A building target damage effect evaluation method based on multi-source information fusion

Abstract

The application relates to a building target damage effect evaluation method based on multi-source information fusion, which comprises the following steps: performing feature recognition on a building target explosion image, determining a broken area recognition area ratio and an initial blast point position; training a neural network model based on a building internal explosion damage sample library to obtain a building internal explosion damage effect prediction model; inputting a current blast point position into the building internal explosion damage effect prediction model to obtain a broken area prediction area ratio; comparing the broken area recognition area ratio and the broken area prediction area ratio, if the deviation of the two area ratios is less than a set threshold, the current blast point position is the final blast point position, the final blast point position is input into the building internal explosion damage effect prediction model to obtain a building component flexural span ratio; and determining a building overall damage grade according to the building component flexural span ratio. The application can quickly evaluate the overall damage grade and predict the internal component damage condition under a building internal explosion scene.

Description

Technical Field

[0001] This application relates to the field of damage assessment, and more specifically, to a method for assessing the damage effects on building targets based on multi-source information fusion. Background Technology

[0002] In high-tech environments, the kill chain—from reconnaissance, decision-making, mission allocation, strike, damage assessment, re-decision, and re-strike—is short, the pace of warfare is fast, and opportunities are fleeting. Damage assessment is an indispensable link in this kill chain. Traditionally, damage assessments are conducted based on single information sources such as images, test data, or prior models. However, images only provide information about the external damage effects of a target and cannot provide internal damage characteristics; the quantity and location of test data collection are limited, failing to provide a comprehensive and continuous characterization of damage effects; and prior models are incomplete and cannot provide predictions of damage effects across all scenarios. Therefore, damage assessments based on single information sources can no longer meet the comprehensive, efficient, and accurate requirements of modern warfare for evaluating the energy conversion effects of explosive detonations. Summary of the Invention

[0003] To overcome at least one deficiency in the prior art, this application provides a method for evaluating the damage effect of building targets based on multi-source information fusion.

[0004] Firstly, a method for assessing the damage effects of building targets based on multi-source information fusion is provided, including: Step 1: Perform feature recognition on the exploded image of the building target to determine the proportion of the identified fragmentation area and the initial explosion point location; use the initial explosion point location as the current explosion point location; Step 2: Train the neural network model based on the sample database of building internal blast damage to obtain a prediction model of building internal blast damage effects; Step 3: Input the current blast point location into the building internal blast damage effect prediction model to obtain the predicted area ratio of the fracture zone; Step 4: Compare the identified area ratio of the broken area with the predicted area ratio of the broken area. If the deviation between the two area ratios is less than the set threshold, the current explosion point location is the final explosion point location, and proceed to Step 5; otherwise, redetermine the explosion point location and return to Step 3. Step 5: Input the final blast point location into the building internal blast damage effect prediction model to obtain the deflection-span ratio of the building components; Step 6: Determine the overall damage level of the building based on the deflection-to-span ratio of the building components.

[0005] In one embodiment, determining the overall damage level of a building based on the deflection-to-span ratio of its building components includes: The damage level of a building component is determined based on its deflection-to-span ratio and the criteria for determining the damage level of the component. The damage level of a room is determined based on the damage level of the components and the damage level criteria for a single room. The overall damage level of the building is determined based on the damage level of the rooms and the criteria for overall building damage.

[0006] Secondly, a building target damage effect assessment device based on multi-source information fusion is provided to implement the above-mentioned method.

[0007] Compared with the prior art, this application has the following beneficial effects: Based on the external image of the building target after being hit, the prediction model of the damage effect of the building implosion components, and the building damage level classification standard, this application can quickly assess the overall damage level and predict the damage of internal components in the building implosion scenario. Attached Figure Description

[0008] This application can be better understood by referring to the description given below in conjunction with the accompanying drawings, which, together with the detailed description below, are incorporated in and form part of this specification. In the drawings: Figure 1 A flowchart of a method for assessing the damage effects of building targets based on multi-source information fusion is shown. Figure 2 The diagram shows an image illustrating the damage to the building target. Figure 3 A schematic diagram showing the proportion of the identified broken zone area is provided. Detailed Implementation

[0009] Exemplary embodiments of the present application will be described below with reference to the accompanying drawings. For clarity and brevity, not all features of the actual embodiments are described in the specification. However, it should be understood that many embodiment-specific decisions can be made in the development of any such actual embodiment to achieve the developer’s specific objectives, and these decisions may vary as the embodiments differ.

