A digitalized engineering safety emergency management method and system

Abstract

The present application relates to the technical field of engineering safety emergency management, and discloses a kind of digital engineering safety emergency management method and system, including emergency positioning conversion module, space distribution evaluation module, emergency crowded analysis module, dispatch time evaluation module, dispatch order optimization module and dispatch instruction generation module.The present application converts accident point and emergency team position into accident reference coordinates containing relative radius and relative azimuth;Based on the above parameters, the order vulnerability of thin-shell defect angle type and the proximity of the same shell and fan are calculated to generate the order vulnerability coefficient;And this is combined with the historical order table to build a single target effective concentration time model;Finally, the priority weight is used as the particle variable for iterative optimization, and the distance fine-tuning term is introduced to stabilize the decoding of the optimal weight to obtain the final single-dimensional dispatch order table.

Description

Technical Field

[0001] This invention relates to the field of engineering safety emergency management technology, and more specifically, to a digital engineering safety emergency management method and system. Background Technology

[0002] Currently, the field of digital twins for engineering safety and emergency response emphasizes that emergency systems must form a closed loop of perception → update → simulation → decision-making. However, the integrity, resolution, frequency, interoperability, and reliability of real-time data remain significant bottlenecks for the industry. Meanwhile, traditional static path solutions are no longer effective in handling smoke, obstacles, and dynamically changing environments. In the spatial positioning stage, ultra-wideband positioning errors under non-line-of-sight conditions typically exceed 30 centimeters. In real indoor construction sites, the two-dimensional root mean square error increases from 0.38 meters to 0.70 meters as the coverage area expands, and in dynamically switching scenarios, the maximum error can even reach 1.5 meters. When these positioning deviations are applied to complex construction sites such as water plants, due to the spatial non-uniformity of harmful gas emissions and environmental constraints such as pool boundaries, pump room perimeters, and temporary safety perimeters, the actual habitable locations of emergency rescue teams are not evenly distributed around the accident point, but are compressed into a few safety perimeters. This results in multiple emergency teams being extremely close to the accident point by radius, forming a very thin ring-like distribution pattern. Furthermore, the safety sectors that can be occupied by teams are often incomplete, resulting in significant unilateral bias and gaps in spatial orientation. When the distance radii of multiple emergency response teams are extremely small, and the system faces dynamic positioning perturbations at the level of 0.3 meters to 1.5 meters, the single-dimensional deployment priority table output by conventional algorithms is prone to order brittleness and chattering. For example, the ranking in the previous scheduling cycle may be frequently swapped or completely replaced in the next cycle. This frequent order chattering is not a generalization error, but directly translates into the cost of frequent switching of on-site rescue commands, severely weakening the stability of emergency dispatch and on-site execution efficiency. Summary of the Invention

[0003] This invention provides a digital engineering safety emergency management method and system, which solves the technical problems mentioned in the background art.

[0004] Firstly, a digital engineering safety emergency management system includes:

[0005] The emergency location conversion module is used to convert the acquired accident point location and the emergency team locations of each emergency team into accident reference coordinates that include relative radius and relative azimuth.

[0006] The spatial distribution assessment module is used to calculate the sequence vulnerability of the thin-shell corner-missing type based on the relative radius and the relative azimuth angle.

[0007] An emergency congestion analysis module is used to calculate the proximity of the same shell and the same fan among the various emergency groups, and combine it with the thin-shell corner-cut sequence vulnerability to generate the sequence vulnerability coefficient of each emergency group.

[0008] The scheduling time evaluation module is used to construct an effective assembly time model for a single target based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient.

[0009] The scrambling order optimization module is used to initialize the bias by introducing the order vulnerability coefficient with priority weight as the particle position variable, and to perform intelligent iterative optimization based on the effective assembly time model to obtain the optimal priority weight.

[0010] The scheduling instruction generation module is used to introduce a distance fine-tuning term into the optimal priority weight for stable decoding to obtain the final order, generate a dispatch order table based on the final order, and use the dispatch order table as the historical dispatch order table for the next cycle.

[0011] Secondly, a digital engineering safety emergency management method, applied in any one of the digital engineering safety emergency management systems described above, includes:

[0012] The obtained accident location and the emergency team locations of each emergency team are converted into accident reference coordinates that include relative radius and relative azimuth.

[0013] Based on the relative radius and the relative azimuth angle, calculate the sequence vulnerability of the thin-shell corner-missing type;

[0014] Calculate the proximity of the same shell and the same fan among the various emergency response teams, and combine it with the thin-shell corner-cut sequence vulnerability to generate the sequence vulnerability coefficient of each emergency response team.

[0015] Based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient, an effective assembly time model for a single target is constructed.

[0016] Using priority weights as particle position variables, the order vulnerability coefficient is introduced for bias initialization, and intelligent iterative optimization is performed based on the effective assembly time model to obtain the optimal priority weights.

[0017] The optimal priority weight is subjected to a distance fine-tuning term for stable decoding to obtain the final order. A deployment order table is generated based on the final order, and the deployment order table is used as the historical deployment order table for the next cycle.

[0018] The beneficial effects of this invention are as follows: By converting the accident point and emergency team location into accident reference coordinates that include relative radius and relative azimuth, under complex working conditions where positioning generates dynamic micro-disturbances and personnel spatial distribution exhibits extremely small radius differences, the effective assembly time model is modified by introducing thin-shell corner-type sequence vulnerability and local congestion. Then, the final deployment order is obtained by using an intelligent optimization engine with bias initialization and a stable decoding mechanism with distance fine-tuning terms. This invention effectively overcomes the problem of frequent jumps in deployment order induced by the superposition of extremely small spacing and positioning errors. Without the need to output complex route plans, it significantly reduces the cost of repeated switching of on-site emergency commands and greatly improves the overall stability and execution efficiency of multi-team collaborative scheduling. Attached Figure Description

[0019] Figure 1 This is a flowchart of a digital engineering safety emergency management system according to the present invention;

[0020] Figure 2 This is a schematic diagram of the spatial distribution of the engineering emergency response team with a thin shell and missing corner under the accident reference coordinate system of the present invention. Detailed Implementation

[0021] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0022] Example 1: As Figure 1 As shown, a digital engineering safety emergency management system includes:

[0023] The emergency location conversion module is used to convert the acquired accident point location and the emergency team locations of each emergency team into accident reference coordinates that include relative radius and relative azimuth.

[0024] The spatial distribution assessment module is used to calculate the sequence vulnerability of the thin-shell corner-missing type based on the relative radius and the relative azimuth angle.

[0025] An emergency congestion analysis module is used to calculate the proximity of the same shell and the same fan among the various emergency groups, and combine it with the thin-shell corner-cut sequence vulnerability to generate the sequence vulnerability coefficient of each emergency group.

[0026] The scheduling time evaluation module is used to construct an effective assembly time model for a single target based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient.

