting condition n is the same as that of blasting condition b, blasting condition b is replaced by blasting condition n, and then S1-2 to S1-4 are re-executed; if blasting condition n falls on the critical damage condition, then the blasting condition n at this time is the critical damage condition.

3. The method for rapid damage assessment of typical engineering targets based on image recognition technology according to claim 2 is characterized in that: The blasting conditions include: selecting a fixed overpressure and different impulses that increase in sequence as the blasting conditions; selecting a fixed impulse and different overpressures that increase in sequence as the blasting conditions; selecting an extreme overpressure and different impulses that increase in sequence as the blasting conditions; selecting an extreme impulse and different overpressures that increase in sequence as the blasting conditions.

4. The method for rapid damage assessment of typical engineering targets based on image recognition technology according to claim 1 is characterized in that: The explosion-resistant damage data include: deflection-span ratio, structural rotation angle and spalling thickness.

5. The method for rapid damage assessment of typical engineering targets based on image recognition technology according to claim 1 is characterized in that: The damage levels are divided into light damage, moderate damage, heavy damage and complete damage.

6. The method for rapid damage assessment of typical engineering targets based on image recognition technology according to claim 5 is characterized in that: The critical state of destruction includes moderate destruction criticality, severe destruction criticality and complete destruction criticality. The critical state between mild destruction and moderate destruction is defined as moderate destruction criticality, the critical state between moderate destruction and severe destruction is defined as severe destruction criticality, and the critical state between severe destruction and complete destruction is defined as complete destruction criticality.

Images