A dynamic evacuation path planning method and system based on multi-source data fusion

Abstract

The application discloses a kind of dynamic evacuation path planning method and system based on multi-source data fusion, it is related to path planning technical field, constructs indoor environment static grid map, defines walkable area, obstacle area and safety exit position;Real-time acquisition and fusion multi-source dynamic data, generate dynamic comprehensive generation value;Dynamic comprehensive generation value is superimposed with static grid map, and constructs real-time dynamic network model;The real-time position of all personnel to be evacuated at current time is acquired as multi-source starting point, and all available safety exit is as multi-harmonic terminal;Using improved DLite algorithm, based on real-time dynamic network model, the safe evacuation path of each evacuation personnel is calculated;The dynamic path planned is real-time issued to corresponding evacuation personnel terminal.The application uses the above-mentioned dynamic evacuation path planning method and system based on multi-source data fusion, can provide safe, efficient and personalized optimal evacuation path for large-scale evacuation crowd in real time.

Description

Technical Field

[0001] This invention relates to the field of route planning technology, and in particular to a dynamic evacuation route planning method and system based on multi-source data fusion. Background Technology

[0002] With the increasing number of large and complex buildings, the safe evacuation of people in the event of indoor fires and other emergencies faces severe challenges. Traditional static evacuation signs and broadcasts are unable to cope with complex situations such as fire, dynamic smoke spread, and crowd congestion. Therefore, real-time, dynamic, and intelligent route planning technology has become a research hotspot in the field of emergency evacuation.

[0003] In existing technologies, algorithms based on Dijkstra's algorithm, A, D, and their variants are widely used for path planning. However, in the specific scenario of indoor fire evacuation, existing methods have significant shortcomings: 1) Limited data: Most of the data only considers static maps or simple fire simulations, without deeply integrating multi-dimensional data such as real-time smoke concentration, accurate crowd density, and passage status. This may result in the planned routes being infeasible in reality due to dense smoke or congestion.

[0004] 2) Efficiency bottleneck: In large-scale personnel evacuation scenarios with multiple sources and multiple sinks (multiple starting points and multiple exits), traditional single-point path planning algorithms are inefficient and cannot meet real-time requirements.

[0005] 3) Insufficient risk avoidance: Most algorithms aim at the shortest path and lack proactive and forward-looking avoidance mechanisms for dynamic spread risks, which may lead to evacuation routes being too close to areas that will soon be covered by fire or smoke. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic evacuation route planning method and system based on multi-source data fusion. It integrates a comprehensive environmental model that incorporates multi-dimensional dynamic information such as fire, smoke, pedestrian flow, and passage status, and adopts an efficient improved route planning algorithm to realize safe, efficient, and personalized optimal evacuation routes for large-scale evacuation groups in real time.

[0007] To achieve the above objectives, this invention provides a dynamic evacuation path planning method based on multi-source data fusion, comprising the following steps: S1. Construct a static grid map of the indoor environment, defining walkable areas, obstacle areas, and the location of safety exits; S2. Real-time collection and fusion of multi-source dynamic data to generate dynamic comprehensive value. The multi-source dynamic data includes at least: fire location and spread information, smoke concentration distribution information, regional population density information, and passage opening / closing or congestion status information. S3. Overlay the dynamic comprehensive cost with the static grid map to construct a real-time dynamic network model, in which the passage cost of each grid node is determined by the basic movement cost and the dynamic comprehensive cost. S4. Obtain the real-time location of all personnel to be evacuated at the current moment as the multi-source starting point, and all available safety exits as the multi-sink endpoint; S5, Adopting an improved D-based approach The Lite algorithm, based on a real-time dynamic network model, calculates the safe evacuation path from each evacuee's starting point to the optimal exit in real time for each evacuee. S6. The planned dynamic path is sent to the corresponding evacuation personnel terminal or guidance device in real time.

[0008] Preferably, in step S2, the formula for calculating the dynamic comprehensive cost value is as follows: ; in, For grid nodes, The current time; This is a fire hazard function calculated based on a fire location and temperature model; The visibility / toxicity cost function is based on smoke concentration data; The congestion delay cost function is based on pedestrian density data; For channel state functions; These are the weighting coefficients for each cost item, and can be dynamically adjusted according to the stage of the disaster.

[0009] Preferred fire hazard function and smoke cost function The input data comes from the real-time output of fire dynamics simulation software or from indoor temperature and smoke sensor networks; pedestrian density function The input data comes from the real-time crowd analysis results of the video surveillance system; channel state function The input data comes from the access control system status or pressure sensor data.