[0027] The scrambling order optimization module is used to initialize the bias by introducing the order vulnerability coefficient with priority weight as the particle position variable, and to perform intelligent iterative optimization based on the effective assembly time model to obtain the optimal priority weight.

[0028] The scheduling instruction generation module is used to introduce a distance fine-tuning term into the optimal priority weight for stable decoding to obtain the final order, generate a dispatch order table based on the final order, and use the dispatch order table as the historical dispatch order table for the next cycle.

[0029] Preferably, the acquired accident location and the locations of each emergency response team are converted into accident reference coordinates that include relative radii and relative azimuth angles, including:

[0030] A two-dimensional displacement vector is generated according to the following formula, and the relative radius and the relative azimuth angle are calculated:

[0031]

[0032]

[0033]

[0034] in, This represents the generated two-dimensional displacement vector; The x-coordinate of each of the aforementioned emergency response teams indicates its location. The x-coordinate of the accident point represents the obtained location of the accident point; The vertical coordinates of the emergency response teams indicate their locations. The vertical coordinate of the accident point represents the obtained location of the accident point; This indicates the relative radius included; This indicates the included relative azimuth angle; This represents the arctangent function.

[0035] The x-coordinate of an accident point is a parameter characterizing its horizontal position in a local Cartesian coordinate system. It can be collected using ultra-wideband positioning equipment, global navigation satellite system positioning equipment, or building information modeling (BIM) planar positioning systems.

[0036] The vertical coordinate of an accident point is a parameter characterizing its position in a local Cartesian coordinate system. It can be collected using ultra-wideband positioning equipment, global navigation satellite system positioning equipment, or building information modeling (BIM) plane positioning systems.

[0037] The x-coordinate of an emergency response team is a parameter representing the horizontal position of each team in a local Cartesian coordinate system. It can be collected using ultra-wideband positioning equipment, global navigation satellite system positioning equipment, or building information modeling (BIM) plane positioning systems.

[0038] The ordinate of an emergency response team is a parameter representing the vertical position of each team in a local Cartesian coordinate system. It can be collected using ultra-wideband positioning equipment, global navigation satellite system positioning equipment, or building information modeling (BIM) plane positioning systems.

[0039] The lateral displacement component is the difference between the horizontal coordinate of the emergency response team and the horizontal coordinate of the accident point. It is a parameter that characterizes the magnitude of the horizontal displacement of the emergency response team relative to the accident point.

[0040] The longitudinal displacement component is the difference between the longitudinal coordinate of the emergency response team and the longitudinal coordinate of the accident point. It is a parameter that characterizes the magnitude of the vertical displacement of the emergency response team relative to the accident point.

[0041] A two-dimensional displacement vector is a vector composed of lateral displacement components and longitudinal displacement components. It is a parameter that characterizes the two-dimensional spatial displacement of the emergency response team relative to the accident point.

[0042] The relative radius is the straight-line distance between the emergency response team and the accident point. It is a parameter calculated by squaring and taking the square root of the lateral and longitudinal displacement components.

[0043] The relative azimuth angle is the azimuth angle of the emergency response team relative to the accident point. It is a parameter that characterizes the spatial orientation of the emergency response team relative to the accident point, calculated by the longitudinal and lateral displacement components through the arctangent function.

[0044] In detail, a two-dimensional displacement vector is first generated through the horizontal and vertical displacement components. Then, the Cartesian coordinate features are transformed into polar coordinate features of relative radius plus relative azimuth. This transformation method directly provides a calculable basis for subsequent identification of thin-shell corner-deficient spatial structures. This is different from the traditional coordinate translation operation without a clear application target. At the same time, the arctangent function used to calculate the relative azimuth can cover the entire range from 0 to 360 degrees, avoiding the quadrant ambiguity caused by the ordinary arctangent function only being able to calculate from 0 to 180 degrees. For example, if an accident occurs at the construction pool of a water plant, the horizontal coordinate of the accident point is 50 and the vertical coordinate is 30. The horizontal coordinate of an emergency team is 45 and the vertical coordinate is 38. Its horizontal displacement component is -5 and its vertical displacement component is 8. The ordinary arctangent function will calculate a negative angle, while the arctangent function used in this scheme can directly obtain the correct azimuth angle of 127.93 degrees, ensuring the accuracy of the circumferential azimuth analysis and making the subsequent azimuth entropy and corner notch calculations have practical engineering significance.

[0045] In detail, the coordinates of all accident points and emergency response teams must be unified to a local Cartesian coordinate system specific to the construction site, with the unit being meters. If the coordinate sources are mixed, rigid body registration is required to map the original coordinates from different sources to this unified coordinate system. Coordinate data only accepts valid data with timestamps no earlier than the start of the current scheduling cycle minus one refresh cycle. The unit for calculating relative azimuth angles is uniformly radians, and all subsequent azimuth-related calculations will use this unit. The scheduling cycle for coordinate acquisition is preferably 2 seconds or 5 seconds, which can be adjusted according to the actual accuracy of the positioning equipment at the construction site. The acquisition accuracy of accident point coordinates must be controlled within 0.01 meters, and the acquisition accuracy of emergency response team coordinates must be controlled within 0.05 meters. If an emergency response team loses coordinates once in the current cycle, the last valid coordinates from the previous cycle will be used to fill the gap. If coordinates are lost for two consecutive cycles, the team will be directly pushed to the end of the subsequent scheduling priority. When there is only one emergency response team, the coordinate transformation calculation of the entire process will still be executed normally. When there is no emergency response team on site, the calculation process of this module will be terminated directly, and a signal that there is no scheduling object will be fed back to the system.

[0046] Preferably, the sequence vulnerability of the thin-shell notched type is calculated based on the relative radius and the relative azimuth angle, including:

[0047] The sequence fragility of the thin-shell notched type is calculated using the following formula:

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054] in, This indicates the fragility of the thin-shell, corner-missing sequence position; Indicates the average relative radius; Indicates the total number of emergency response teams; and Both represent the relative radii; Indicates radial dispersion; Indicates the maximum relative radius; Indicates the minimum relative radius; Indicates the first bias; Represents the directional distribution entropy; Indicates the total number of partitions; Indicates the percentage of the group; Indicates the second bias; Indicates the maximum gap ratio; and Indicates adjacent ascending azimuth angles; Represents pi; Indicates the minimum spacing; Indicates the third bias; represents the natural logarithm function; max represents the maximum value function; min represents the minimum value function.

[0055] The first bias is a very small numerical parameter set to avoid the denominator being zero in radial dispersion and subsequent related calculations. It is preferably 0.05 meters to accommodate the centimeter-level error of ultra-wideband positioning at the construction site. This value can effectively avoid the denominator being zero and will not affect the quantification accuracy of radial dispersion.

