CN-122022088-A - Nuclear accident emergency path planning method, medium, equipment and product

CN122022088ACN 122022088 ACN122022088 ACN 122022088ACN-122022088-A

Abstract

The specification discloses a nuclear accident emergency path planning method, medium, equipment and product, and relates to the technical field of nuclear accident emergency management and intelligent decision making. An evacuation decision dataset for a deep learning model is constructed by collecting user data, resource information, environmental meteorological data, environmental radiation data, and map information, and acquiring concentration field prediction data and/or dose field prediction data. And carrying out joint modeling on the radiation diffusion trend, the road space structure and the personnel state by using the model, and outputting probability distribution of candidate evacuation azimuth angles. And finally, carrying out safety evaluation on the evacuation route to determine a final evacuation route. According to the invention, under a complex dynamic environment after nuclear accident occurrence, radiation diffusion prediction information and deep learning decision-making capability are fully utilized, prospective planning and dynamic optimization of a personnel evacuation path are realized, the radiation exposure risk of personnel is effectively reduced, and the safety and reliability of nuclear accident emergency evacuation are improved.

Inventors

LV MINGHUA
ZHANG YUETING
LI YU
PAN YANHUI
NIU YANJING
GUO HUAN
LI QINGYUN
YAO RENTAI
LI MINGYE
LIAN BING
ZHANG JUNFANG
Lv Xindong
QIU ZHIXIN
HE XIN
ZHAO DUOXIN

Assignees

中国辐射防护研究院

Dates

Publication Date: 20260512
Application Date: 20251229

Claims (12)

1. A nuclear accident emergency path planning method, comprising: Acquiring and acquiring user data and/or resource information and/or environmental meteorological data and/or environmental radiation data and/or map information, wherein the environmental radiation data comprises concentration field prediction data and/or dose field prediction data with future time steps acquired from a nuclear accident consequence evaluation system and is used for representing the diffusion state of radiation substances at a plurality of future time nodes after a nuclear accident; Constructing an evacuation decision data set for a deep learning model based on the environmental meteorological data, the environmental radiation data and the map information, wherein the evacuation decision data set at least comprises multi-time-step radiation prediction features, road space structure features and evacuation decision samples corresponding to the multi-time-step radiation prediction features and/or the road space structure features; The method comprises the steps of inputting a current position of a user, user state information and the evacuation decision data set into a deep learning model, and determining an evacuation decision result of the user at the current position, wherein the deep learning model adopts a time sequence prediction and space feature fusion framework suitable for a nuclear accident prediction scene to carry out joint modeling on a future radiation diffusion trend and personnel evacuation behavior; Based on the evacuation decision result, probability distribution output is carried out on azimuth angles of a plurality of candidate evacuation paths by using a deep learning model, an optimal evacuation azimuth is determined according to the probability result of the azimuth angles, and a corresponding optimal evacuation route is generated by combining a road network structure; and carrying out safety evaluation on the optimal evacuation route, and determining a final evacuation route according to an evaluation result, wherein the safety evaluation comprises an overall optimal evaluation based on a future time scale and an instant optimal evaluation based on the current moment.
2. The nuclear accident emergency path planning method according to claim 1, wherein the concentration field prediction data and/or the dose field prediction data comprise spatially distributed data at a plurality of preset time steps for characterizing a trend of the radiation level over time.
3. A method of nuclear accident emergency path planning according to claim 1, wherein each data sample in the evacuation decision data set includes current time instant state data and radiation prediction data corresponding to at least one future time step to construct an evacuation decision sample having a time series characteristic.
4. The nuclear accident emergency path planning method according to claim 1, wherein the deep learning model comprises a time sequence modeling unit for processing the radiation prediction time sequence and a spatial feature extraction unit for processing the road spatial structure, and the time sequence modeling unit and the spatial feature extraction unit perform feature fusion to achieve modeling of the association relationship between the nuclear accident radiation prediction field and the evacuation behavior.
5. The nuclear accident emergency path planning method of claim 1, wherein the deep learning model is configured to process radiation spread data including prediction uncertainty and to improve the robustness of evacuation decisions in a nuclear accident environment by learning the effect of radiation prediction errors on the evacuation decisions.
6. The nuclear accident emergency path planning method according to claim 1, wherein the deep learning model outputs a probability value for an azimuth angle corresponding to each candidate evacuation path, characterizes the safety reliability of the azimuth angle as an evacuation direction, and determines an optimal evacuation azimuth based on the probability value.
7. The nuclear accident emergency path planning method of claim 6, wherein the optimal evacuation route is generated based on a matching relationship of the optimal evacuation azimuth to road azimuth in a road network.
8. A nuclear accident emergency path planning method according to claim 1, wherein the overall optimal assessment is used to assess the cumulative radiation risk and overall safety of the evacuation path throughout the evacuation process, and the instantaneous optimal assessment is used to assess the safety status and feasibility of the evacuation path at the current time.
9. The nuclear accident emergency path planning method of claim 1, wherein during the user evacuation process, based on the updated radiation prediction data and the user location information, the step of inputting the user current location, the user status information, and the evacuation decision data set into a deep learning model is returned, and the evacuation decision result of the user at the current location is determined, so as to dynamically adjust the evacuation path.
10. A computer readable storage medium, characterized in that the storage medium stores a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-9.
11. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of the preceding claims 1-9 when executing the program.
12. A computer program product, characterized in that instructions in the computer program product, when executed by a processor of an electronic device, cause the electronic device to perform the method of any of claims 1-9.