[0010] It should also be noted that, in order to avoid obscuring this application with unnecessary details, only the device structure closely related to the solution of this application is shown in the accompanying drawings, while other details that are not closely related to this application are omitted.

[0011] It should be understood that this application is not limited to the described embodiments by virtue of the following description with reference to the accompanying drawings. In this document, embodiments may be combined with each other, features may be substituted or borrowed between different embodiments, and one or more features may be omitted in one embodiment, where feasible.

[0012] This application provides a method for assessing the damage effect of building targets based on multi-source information fusion. Figure 1A flowchart illustrating a method for assessing the damage effects of building targets based on multi-source information fusion is shown. (See attached diagram.) Figure 1 The method mainly includes the following steps: Step 1: Perform feature recognition on the exploded image of the building target to determine the proportion of the identified fragmentation area and the initial explosion point location; use the initial explosion point location as the current explosion point location.

[0013] Here, a method for identifying typical building target damage features based on visible light images is used to identify and extract features such as the location, area, and number of structural damages. Simultaneously, the initial blast point location, i.e., the location of the explosion room and the relative position of the blast center within the explosion room, is provided.

[0014] Step 2: Train the neural network model based on the sample database of building internal explosion damage to obtain a prediction model of building internal explosion damage effects.

[0015] A numerical simulation model of building implosion was established using traditional simulation methods (and its effectiveness was verified). The propagation law of shock wave under implosion load and the damage mode of adjacent rooms were analyzed, laying the foundation for the construction of physical loss function in physical information neural network.

[0016] The first phase involves constructing a sample database of building implosion damage. First, the propagation patterns of the implosion shock wave are analyzed. A fluid-structure interaction numerical simulation model is established, incorporating the explosive, air domain, and building structure (including adjacent rooms, doorways, and windows). Simulations briefly analyze the formation and development of the shock wave and detonation products in the blast center room, and their propagation through reflection and diffraction at doorways, windows, and corridors. Second, numerical simulation studies under typical operating conditions are conducted. Combined with experimental data, the structural response and damage modes of adjacent rooms under different blast center parameters are analyzed. Finally, large-scale simulation studies are performed to generate a database of adjacent room component damage under different operating conditions. The input parameters are tentatively set as the TNT charge at the blast center (W), the blast center location coordinates (x0, y0, z0) (dimensionless representation can be considered), and the output data are the component deflection-to-span ratio and the predicted area of ​​the fracture zone. 80% of the database data is used for training, and 20% for validation, using random sampling for classification.

[0017] The second stage involves constructing a neural network model. Here, the neural network model can employ traditional deep learning-based neural network data. Through activation functions, optimizers, and learning rates, the neural network is trained and optimized to obtain a set of optimal parameters.

[0018] Step 3: Input the current blast point location into the building internal blast damage effect prediction model to obtain the predicted area ratio of the fracture zone.

[0019] Step 4: Compare the identified area ratio of the broken area with the predicted area ratio of the broken area. If the deviation between the two area ratios is less than a set threshold (e.g., 10%), then the current explosion point position is the final explosion point position, and proceed to Step 5. Otherwise, redetermine the explosion point position through a traversal algorithm and return to Step 3.

[0020] Step 5: Input the final blast point location into the building internal explosion damage effect prediction model to obtain the deflection-span ratio of the building components (i.e., the component damage result).

[0021] Step 6: Determine the overall damage level of the building based on the deflection-to-span ratio of the building components.

[0022] Specifically, the damage level of a component is determined by combining the component damage level criteria; the damage level of a room is determined by combining the component damage level with the individual room damage level criteria; and the overall damage level of the building is determined by combining the room damage level with the overall building damage criteria.

[0023] Here, the criteria for determining the damage level of components can be, for example: setting thresholds for different damage levels of components (not specifically limited here), and determining the damage level of components based on the relationship between the component's deflection-to-span ratio and the threshold; the criteria for determining the damage level of a single room can be: counting the number of components in a single room with a damage level of moderate or above, and determining the room's damage level based on the number (not specifically limited here); counting the number of rooms with a damage level of moderate or above, calculating the ratio of this number to the total number of rooms, and determining the overall damage level of the building based on the range of the ratio. Different overall damage levels of the building correspond to different ranges (not specifically limited here).