[0056] The total number of partitions is a parameter representing the number of regions in a circle that are equally divided to quantify the directional distribution characteristics. It is preferably set to 8 when the number of emergency response teams is no more than 8, and to 12 when the number of emergency response teams is greater than 8, to balance the granularity of directional partitioning and computational efficiency. When the number of teams is small, overly fine partitioning is meaningless; when the number of teams is large, increasing the number of partitions can improve the accuracy of directional distribution representation.

[0057] The second bias is a very small numerical parameter set to avoid the logarithmic term being zero in the calculation of azimuth spread entropy. It is preferably 10 to the power of -6, a very small value that effectively avoids meaningless logarithmic terms and does not affect the quantification result of the azimuth spread entropy.

[0058] The third bias is a very small numerical parameter set to avoid the denominator being zero in the calculation of the fragility of the thin-shell missing corner type. It is preferably 10 to the power of negative 6. This value is extremely small, which can effectively avoid the denominator being zero and will not change the relative magnitude and quantification trend of the fragility of the thin-shell missing corner type.

[0059] The average relative radius is obtained by taking the arithmetic mean of the relative radii of all emergency response teams, and is a parameter that characterizes the overall distance of the emergency response team relative to the accident point.

[0060] The maximum relative radius is the maximum value extracted from the relative radii of all emergency response teams, representing the farthest distance parameter of an emergency response team relative to the accident point.

[0061] The minimum relative radius is the minimum value extracted from the relative radii of all emergency response teams, representing the closest distance parameter between an emergency response team and the accident point.

[0062] Radial dispersion is a dimensionless parameter that characterizes the degree of dispersion of the relative radius distribution of emergency response teams. It is obtained by dividing the difference between the maximum and minimum relative radii by the sum of the average relative radius and the first bias.

[0063] The group ratio is the ratio of the number of emergency groups in each directional zone to the total number of emergency groups. It is a dimensionless parameter that characterizes the distribution ratio of emergency groups in each directional zone.

[0064] The azimuth distribution entropy is a dimensionless parameter that characterizes the uniformity of the azimuth distribution of emergency response teams. It is calculated by combining the proportion of teams, the total number of zones, and the second bias. The smaller the value, the more concentrated the azimuth distribution.

[0065] Ascending azimuth is an angle sequence parameter obtained by arranging the relative azimuths of all emergency response teams in ascending order.

[0066] The maximum gap ratio is the maximum value extracted by dividing the difference between adjacent ascending azimuth angles by the circumference angle. It is a dimensionless parameter that characterizes the size of the gap in the azimuth distribution of the emergency response team.

[0067] Minimum spacing is the minimum value extracted from the absolute value of the relative radius difference between any two emergency response teams, and is a parameter characterizing the minimum difference in relative radius between emergency response teams.

[0068] The sequence flutter vulnerability of thin-shell missing corner type is a dimensionless parameter that comprehensively characterizes the spatial distribution characteristics of thin-shell missing corner and the degree of sequence flutter by combining the maximum notch ratio, azimuth scattering entropy, radial dispersion, minimum spacing and various offsets. The larger the value, the higher the risk of sequence flutter.

[0069] In detail, unlike traditional range coefficients of variation, this method adapts to the thin-shell radius distribution characteristics of emergency response teams on construction sites. It introduces a second bias and normalizes it according to the total number of zones, transforming the non-uniform azimuth distribution into a dimensionless value of zero to one, accurately characterizing the missing corner feature. The maximum gap ratio uses a closed-loop calculation rule, calculating the gap between the last ascending azimuth angle and the first ascending azimuth angle by adding 360 degrees to the first ascending azimuth angle and then subtracting the last ascending azimuth angle, ensuring the completeness of the circumferential azimuth gap analysis. Furthermore, the thin-shell missing corner type sequence vulnerability uniquely incorporates the missing corner feature... Coupled with thin shell and sequence gap features, quantization achieves single-valued continuous representation of dual features, which can be directly passed to subsequent objective functions. For example, if there are 5 emergency response teams around a construction accident site in a water plant, with relative radii of 9.9, 10.0, 10.1, 10.0, and 10.2 meters and relative azimuth angles of 350 degrees, 355 degrees, 4 degrees, 10 degrees, and 18 degrees, the quantization system can calculate a very small radial dispersion, a large maximum gap ratio, and a small azimuth spread entropy, ultimately resulting in a large thin shell corner-type sequence vulnerability, accurately identifying the high-sequence flutter risk in this scenario.

[0070] In detail, the calculation method for the group proportion is to count the number of emergency groups in each equiangular azimuth zone and then divide it by the total number of emergency groups. The natural logarithm in the azimuth spread entropy calculation uses the natural constant as the base. The closure processing of ascending azimuth angles requires calculating the difference between the last and first ascending azimuth angles by adding 360 degrees to the first ascending azimuth angle and then subtracting the last ascending azimuth angle. The minimum spacing is obtained by iterating through all emergency groups, pairwise combining them, calculating the absolute value of the relative radius difference for each group, and then extracting the minimum value. (Thin-shell corner missing) The value of the sequence flutter is greater than or equal to 0 with no upper limit. The larger the value, the more obvious the thin-shell corner-missing feature of the emergency team, and the higher the risk of sequence flutter. When the relative radii of all emergency teams are the same, the radial dispersion is 0. At this time, the original formula is still used to calculate the sequence flutter of the thin-shell corner-missing type. This situation will make the flutter value larger and more accurately characterize the high-sequence flutter risk of pure thin-shell distribution. All orientation-related calculations are performed in radians, and the circumferential angle is 6.283185307 radians. All numerical results are retained to 6 decimal places during the calculation process.

[0071] Preferably, the proximity of the same shell and the same sector among the various emergency response teams is calculated, and combined with the thin-shell corner-canceling sequence vulnerability, a sequence vulnerability coefficient for each emergency response team is generated, including:

[0072] The proximity of the same shell and the same sector, and the order vulnerability coefficient are calculated using the following formula:

[0073]

[0074]

[0075]

[0076]

[0077]

[0078] in, Indicates the azimuth angle between emergency response teams; and Both represent the relative azimuth angles; This indicates the proximity of the components in the same casing and sector; Indicates the radial attenuation scale; Indicates the angular attenuation scale; Indicates the local fan cluster crowding degree; Indicates the local sequence gap coefficient; Indicates the fourth bias; This represents the order vulnerability coefficient; This represents an exponential function.

[0079] The radial attenuation scale is a scale parameter used to quantify the impact of relative radius differences on spatial proximity. It is preferably set to the larger of 0.5 meters and 0.15 times the average relative radius to accommodate the error magnitude of ultra-wideband positioning at construction sites. Simultaneously, it is dynamically adjusted in conjunction with the average relative radius to take into account the radius scale differences in different accident scenarios, ensuring the rationality of the quantification of the radial proximity component.