[0010] Preferably, the improvements to the algorithm in step S5 include: S51. Introduce a search rectangle constraint based on the starting point and the target exit to dynamically limit the initial search range of the algorithm; S52. A two-layer risk avoidance mechanism is adopted. When calculating the cost of a node, the danger level of the current node and the danger information of its adjacent second-layer nodes are considered. The penalty term guides the path away from the potential danger area. S53. Introduce virtual nodes to connect all actual exits, transforming the multi-source, multi-sink path planning problem into a multi-source, single-sink problem; S54. A shared environment graph mechanism is adopted to store and reuse the path cost information of traversed nodes for subsequent path planning of other starting points, thereby reducing redundant calculations.

[0011] Preferably, the method for constructing and dynamically expanding the search rectangle constraint in step S51 is as follows: Construct an initial rectangular search region with the starting point and the target exit as diagonal vertices; if the rectangle is divided by obstacles, causing the starting point and the exit to be disconnected, expand the rectangle boundary outward by a preset step size until the starting point and the exit are in the same connected region composed of walkable grids; use the final connected region as the constraint region for this path search.

[0012] Preferably, in step S52, the two-layer risk avoidance mechanism modifies the nodes in the path planning algorithm. right-hand side value The calculation formula is implemented as follows: cost Penalty ; in, It is a node The set of successor nodes, From arrive The sum of the basic movement cost and the dynamic comprehensive cost. It is a node The current minimum cost, Penalty Based on nodes Second Neighborhood The penalty items for calculating the degree of internal danger, if If a high-risk node is found in memory, a penalty value greater than the difference between the maximum possible heuristic difference and the movement cost is applied. .

[0013] Preferably, in step S54, the shared environment graph is a global data structure that stores all the grid nodes that have been visited during the planning process. Value and When planning a path for a new evacuation starting point, the algorithm first queries the node information already stored in the shared environment graph and directly reuses valid information. Value and The value is only recalculated and updated for nodes whose information is invalid due to environmental changes or different starting points.

[0014] A dynamic evacuation route planning system based on multi-source data fusion includes: The data sensing and fusion module is used to collect data from multiple heterogeneous sensors and systems in real time, and perform fusion processing according to step S2 to output a dynamic comprehensive cost layer. The dynamic environment modeling module is used to load static grid maps and receive the dynamic comprehensive cost layer, and then overlay them to generate a real-time dynamic network model. The core path planning module has an improved path planning algorithm built in. It receives the real-time dynamic network model, the real-time location of evacuees and exit information, and performs parallel or serial real-time path calculation according to step S5. The route distribution and guidance module is used to distribute the planned personalized evacuation routes to user terminals, public displays, or audio-visual guidance devices in the form of graphics, instructions, or signals.

[0015] Therefore, the present invention employs the above-mentioned dynamic evacuation path planning method and system based on multi-source data fusion, which has the following beneficial effects: (1) By integrating real-time data from multiple dimensions such as fire, smoke, pedestrian flow, and passage status, the constructed environmental model is closer to the real disaster situation, making path planning decisions more scientific and safer; (2) The two-layer risk avoidance mechanism provides a forward-looking risk avoidance capability, which can effectively guide evacuees away from current and near-future high-risk areas and improve the safety of the route; (3) Combining the three strategies of search rectangle constraint, virtual node transformation and shared environment graph reuse, the computational performance of large-scale evacuation scenarios with multiple sources and sinks is optimized to meet the requirements of real-time dynamic replanning.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] Figure 1 This is a flowchart of a dynamic evacuation path planning method based on multi-source data fusion according to the present invention. Figure 2 This is a dynamic integrated cost fusion logic diagram of an embodiment of the present invention; Figure 3 This is an architecture diagram of a dynamic evacuation path planning system based on multi-source data fusion according to the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] Example Please see Figures 1-2 This invention provides a dynamic evacuation path planning method based on multi-source data fusion, comprising the following steps: S1. Construct a static grid map of the indoor environment, defining walkable areas, obstacle areas, and the locations of safety exits. First, digitally model the target building to generate a grid map. Each grid cell records its coordinates and type (corridor, room, wall, exit, etc.). Set basic movement costs, such as a cost of 1 for four-way movement and 1 for diagonal movement. .

[0021] S2. Collect and integrate multi-source dynamic data in real time to generate dynamic comprehensive value. The multi-source dynamic data includes at least: fire location and spread information, smoke concentration distribution information, regional population density information, and passage opening / closing or congestion status information.