Description

Nuclear accident emergency path planning method, medium, equipment and product Technical Field The specification relates to the technical field of nuclear accident handling, in particular to a nuclear accident emergency path planning method, medium, equipment and product. Background After a nuclear accident, the release and diffusion of radioactive substances in the environment can continuously and dynamically influence the accident area and the surrounding environment, and how to provide a safe and reasonable path planning scheme for personnel evacuation in a complex and changeable radiation environment in the emergency treatment process of the accident is one of the important technical problems facing the emergency management field of the nuclear accident. The conventional nuclear accident emergency decision system can generally evaluate the result of the diffusion range and the radiation level of the radioactive substances based on accident source parameters, meteorological conditions and topography conditions, but in the actual personnel evacuation process, the result evaluation result is still required to be combined with a specific evacuation path planning problem, so that the radiation exposure risk of the personnel in the evacuation process is reduced as much as possible while the evacuation efficiency is ensured. In the prior art, a decision mode based on rules or weights is mostly adopted in a personnel evacuation path planning method, and a plurality of candidate evacuation paths are comprehensively scored by setting fixed or manually set weights on factors such as radiation dose, travel distance, travel time, road traffic conditions and the like, so that a path with an optimal scoring result is selected as a recommended evacuation scheme. However, in practical application, the method often relies on manual experience to set weight parameters, so that complex characteristics of radiation diffusion changing along with time after nuclear accident occurrence are difficult to comprehensively reflect, and when meteorological conditions change or accident working conditions evolve, the set weight parameters are difficult to timely adjust, and deviation between a path planning result and actual radiation risk is easy to occur. In addition, in some of the prior art, decisions are mainly made according to radiation monitoring data at the current moment or in a short time in the path planning process, and the diffusion trend of radioactive substances on a future time scale cannot be fully considered, so that the situation that a path selected by a person in the initial stage of evacuation gradually enters a high-radiation area in a subsequent period can occur, and the accumulated radiation exposure risk of the person in the evacuation process is increased. Meanwhile, as the urban road network structure is increasingly complex, the evacuation safety, the instantaneity and the feasibility are difficult to be considered under the emergency scene of the nuclear accident by simply relying on the traditional shortest path or static risk avoidance algorithm. In recent years, with the development of artificial intelligence technology, part of research attempts to introduce a machine learning or deep learning method into the field of path planning or decision support, but the related art is concentrated on traffic navigation, robot path planning or emergency decision under a general disaster scene, and often lacks consideration of radiation prediction uncertainty, time sequence characteristics and result evaluation result depth fusion in a nuclear accident emergency scene, and a technical scheme capable of effectively utilizing a prediction result of a nuclear accident result evaluation system and performing prospective and dynamic optimization on a personnel evacuation direction and a path is not formed. Therefore, how to fully combine the prediction field information output by the result evaluation system in the nuclear accident emergency path planning process and construct an intelligent decision method suitable for the nuclear accident scene is still to be further researched and perfected. Disclosure of Invention The present disclosure provides a method, medium, apparatus and product for planning a nuclear accident emergency path to at least partially solve the above-mentioned problems of the prior art. The technical scheme adopted in the specification is as follows: The specification provides a nuclear accident emergency path planning method, which comprises the following steps: Acquiring and acquiring user data and/or resource information and/or environmental meteorological data and/or environmental radiation data and/or map information, wherein the environmental radiation data comprises concentration field prediction data and/or dose field prediction data with future time steps acquired from a nuclear accident consequence evaluation system and is used for representing the diffusion state of radiation substan