[0080] The angular attenuation scale is a scale parameter used to quantify the impact of differences in azimuth angles on spatial proximity. A value of 0.5236 radians is preferred because in engineering emergency scenarios, the angle definition of the same sector cluster is typically within a reasonable range of 30 degrees. This value accurately characterizes the degree of proximity in terms of azimuth and is suitable for the equiangular partitioning characteristics of a circle.

[0081] The fourth bias is a very small numerical parameter set to avoid the denominator being zero in the calculation of the ordinal fragility coefficient. It is preferably 10 to the power of -6. This value is extremely small, which can effectively avoid the case where the denominator is zero, and will not affect the relative size and quantification trend of the ordinal fragility coefficient, and the value logic of other biases is consistent.

[0082] The azimuth angle between emergency response teams is a parameter that characterizes the azimuth difference between two emergency response teams by calculating the difference in azimuth angle between any two emergency response teams and taking a reasonable angle value.

[0083] The radial proximity component is calculated from the difference in the relative radii of the two emergency response teams and is a dimensionless parameter characterizing the degree of radial proximity between the two teams.

[0084] The angular proximity component is calculated from the azimuth angle between the two emergency response teams and is a dimensionless parameter characterizing the degree of proximity between the two emergency response teams in terms of azimuth angle.

[0085] The proximity of the same shell and the same sector is obtained by multiplying the radial proximity component and the angular proximity component. It is a dimensionless parameter that comprehensively characterizes the degree of common proximity of the two emergency response teams in the radial and azimuth directions.

[0086] Local fan cluster congestion is a dimensionless parameter that characterizes the degree of congestion in the space surrounding a single emergency group by summing the proximity of the single emergency group to the same shell and fan of all other emergency groups.

[0087] The local sequence gap coefficient is obtained by normalizing the minimum relative radius difference between a single emergency response team and other teams, and is a dimensionless parameter characterizing the sequence stability of a single emergency response team.

[0088] The vulnerability-crowding product is obtained by multiplying the thin-shell corner-missing type sequence vulnerability by the local fan cluster crowding, and is a parameter that characterizes the degree of coupling between global vulnerability features and local crowding features.

[0089] The sequence fragility coefficient is a dimensionless parameter that characterizes the risk of sequence flutter caused by a single emergency response team. It is calculated by combining the product of fragility and crowding with the local sequence gap coefficient.

[0090] In detail, the minimum value between the relative azimuth angle difference and the circumferential angle minus the azimuth angle of the two emergency response teams is first taken to match the angular characteristics of the circumferential azimuth, avoiding unreasonable angles greater than 180 degrees. Then, the radial and angular proximity components are coupled into the same shell and same fan proximity degree through a double exponential decay method, accurately representing the same shell and same fan characteristics in the engineering scenario. Subsequently, the global thin-shell corner-deficient sequence vulnerability is combined with the local fan cluster congestion degree, and a saturated sequence vulnerability coefficient calculation method is introduced by introducing the local sequence gap coefficient, so that the coefficient value is limited to between 0 and 1 to avoid numerical distortion. For example, there are 4 emergency response teams with thin-shell corner-deficient distribution around the construction accident point of a water plant. The relative radius difference between team 1 and teams 2 and 3 is within 0.2 meters, and the azimuth angle is within 20 degrees. Through this method, it can be found that the same shell and same fan proximity degree and local fan cluster congestion degree of team 1 are relatively large. After combining the global vulnerability, the final sequence vulnerability coefficient is close to 1, accurately identifying the high sequence flutter risk of this team.

[0091] In detail, the base of the exponential operation in the calculation of the proximity of the same shell and the same fan is a natural constant. The unit of calculation for the azimuth angle between emergency groups is uniformly set to radians, consistent with the unit of the previous azimuth-related calculations. When the number of emergency groups is 1, the local fan cluster congestion degree and the sequence vulnerability coefficient are both set to 0. When the number of emergency groups is 2, the local fan cluster congestion degree of a single group is directly taken as the proximity of the same shell and the same fan between the two groups. The local sequence gap coefficient is taken as the normalized value of the relative radius difference between the two groups. The value range of the proximity of the same shell and the same fan is 0 to 1. The larger the value, the more obvious the same shell and the same fan characteristics of the two groups are. The value range of the sequence vulnerability coefficient is 0 to 1. The larger the value, the higher the risk of the group causing sequence flutter. The radial and angular proximity components adopt the logic of exponential decay because exponential decay can reflect the nonlinear characteristics of spatial proximity. The closer the distance and angle between the two groups, the slower the decay of the proximity component. The larger the value, the faster the decay. The smaller the value, the closer the decay. This is more in line with the actual spatial distribution of the engineering site. All numerical results in the calculation process are retained to 6 decimal places.

[0092] Preferably, based on the preset nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient, an effective assembly time model for a single target is constructed, including:

[0093] The effective assembly time model is calculated using the following formula:

[0094]

[0095]

[0096]

[0097]

[0098] in, and Both refer to the nominal assembly time; Indicates the relative radius; Indicates a uniform reference response speed; Indicates the grouping overlap coefficient; This indicates the current group ranking under any candidate order; This indicates that group seating has been assigned; This indicates the proximity of the components in the same casing and sector; Indicates the distance of the order rearrangement; Indicates the total number of emergency response teams; This indicates the candidate deployment ranking for each of the aforementioned emergency response teams; This indicates the historical deployment ranking in the historical deployment order table; This represents the effective assembly time model; This represents the order vulnerability coefficient; This represents the vulnerability penalty weight; Indicates the weight of the congestion penalty; This represents the time coefficient for sequential switching.

[0099] The unified reference response speed is a user-defined parameter used to calculate the theoretical assembly time of emergency response teams. It is preferably 1.5 meters per second. This value is suitable for the normal movement speed of emergency personnel at the construction site, closely matches the actual assembly efficiency of the project site, and accurately represents the theoretical assembly time when there are no additional obstacles.

[0100] The vulnerability penalty weight is a weight parameter used to quantify the impact of the ordinal vulnerability coefficient on the assembly time delay. It is a custom parameter, preferably configurable within the range of 0.2 to 0.8. It can be dynamically adjusted according to the severity of the thin-shell corner defects on site. The more obvious the defect, the larger the weight value, which can enhance the time penalty effect on high-fragility risk groups.

[0101] The congestion penalty weight is a weight parameter used to quantify the impact of the group overlap coefficient on the assembly time delay. It is a custom parameter, preferably configurable within the range of 0.1 to 0.5. It can be flexibly adjusted according to the congestion level of the field emergency team's fan clusters. The higher the congestion level, the larger the weight value, adapting to different spatial distribution scenarios.

[0102] The sequence reordering time coefficient is a parameter used to convert the dimensionless sequence reordering distance into actual time cost. It is a custom parameter, preferably configurable within a range of 5 to 20 seconds. This value closely matches the actual reordering time for transmitting and executing emergency instructions at the construction site, and can accurately quantify the on-site time cost of sequence reordering.