[0022] Fire Data: Acquires the initial fire location from the building's automatic fire alarm system (heat and smoke detectors), or combines pre-stored building information with fire dynamics models (such as FDS) to perform real-time spread simulation, outputting the predicted temperature field for each grid cell. And the time of fire arrival. Define the fire hazard function: ,in The safe temperature threshold (e.g., 60℃). , These are the model parameters.

[0023] SmokeData: Real-time smoke concentration data for various areas obtained from networked smart smoke sensors. Define the smoke cost function: Visibility and Inversely proportional For coefficients, To prevent small constants from being divided by zero.

[0024] CrowdData: Real-time estimation of crowd density in various channels and areas using deployed cameras and computer vision algorithms (such as crowd counting and optical flow). Define the congestion delay function: (Simulating a nonlinear relationship where velocity is inversely proportional to density) is a coefficient.

[0025] PassageData: Obtains the open / closed status of doors from the electronic access control system, or determines whether a blockage has occurred via floor pressure sensors in critical passageways. Defines the passage status function: , where M is a relatively large cost.

[0026] The formula for calculating the dynamic comprehensive cost value is as follows: ; in, For grid nodes, The current time; This is a fire hazard function calculated based on a fire location and temperature model; The visibility / toxicity cost function is based on smoke concentration data; The congestion delay cost function is based on pedestrian density data; For channel state functions; These are the weighting coefficients for each cost item, and can be dynamically adjusted according to the stage of the disaster.

[0027] Fire hazard function and smoke cost function The input data comes from the real-time output of fire dynamics simulation software or from indoor temperature and smoke sensor networks; pedestrian density function The input data comes from the real-time crowd analysis results of the video surveillance system; channel state function The input data comes from the access control system status or pressure sensor data.

[0028] S3. Overlay the dynamic aggregated cost with the static grid map to construct a real-time dynamic network model, where the passage cost of each grid node is determined by both the base movement cost and the dynamic aggregated cost. Update the final passage cost of each walkable grid in the GridMap as follows: The cost of a non-walkable grid (wall) is infinite.

[0029] S4. Obtain the real-time location of all personnel to be evacuated at the current moment as a multi-source starting point, and all available safety exits as multi-source ending points. When a fire alarm is triggered, or when the system is running periodically, obtain the real-time coordinates of all personnel to be evacuated from the personnel positioning system (such as Wi-Fi, Bluetooth beacons, camera re-identification). This serves as the starting set, Sources. It retrieves all exits from the map that are not blocked by fire. , as the final set Sinks.

[0030] S5, Adopting an improved D-based approach The Lite algorithm, based on a real-time dynamic network model, calculates the safe evacuation path from each evacuee's starting point to the optimal exit in real time. Improvements include: S51. Introduce a search rectangle constraint based on the starting point and the target exit to dynamically limit the initial search range of the algorithm. The method for constructing and dynamically expanding the search rectangle constraint is as follows: Construct an initial rectangular search region with the starting point and the target exit as diagonal vertices; if the rectangle is divided by obstacles, causing the starting point and the exit to be disconnected, expand the rectangle boundary outward by a preset step size until the starting point and the exit are in the same connected region composed of walkable grids; use the final connected region as the constraint region for this path search.

[0031] S52. A two-layer risk avoidance mechanism is adopted. When calculating the cost of a node, the danger level of the current node and the danger information of its adjacent second-layer nodes are considered. The penalty term guides the path away from the potential danger area.

[0032] The two-layer risk avoidance mechanism modifies the nodes in the path planning algorithm. right-hand side value The calculation formula is implemented as follows: cost Penalty ; in, It is a node The set of successor nodes, From arrive The sum of the basic movement cost and the dynamic comprehensive cost. It is a node The current minimum cost, Penalty Based on nodes Second Neighborhood The penalty items for calculating the degree of internal danger, if If a high-risk node is found in memory, a penalty value greater than the difference between the maximum possible heuristic difference and the movement cost is applied. .

[0033] S53. Introduce virtual nodes to connect all actual exits, transforming the multi-source, multi-sink path planning problem into a multi-source, single-sink problem.

[0034] S54. A shared environment graph mechanism is adopted to store and reuse the path cost information of traversed nodes for subsequent path planning of other starting points, reducing redundant calculations. Value and When planning a path for a new evacuation starting point, the algorithm first queries the node information already stored in the shared environment graph and directly reuses valid information. Value and The value is only recalculated and updated for nodes whose information is invalid due to environmental changes or different starting points.

[0035] Create a virtual target node And match it with all actual exit nodes on the map. Connect them with edges of zero cost. The problem is transformed into connecting each Pi to... Path planning.

[0036] Sequential or parallel planning: Sort the starting point set (e.g., by distance from the fire source). Plan the path for the first starting point P1.