[0024] This embodiment uses external images of the building target after it is hit, a prediction model of the damage effect of the building implosion components, and the building damage level classification standard to quickly assess the overall damage level and predict the damage of internal components in the building implosion scenario.

[0025] The method of this application will be further described in detail below with reference to the accompanying drawings and preferred embodiments. Based on the above method, the damage effects of a 15kg TNT explosion inside a building can be predicted.

[0026] Figure 2 A schematic diagram of an exploded building target is shown. Figure 3 The diagram shows the percentage of the identified fragmented area. After image recognition, the total number of rooms is: 6. 7 2=84; Explosion room location: 2,6 (second floor, 6th room); Initial blast point: (0,0,z); Area percentage: 7.26%.

[0027] Based on the established neural network model, the damage effects on building components are predicted. Assuming the building structure is six stories high, with two rows of seven rooms per row, the input explosive charge mass is 20 kg, and the explosion center coordinates are 5,5 (the explosion room is located on the fifth floor, in the fifth room) and (0,0,0) (with the center of the room as the origin of the coordinate system). The input parameters are shown in Table 1.

[0028] Table 1 Input Parameter Table

[0029] The output results calculated by the model are shown in Table 2.

[0030] Table 2 Model Output Parameter Table

[0031] Setting the initial explosion point z-value to 0, the initial explosion point is (0,0,0). Based on the established sample library, a prediction model for the damage effect of building implosion components is formed through training with the PINNs algorithm. Substituting the initial explosion point (0,0,0) into the prediction model, the area proportion is 7.95%, (7.95%-7.26%) / 7.95%=9.5%, and 9.5%≤10%, that is, (0,0,0) is the determined final explosion point.

[0032] The final explosion point (0,0,0) was substituted into the building implosion component damage effect prediction model to calculate the damage results of the walls and floors of each adjacent room, as shown in Table 3.

[0033]

[0034] Based on the standards for classifying the damage levels of walls and panels, the damage levels of individual rooms are given in the table below.

[0035] Table 4 Damage Levels for Individual Rooms

[0036] The number of rooms with moderate or severe damage (i.e., the number of rooms with damage 1) is 10. 10 / 84 = 11.9%, which is between 10% and 30% of the rooms with minor damage. Therefore, the building is classified as slightly damaged.

[0037] This application also provides a building target damage effect assessment device based on multi-source information fusion, used to implement the building target damage effect assessment method based on multi-source information fusion of the aforementioned embodiments.

[0038] The above descriptions are merely various embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for assessing the damage effect of building targets based on multi-source information fusion, characterized in that, include: Step 1: Perform feature recognition on the exploded image of the building target to determine the proportion of the identified fragmentation area and the location of the initial explosion point; The initial explosion point location is taken as the current explosion point location; Step 2: Train the neural network model based on the sample database of building internal blast damage to obtain a prediction model of building internal blast damage effects; Step 3: Input the current blast point location into the building internal blast damage effect prediction model to obtain the predicted area ratio of the fracture zone; Step 4: Compare the identified area ratio of the broken zone with the predicted area ratio of the broken zone. If the deviation between the two area ratios is less than a set threshold, the current explosion point location is the final explosion point location, and proceed to Step 5; otherwise, redetermine the explosion point location and return to Step 3. Step 5: Input the final explosion point location into the building internal explosion damage effect prediction model to obtain the deflection-span ratio of the building components; Step 6: Determine the overall damage level of the building based on the deflection-to-span ratio of the building components.

2. The method as described in claim 1, characterized in that, The overall damage level of the building is determined based on the deflection-to-span ratio of the building components, including: The damage level of the building component is determined based on the deflection-to-span ratio of the building component and the damage level criterion. Based on the damage level of the components and the damage level criteria for individual rooms, the damage level of the room is determined. Based on the damage level of the rooms and the criteria for overall building damage, the overall building damage level is determined.

3. A device for assessing the damage effect of building targets based on multi-source information fusion, characterized in that, Used to implement the method of claim 1 or 2.

Images