[0103] The nominal assembly time is a parameter obtained by dividing the relative radius of the emergency response team by the uniform reference response speed. It represents the theoretical assembly time of the emergency response team to the accident site under conditions of no interference.

[0104] The candidate order is an arrangement obtained by sorting the priority weights of the emergency response teams, representing the candidate schemes for the order in which the emergency response teams are deployed.

[0105] The grouping overlap coefficient is the average of the proximity of the current emergency group to all the emergency groups preceding it in the candidate order, representing the degree of overlap between the current emergency group and the fan cluster of the preceding group.

[0106] The candidate deployment ranking is the specific position of the emergency response team in the candidate order, representing the order in which the emergency response team is selected from the candidate deployment plans.

[0107] The historical deployment ranking is the specific position of the emergency response team in the historical deployment ranking table of the previous cycle, representing the actual deployment order of the emergency response team in the previous cycle.

[0108] The total ranking deviation is a parameter obtained by summing the absolute values ​​of the differences between the candidate deployment ranking and the historical deployment ranking of all emergency response teams. It represents the overall degree of deviation between the candidate order and the historical order.

[0109] The ranking rearrangement distance is a dimensionless parameter obtained by normalizing the sum of ranking deviations, and it characterizes the degree of relative deviation between the candidate order and the historical order.

[0110] The delay ratio is a coefficient obtained by adding the base value, the brittleness penalty, and the congestion penalty, and it represents the delay ratio of the emergency team assembly time.

[0111] The total assembly time is a parameter obtained by globally summing the nominal assembly times of all emergency response teams by their respective delay ratios, and it represents the total delayed assembly time of all emergency response teams.

[0112] The switching cost time is a parameter obtained by multiplying the order rearrangement distance by the order switching time coefficient, which represents the time cost of switching field instructions generated by replacing the historical order with the candidate order.

[0113] The effective assembly time model is a single-objective model obtained by adding the total assembly time to the switching cost time, and it represents the total effective time cost in the emergency dispatch process.

[0114] In detail, the nominal assembly time is calculated by dividing the relative radius by the unified reference response speed, which aligns with the linear assembly characteristics of emergency engineering sites without additional road network constraints, unlike traditional road network path time calculations. The group overlap coefficient is calculated by averaging the proximity of the current group to all preceding groups within the same shell and sector, avoiding a one-sided representation of proximity by a single group. The order reordering distance is normalized by dividing the sum of the ranking deviations by the product of the total number of emergency groups and the total number minus one, which can adapt to the quantification of order deviations under different numbers of emergency groups. The delay ratio transforms the brittleness and congestion penalties into a time delay ratio rather than a direct time summation, which is more in line with the time loss logic on site. At the same time, the introduction of the order switching time coefficient transforms the abstract deviation of order reordering into actual deviations. The model directly addresses the engineering problem of solving the cost of on-site command switching, and when there is no historical dispatch priority table in the first round of scheduling, the priority rearrangement distance is directly set to 0. The model can automatically degenerate into minimizing the assembly time after brittleness correction, without manual intervention, ensuring the automated operation of the process. For example, at a water plant construction accident site, there are 6 emergency teams, 3 of which are distributed with thin shells and high order fragility coefficients, and are also in the same sector cluster, resulting in congestion. Through this model, the brittleness and congestion characteristics of these 3 teams can be transformed into the delay ratio of assembly time, and the deviation between the candidate order and the historical order can be transformed into the switching cost time. Finally, a single total effective assembly time is obtained, achieving a high degree of fit between algorithm optimization and actual engineering needs.

[0115] In detail, the unified reference response speed can be determined by the implementing unit based on the terrain features of the construction site and the mobility of emergency personnel. The vulnerability penalty weight, congestion penalty weight, and priority switching time coefficient are all adjusted according to the actual site scenario within the optimal value range. The more obvious the thin-shell corner defects, the greater the vulnerability penalty weight; the higher the congestion level of the fan clusters, the greater the congestion penalty weight; the lower the efficiency of on-site command transmission, the greater the priority switching time coefficient. If there is no historical deployment priority table for the first round of dispatch, the priority rearrangement distance is directly assigned a value of 0, and the effective assembly time model automatically degenerates into the brittleness-corrected assembly time. Minimize the number of emergency groups. The candidate order is generated by sorting the priority weights output by the subsequent particle swarm optimization algorithm in descending order. When the number of emergency groups is 1, the group overlap coefficient is 0, the order rearrangement distance is 0, the delay ratio only includes the base value 1 and the brittleness penalty term, the total assembly time is the nominal assembly time of the group multiplied by the delay ratio, the optimization objective of the effective assembly time model is to minimize the total effective time cost, the global summation range of the total assembly time is all emergency groups in the candidate order, and it is necessary to traverse each group to complete the calculation, the calculation results of all parameters are retained to 6 decimal places, and the calculation unit of all time-related parameters is unified to seconds.

[0116] Preferably, the priority weight is used as the particle position variable, the order vulnerability coefficient is introduced for bias initialization, and intelligent iterative optimization is performed based on the effective assembly time model to obtain the optimal priority weight, including:

[0117] The optimal priority weight is obtained by initializing and iteratively optimizing according to the following formula:

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] in, Indicates the initial priority weight; Indicates the initial random number; Indicates the embrittlement bias coefficient; This represents the order vulnerability coefficient; Indicates the candidate deployment sequence; This represents a function to sort in descending order. and Each represents the priority weight of the current iteration; Indicates the fitness value; This represents the effective aggregation time model under the corresponding sequence; Indicates update speed; Indicates inertia weight; Indicates the optimization speed; Represents an individual's learning factor; Represents the first random number; This represents the optimal weight for an individual; Represents the global learning factor; Indicates the second random number; Indicates the globally optimal weight; This indicates an update to the priority weight.

[0124] The brittleness bias coefficient is a weighted parameter used to introduce a positional fragility bias during the particle swarm initialization phase. It is a user-defined parameter. A value of 0.5 is preferred to match the range of 0 to 1 for the positional fragility coefficient, balancing the diversity of random initialization with the specificity of the brittleness bias, and preventing the bias from being too strong or too weak, which could negatively impact optimization efficiency.

[0125] Inertia weight is a parameter used to adjust the degree of influence of historical velocity on the current velocity update during particle swarm optimization. It is a user-defined parameter. The preferred value is 0.7, which is a classic fit value for the particle swarm optimization algorithm, balancing the algorithm's global exploration capability and local development capability, and adapting to the optimization needs of emergency engineering scheduling.