[0037] Search rectangle constraints: with P1 and Construct an initial rectangle R0 using the coordinates of (or its nearest actual exit). Check if P1 within R0 is connected to the target (based on the walkable area in the GridMap). If not connected, progressively expand the rectangle's boundary symmetrically to obtain... until the connected component Includes both. Subsequent searches are limited to Inside.

[0038] Incremental search and two-layer risk avoidance: at computing nodes of When the value is high, implement a two-layer risk avoidance strategy. Define nodes. Second Neighborhood (distance (For all nodes within 2 steps). Calculate the penalty term. if All nodes within All are below the safety threshold; otherwise .Revise The calculation is as follows: The penalty value P needs to be set to a constant greater than the maximum possible key value difference in the algorithm (e.g., calculated based on the heuristic function and the range of movement costs). (The value can be an integer, 4 or 5), to ensure that potentially risky nodes are prioritized later in the priority queue. After completing the path planning for P1, the values ​​of all visited and computed nodes during the algorithm are calculated. Value and The value is stored in the SharedMap.

[0039] Planning for the next starting point: During the planning phase for the next starting point P2, the algorithm initialization stage loads the existing node information in the SharedMap. Due to the environment model ( The value may have changed over time; the algorithm will check. and The system ensures consistency and only updates nodes and their affected areas that are invalidated due to environmental changes or from a new starting point perspective. Simultaneously, a new search rectangle constraint is established for P2. This process is repeated until all personnel have completed path planning.

[0040] S6. The planned dynamic path is sent to the corresponding evacuation personnel terminals or guidance devices in real time. For each person... Generate a series of ordered raster coordinate sequences as the path. Through indoor positioning and communication networks, The next key turning point or simplified directional instructions can be pushed to the person's smartphone app, or visually guided through smart evacuation signs (dynamic arrows, light strips) in the area where the person is located.

[0041] like Figure 3 The present invention also provides a dynamic evacuation path planning system based on multi-source data fusion, comprising: The data sensing and fusion module is used to collect data from multiple heterogeneous sensors and systems in real time, and perform fusion processing according to step S2 to output a dynamic comprehensive cost layer. It integrates multiple data adapters to receive raw data from fire alarm control panels, IoT smoke sensors, IP camera video streams, and access control controllers. Internally, it includes a data preprocessing submodule, a spatiotemporal registration submodule, and a fusion calculation submodule based on a weighted model.

[0042] The dynamic environment modeling module loads static raster maps and receives the dynamic comprehensive cost layer, overlaying them to generate a real-time dynamic network model. It includes a static map database and a real-time cost layer management submodule. It is responsible for maintaining... The matrix provides a unified query interface to the outside world.

[0043] The core path planning module incorporates an improved path planning algorithm. It receives the real-time dynamic network model, the real-time location of evacuees, and exit information, and performs parallel or serial real-time path calculations according to step S5. It incorporates the improved D algorithm of this invention. The Lite algorithm is implemented. This module supports multithreading or distributed computing and can handle path planning requests from multiple starting points simultaneously. It reads from the modeling module. It receives Sources and Sinks from the management platform and outputs PathSet.

[0044] The route distribution and guidance module is used to distribute planned personalized evacuation routes to user terminals, public displays, or audio-visual guidance devices in the form of graphics, instructions, or signals. It includes a user terminal interface and a field equipment control interface. The user terminal interface pushes the route to a mobile app via protocols such as HTTP / WebSocket. The field equipment control interface controls the display content and direction of electronic evacuation signs and audio-visual alarms via a dedicated network (such as a fire alarm bus).

[0045] Provides a human-computer interaction interface for monitoring the overall system's operational status, visualizing real-time disaster situation (heatmap, path overlay), and manually adjusting algorithm parameters (such as weights). ) and perform system configuration.

[0046] Therefore, the present invention adopts the above-mentioned dynamic evacuation route planning method and system based on multi-source data fusion, which integrates a comprehensive environmental model that incorporates multi-dimensional dynamic information such as fire, smoke, pedestrian flow, and passage status, and uses an efficient improved route planning algorithm to realize the real-time planning of safe, efficient, and personalized optimal evacuation routes for large-scale evacuation groups.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A dynamic evacuation path planning method based on multi-source data fusion, characterized in that, Includes the following steps: S1. Construct a static grid map of the indoor environment, defining walkable areas, obstacle areas, and the location of safety exits; S2. Real-time collection and fusion of multi-source dynamic data to generate dynamic comprehensive value. The multi-source dynamic data includes at least: fire location and spread information, smoke concentration distribution information, regional population density information, and passage opening / closing or congestion status information. S3. Overlay the dynamic comprehensive cost with the static grid map to construct a real-time dynamic network model, in which the passage cost of each grid node is determined by the basic movement cost and the dynamic comprehensive cost. S4. Obtain the real-time location of all personnel to be evacuated at the current moment as the multi-source starting point, and all available safety exits as the multi-sink endpoint; S5. Adopting an improved D The Lite algorithm, based on a real-time dynamic network model, calculates the safe evacuation path from each evacuee's starting point to the optimal exit in real time for each evacuee. S6. The planned dynamic path is sent to the corresponding evacuation personnel terminal or guidance device in real time.