[0126] The individual learning factor is a parameter used to adjust the degree to which the optimal position of an individual particle affects its own optimization direction and velocity. It is a user-defined parameter. The preferred value is 1.5. This value can ensure the learning strength of the particle for its own optimal solution, while avoiding local optima problems caused by over-convergence. It works in conjunction with the global learning factor to achieve optimization balance.

[0127] The global learning factor is a parameter used to adjust the degree to which the global optimal position of the particle swarm influences the optimization direction and velocity of individual particles. It is a user-defined parameter. A value of 1.5 is preferred, as this ensures the learning intensity of particles towards the global optimal solution, matches the individual learning factor, and allows the particle swarm to achieve a balance between global exploration and local development, thus improving optimization accuracy.

[0128] The initial random number is a basic random value generated for each particle during the particle swarm initialization phase, and is randomly generated by the algorithm.

[0129] The initial priority weight is the value obtained by subtracting the vulnerability bias term from the initial random number, and it is the emergency group priority weight in the initial state of particle swarm iteration.

[0130] The current iteration priority weight is the real-time priority weight of an emergency response team for a single particle in each iteration of the particle swarm.

[0131] The candidate deployment sequence is a candidate scheme for the deployment order of emergency teams obtained by sorting the current iteration priority weights in descending order.

[0132] The fitness value is a numerical value obtained by substituting the candidate mobilization sequence into the effective assembly time model. It is a quantitative indicator that characterizes the quality of the candidate mobilization sequence.

[0133] The optimization speed is the velocity parameter of a single particle in each iteration of the particle swarm, representing the magnitude and direction of the priority weight update.

[0134] The inertial component is the value obtained by multiplying the inertial weight by the optimization velocity, and it represents the influence of the particle's historical optimization velocity on the current velocity update.

[0135] The individual cognitive component is a value obtained by multiplying the individual learning factor, the first random number, and the difference between the individual's optimal weight and the current priority weight. It represents the influence of the particle's own optimal solution on the current velocity update.

[0136] The global social component is a value obtained by multiplying the global learning factor, the second random number, the difference between the global optimal weight and the current priority weight, and it represents the influence of the global optimal solution of the particle swarm on the current velocity update.

[0137] The update speed is the value obtained by adding the inertial component, the individual cognitive component, and the global social component. It is the optimization speed of the particle in the next iteration.

[0138] The update priority weight is the value obtained by adding the current iteration priority weight to the update speed, and it is the emergency group priority weight for the particle in the next iteration.

[0139] The individual optimal weight is the priority weight that corresponds to the best fitness value for a single particle throughout all iterations.

[0140] The global optimal weight is the priority weight with the best corresponding fitness value among all the individual optimal weights of all particles.

[0141] The optimal priority weight is the final globally optimal weight extracted after the particle swarm iteration meets the termination condition, and it is the final quantitative basis for the priority of the emergency response team.

[0142] The first random number is a random value generated during the particle swarm velocity update phase. It is randomly generated by the algorithm and is used to increase the randomness of individual cognitive components.

[0143] The second random number is a random value generated during the particle swarm velocity update phase. It is randomly generated by the algorithm and is used to increase the randomness of the global social component.

[0144] In detail, firstly, a fragility bias coefficient is introduced during the initialization phase. This transforms the order fragility coefficient into a fragility bias term that participates in the initial priority weight calculation. This reduces the initial weight of emergency response teams with high order fragility coefficients, allowing the particle swarm optimization to avoid highly fragile point sets from the beginning of the iteration, thus improving the targeting and efficiency of optimization. Secondly, a core mechanism of sorting before evaluation is designed. Continuous priority weights are first arranged in descending order to form a discrete candidate deployment sequence, and then substituted into the effective assembly time model to calculate the fitness value. This perfectly solves the adaptation problem between continuous particle swarm optimization algorithms and discrete emergency deployment order optimization. At the same time, the particle position variable is directly set as the priority weight of the emergency response team, achieving a lightweight approach while retaining the core framework of the particle swarm optimization algorithm. To adapt and reduce the difficulty of implementation in engineering projects, the effective assembly time model is used as the sole evaluation criterion for fitness value. This ensures that the optimization direction of the particle swarm always points to the engineering goal of minimizing the total effective assembly time. Individual and global optimal weights are updated only based on fitness value, ensuring the singularity and determinism of optimization. For example, in a water plant construction accident site, there are 5 emergency response teams, 2 of which have a fragility coefficient close to 1. During particle swarm initialization, the initial priority weights of these 2 teams will be reduced due to fragility bias. The initial search of the particle swarm will then move away from the highly fragile combination of these two teams. In subsequent iterations, by sorting first and then evaluating, the solution of each continuous weight can be transformed into an actual deployment order scheme.

[0145] In detail, the initial random numbers, the first and second random numbers, are both in the range of 0 to 1 and are generated by a uniform random algorithm. The particle swarm optimization algorithm preferably has 30 particles and a maximum number of iterations preferably of 80, balancing optimization accuracy and computational efficiency. The optimization speed has a boundary value of -0.3 to 0.3, and the priority weight has a boundary value of -1 to 1. If the update speed or the updated priority weight exceeds the boundary after calculation, it is directly truncated to the corresponding boundary value. The iteration termination condition is reaching the preset maximum number of iterations or the change in fitness value corresponding to the global optimal weight for 10 consecutive rounds being less than 10 to the power of -6. The iteration terminates when either condition is met. When stopping the iteration, if the priority weights are sorted in descending order and the weight values ​​are equal, they are sorted in lexicographical order according to the emergency group identifier. If the individual optimal weights of multiple particles correspond to the same fitness value, the weight of the particle with the smallest combination of emergency group identifiers is selected as the global optimal weight. The individual optimal weight is only updated when the fitness value of the new iteration is better than the historical optimal value. It is not allowed to be updated to a worse result. The running scheduling cycle of the particle swarm optimization algorithm is synchronized with the scheduling cycle of coordinate acquisition, which is 2 seconds or 5 seconds, consistent with the update rhythm of the emergency positioning conversion module. The calculation results of all parameters are retained to 6 decimal places.

[0146] Preferably, a distance fine-tuning term is introduced into the optimal priority weight for stable decoding to obtain the final order, and a dispatch order table is generated based on the final order, including:

[0147] The final order of deployment is calculated using the following formula, and the deployment order table is generated accordingly:

[0148]

[0149]

[0150]

[0151]

[0152]

[0153] in, and All of these indicate stable ranking scores; This represents the optimal priority weight; Indicates stable decoding coefficients; This represents the average relative radius; Indicates the relative radius; This represents the first bias amount; as well as Both indicate the final order; This indicates an indicator function that takes the value of one when the condition inside the parentheses is true, and zero otherwise. This represents the dispatch order table; Indicates the emergency response team's identification; Indicates the total number of emergency response teams; The historical deployment order table indicates the next cycle.