2. The dynamic evacuation path planning method based on multi-source data fusion according to claim 1, characterized in that, In step S2, the formula for calculating the dynamic comprehensive cost value is as follows: ； in, For grid nodes, The current time; This is a fire hazard function calculated based on a fire location and temperature model; The visibility / toxicity cost function is based on smoke concentration data; The congestion delay cost function is based on pedestrian density data; For channel state functions; These are the weighting coefficients for each cost item, and can be dynamically adjusted according to the stage of the disaster.

3. The dynamic evacuation path planning method based on multi-source data fusion according to claim 2, characterized in that: Fire hazard function and smoke cost function The input data comes from the real-time output of fire dynamics simulation software or from indoor temperature and smoke sensor networks; pedestrian density function The input data comes from the real-time crowd analysis results of the video surveillance system; channel state function The input data comes from the access control system status or pressure sensor data.

4. The dynamic evacuation path planning method based on multi-source data fusion according to claim 1, characterized in that, The improvements to the algorithm in step S5 include: S51. Introduce a search rectangle constraint based on the starting point and the target exit to dynamically limit the initial search range of the algorithm; S52. A two-layer risk avoidance mechanism is adopted. When calculating the cost of a node, the danger level of the current node and the danger information of its adjacent second-layer nodes are considered. The penalty term guides the path away from the potential danger area. S53. Introduce virtual nodes to connect all actual exits, transforming the multi-source, multi-sink path planning problem into a multi-source, single-sink problem; S54. A shared environment graph mechanism is adopted to store and reuse the path cost information of traversed nodes for subsequent path planning of other starting points, thereby reducing redundant calculations.

5. The dynamic evacuation path planning method based on multi-source data fusion according to claim 4, characterized in that, The method for constructing and dynamically expanding the search rectangle constraint in step S51 is as follows: Construct an initial rectangular search region with the starting point and the target exit as diagonal vertices; if the rectangle is divided by obstacles, causing the starting point and the exit to be disconnected, expand the rectangle boundary outward by a preset step size until the starting point and the exit are in the same connected region composed of walkable grids; use the final connected region as the constraint region for this path search.

6. The dynamic evacuation path planning method based on multi-source data fusion according to claim 5, characterized in that, In step S52, the two-layer risk avoidance mechanism modifies the nodes in the path planning algorithm. right-hand side value The calculation formula is implemented as follows: ； in, It is a node The set of successor nodes, From arrive The sum of the basic movement cost and the dynamic comprehensive cost. It is a node The current minimum cost, Penalty Based on nodes Second Neighborhood The penalty items for calculating the degree of internal danger, if If a high-risk node is found in memory, a penalty value greater than the difference between the maximum heuristic difference and the movement cost that occurs in the algorithm is applied. .

7. The dynamic evacuation path planning method based on multi-source data fusion according to claim 6, characterized in that: In step S54, the shared environment graph is a global data structure that stores all the grid nodes that have been visited during the planning process. Value and When planning a path for a new evacuation starting point, the algorithm first queries the node information already stored in the shared environment graph and directly reuses valid information. Value and The value is only recalculated and updated for nodes whose information is invalid due to environmental changes or different starting points.

8. A dynamic evacuation route planning system based on multi-source data fusion, employing the dynamic evacuation route planning method based on multi-source data fusion as described in any one of claims 1-7, characterized in that: include: The data sensing and fusion module is used to collect data from multiple heterogeneous sensors and systems in real time, and perform fusion processing according to step S2 to output a dynamic comprehensive cost layer. The dynamic environment modeling module is used to load static grid maps and receive dynamic comprehensive cost layers, which are then overlaid to generate a real-time dynamic network model. The core path planning module has an improved path planning algorithm built in. It receives real-time dynamic network models, real-time locations of evacuees and exit information, and performs parallel or serial real-time path calculations according to step S5. The route distribution and guidance module is used to distribute the planned personalized evacuation routes to user terminals, public displays, or audio-visual guidance devices in the form of graphics, instructions, or signals.

Images