[0154] The stable decoding coefficients are minimal coefficient parameters used to generate the distance fine-tuning term, and are user-defined parameters. The preferred value is 10 to the power of -3, because this value is a very small positive number, which only breaks the approximate parallel state of priority weights without changing the overall sorting logic. It also adapts to the precision characteristics of floating-point calculations, avoiding the introduction of new sorting deviations.

[0155] The radius difference is the value obtained by subtracting the relative radius of an individual emergency response team from the average relative radius, and it characterizes the degree of difference between the relative radius of an individual emergency response team and the overall average radius.

[0156] The distance fine-tuning term is a value obtained by multiplying the normalized radius difference by the stable decoding coefficient. It is used to break the near-parallel alignment of optimal priority weights.

[0157] The ranking stability score is the value obtained by adding the optimal priority weight and the distance fine-tuning term, and it serves as the core basis for the emergency team's final ranking.

[0158] The target number is the number of other emergency response teams whose ranking scores are consistently higher than the current emergency response team, representing the ranking reference base for the current emergency response team.

[0159] The final order is the integer obtained by adding one to the target number, representing the final order of deployment of the emergency response team.

[0160] The emergency response team identifier is a unique identifier used to distinguish each emergency response team. It can be determined through system configuration using either a numeric code or an alphanumeric code.

[0161] The deployment priority table is a one-dimensional dispatch table composed of emergency team identifiers and corresponding final priority, arranged from smallest to largest according to the final priority. It is the core instruction basis for on-site emergency dispatch.

[0162] In detail, firstly, a very small, stable decoding coefficient is introduced to generate a distance fine-tuning term. This term makes a slight adjustment to the optimal priority weight based only on the difference in relative radii, and is only used to break numerical ties without changing the overall ranking result. Then, a unique mapping from continuous values ​​to discrete final order is achieved through the counting method of stable ranking scores, ensuring the determinism of the ranking result. Subsequently, a one-dimensional deployment order table containing only emergency team identifiers and final order is designed to adapt to the simple scheduling requirements of engineering sites and avoid complex data interfering with on-site execution. Most importantly, the deployment order table of the current cycle is updated to the historical order of the next cycle. The historical order table allows subsequent scheduling processes to consider the switching costs of historical orders, thus suppressing order flutter from a process perspective. For example, at a water plant construction accident site, there are three emergency response teams distributed in a thin shell pattern. The optimal priority weights obtained after particle swarm optimization are extremely close, and direct sorting will result in frequent order swaps. Through the stable decoding mechanism of this scheme, after adding a distance fine-tuning term based on relative radius, the sorting stable score produces a small and definite difference, which can generate a unique final order, completely avoiding the problem of order flutter. At the same time, this order table will participate in the next round of scheduling as historical data, effectively controlling the instruction switching cost.

[0163] In detail, when the scores are completely equal, the final order is determined by the lexicographical order of the emergency team identifiers. Numerical codes are sorted in ascending order of value, while alphanumeric codes are sorted first by number and then by letter. Each emergency team identifier uses a unique, fixed code. The coding rules are determined during system initialization and remain unchanged throughout the system. The output format of the deployment order table is a two-column one-dimensional table: the first column is the emergency team identifier, and the second column is the final order. There are no other redundant fields. The deployment order table is immediately stored in the system's local storage module after generation, and is also stored after the end of the scheduling cycle and before the start of the next scheduling cycle. The list is updated to reflect the historical deployment order. The maximum value of the stable decoding coefficient does not exceed 10 to the power of -2 to avoid excessively large distance fine-tuning terms that could alter the overall sorting logic. The calculation precision of the stable score is retained to 6 decimal places. If an emergency team loses its coordinates and fails to recover them for two consecutive rounds, its final order is placed directly after all normal teams. The deployment order list is transmitted to the field dispatch terminal in real time via a wireless communication module, with a transmission delay of no more than 1 second to ensure the real-time performance of field dispatch. When there is no emergency team, the deployment order list is empty, and the system sends a signal to the dispatch terminal that there is no dispatchable team.

[0164] Example 2: A digital engineering safety emergency management method, applied to any of the digital engineering safety emergency management systems described in this example, comprising:

[0165] The obtained accident location and the emergency team locations of each emergency team are converted into accident reference coordinates that include relative radius and relative azimuth.

[0166] Based on the relative radius and the relative azimuth angle, calculate the sequence vulnerability of the thin-shell corner-missing type;

[0167] Calculate the proximity of the same shell and the same fan among the various emergency response teams, and combine it with the thin-shell corner-cut sequence vulnerability to generate the sequence vulnerability coefficient of each emergency response team.

[0168] Based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient, an effective assembly time model for a single target is constructed.

[0169] Using priority weights as particle position variables, the order vulnerability coefficient is introduced for bias initialization, and intelligent iterative optimization is performed based on the effective assembly time model to obtain the optimal priority weights.

[0170] The optimal priority weight is subjected to a distance fine-tuning term for stable decoding to obtain the final order. A deployment order table is generated based on the final order, and the deployment order table is used as the historical deployment order table for the next cycle.

[0171] like Figure 2 As shown, Figure 2This is a polar coordinate distribution diagram centered on the accident point. The radial ring represents the straight-line distance of the emergency response team relative to the accident point, and the angular partition represents the relative orientation, intuitively presenting the spatial distribution characteristics of the emergency response team in an engineering emergency scenario. In the figure, the emergency response teams are concentrated in a few directional sectors and their relative radii and heights are similar, forming a typical spatial structure of thin-shell corner missing. The densely populated areas of local teams are marked as congested areas. This distribution is the core scenario that triggers the sortie order flutter and can intuitively support the quantitative analysis of the thin-shell corner missing type sequence embrittlement characteristics.

[0172] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0173] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A digital engineering safety emergency management system, characterized in that, include: The emergency location conversion module is used to convert the acquired accident point location and the emergency team locations of each emergency team into accident reference coordinates that include relative radius and relative azimuth. The spatial distribution assessment module is used to calculate the order vulnerability of the thin-shell corner-missing type based on the relative radius and the relative azimuth angle, including: Calculate the mean of the relative radii as the average relative radius; Extract the maximum and minimum relative radii from the relative radii corresponding to each of the aforementioned emergency response teams; The radial dispersion is obtained by dividing the difference between the maximum relative radius and the minimum relative radius by the sum of the average relative radius and the first offset. The group percentage is calculated based on the relative azimuth angle, and the azimuth spread entropy is obtained by combining the total number of partitions with the second offset. Arrange the relative azimuth angles in ascending order to obtain the ascending azimuth angles; Divide the difference between adjacent ascending azimuth angles by the circumference angle and extract the maximum value to obtain the maximum notch ratio; Extract the minimum value among the absolute values ​​of any two relative radius differences as the minimum spacing; The difference between the azimuth spread entropy and the maximum gap ratio is multiplied by the difference, and divided by the sum of the radial dispersion, the minimum spacing, the average relative radius and the first bias, and the third bias, to obtain the thin-shell corner-type sequence vulnerability. An emergency congestion analysis module is used to calculate the proximity of the same shell and the same fan among the various emergency groups, and, in combination with the thin-shell corner-missing sequence vulnerability, to generate the sequence vulnerability coefficient of each emergency group, including: Calculate the absolute value of the difference between the relative azimuth angles of any two emergency response teams, compare the absolute value with the difference between the circumferential angle and the absolute value, and extract the minimum value to obtain the azimuth angle between the emergency response teams; Divide the absolute value of the difference between the relative radii of any two emergency response teams by the radial attenuation scale, take the opposite number, and then perform an exponential operation to obtain the radial proximity component; divide the azimuth angle between the emergency response teams by the angular attenuation scale, take the opposite number, and then perform an exponential operation to obtain the angular proximity component; multiply the radial proximity component and the angular proximity component to obtain the proximity degree of the same shell and the same fan. The local fan cluster congestion degree is obtained by summing the proximity of a single emergency response team to the same shell and the same fan among all other emergency response teams. Extract the minimum absolute value of the difference between the relative radii of a single emergency response team and all other emergency response teams, and divide it by the sum of the average relative radius and the first offset to obtain the local order gap coefficient; Multiply the thin-shell corner-shaped sequence vulnerability by the local fan cluster crowding to obtain the vulnerability-crowding product; divide the vulnerability-crowding product by the sum of the vulnerability-crowding product, the local sequence gap coefficient, and the fourth bias to obtain the sequence vulnerability coefficient. The scheduling time evaluation module is used to construct an effective assembly time model for a single target based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient, including: Divide the relative radius by the uniform reference response speed to obtain the nominal assembly time of each emergency response team; In any candidate order, the proximity between the current emergency group and other emergency groups ranked before the current emergency group is extracted, and the grouping overlap coefficient is obtained by calculating the mean of the proximity between the current emergency group and other emergency groups ranked before the current emergency group. The total number of emergency response teams is obtained by counting the number of such teams. Extract the candidate deployment ranking of each emergency response team in the candidate order and the historical deployment ranking in the historical deployment ranking table. Sum the absolute values ​​of the differences between the candidate deployment ranking and the historical deployment ranking to obtain the total ranking deviation. Divide the total ranking deviation by the product of the total number of emergency response teams and the difference between the total number of emergency response teams and the total number of emergency response teams minus one to obtain the ranking rearrangement distance. The delay ratio is obtained by adding the product of the order vulnerability coefficient and the vulnerability penalty weight, and the product of the grouping overlap coefficient and the crowding penalty weight. The delay ratio is multiplied by the nominal assembly time and summed globally to obtain the total assembly time. The order rearrangement distance is multiplied by the order switching time coefficient to obtain the switching cost time. The total assembly time is added to the switching cost time to obtain the effective assembly time model. The scrambling order optimization module is used to initialize the order vulnerability coefficient with priority weight as the particle position variable, and to perform intelligent iterative optimization based on the effective assembly time model to obtain the optimal priority weight, including: Get the initial random number; Multiply the order fragility coefficient by the fragility bias coefficient to obtain the fragility bias term, and subtract the fragility bias term from the initial random number to obtain the initial priority weight; The priority weights of the current iteration are sorted in descending order to generate a candidate deployment sequence; The fitness value is calculated by substituting the candidate mobilization sequence into the effective assembly time model. The inertial component is obtained by multiplying the inertial weight by the optimization speed. The individual cognitive component is obtained by multiplying the individual learning factor, the first random number, and the difference between the individual optimal weight and the priority weight. The global social component is obtained by multiplying the global learning factor, the second random number, and the difference between the global optimal weight and the priority weight. The update speed is obtained by adding the inertial component, the individual cognitive component, and the global social component. The update priority weight is obtained by adding the update speed to the priority weight; When the iteration termination condition is met, the final global optimal weight is extracted as the optimal priority weight; The scheduling instruction generation module is used to introduce a distance fine-tuning term into the optimal priority weight for stable decoding to obtain the final order, generate a deployment order table based on the final order, and use the deployment order table as the historical deployment order table for the next cycle, including: The radius difference is obtained by subtracting the relative radius from the average relative radius. Divide the radius difference by the sum of the average relative radius and the first bias, and multiply the resulting quotient by the stable decoding coefficient to obtain the distance fine-tuning term; The optimal priority weight is added to the distance fine-tuning term to obtain the ranking stability score; The target number is obtained by counting the number of emergency groups whose ranking stability score is greater than that of the current emergency group among all other emergency groups. The target number is then added to one to obtain the final ranking. The emergency response teams are arranged in ascending order of their final priority, and their identifiers are extracted. The deployment priority table is then generated based on the arranged emergency response team identifiers and the final priority. The generated deployment order table is updated to the historical deployment order table for the next cycle.

2. The digital engineering safety emergency management system according to claim 1, characterized in that, The emergency positioning conversion module converts the acquired accident location and the emergency team locations of each emergency team into accident reference coordinates that include relative radius and relative azimuth, including: Obtain the x-coordinate and y-coordinate of the accident point location, as well as the x-coordinate and y-coordinate of each emergency response team location. Subtract the abscissa of the accident point from the abscissa of the emergency response team to obtain the lateral displacement component, and subtract the ordinate of the accident point from the ordinate of the emergency response team to obtain the longitudinal displacement component. A two-dimensional displacement vector is generated based on the lateral displacement component and the longitudinal displacement component; The relative radius is calculated by summing the square of the lateral displacement component and the square of the longitudinal displacement component, and then taking the square root of the summation result. The relative azimuth angle is obtained by performing an arctangent function calculation using the longitudinal displacement component and the lateral displacement component as input parameters.

3. A digital engineering safety emergency management method, applied in the digital engineering safety emergency management system described in any one of claims 1-2, characterized in that, include: The obtained accident location and the emergency team locations of each emergency team are converted into accident reference coordinates that include relative radius and relative azimuth. Based on the relative radius and the relative azimuth angle, calculate the sequence vulnerability of the thin-shell corner-missing type; Calculate the proximity of the same shell and the same fan among the various emergency response teams, and combine it with the thin-shell corner-cut sequence vulnerability to generate the sequence vulnerability coefficient of each emergency response team. Based on the nominal assembly time, the historical deployment order table of the previous cycle, and the order vulnerability coefficient, an effective assembly time model for a single target is constructed. Using priority weights as particle position variables, the order vulnerability coefficient is introduced for bias initialization, and intelligent iterative optimization is performed based on the effective assembly time model to obtain the optimal priority weights. The optimal priority weight is subjected to a distance fine-tuning term for stable decoding to obtain the final order. A deployment order table is generated based on the final order, and the deployment order table is used as the historical deployment order table for the next cycle.

Images