Disaster evacuation route planning method and system for vulnerable people
By constructing a knowledge graph of behavioral influencing factors and making predictions, combined with macro evacuation strategies, the problem that traditional algorithms are difficult to accurately analyze the behavior of vulnerable people is solved, and a more efficient and safe evacuation route planning is achieved.
Patent Information
- Application Number
- CN202411610500.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-02-18
AI Technical Summary
The traditional AlphaPose algorithm is difficult to accurately capture and analyze the posture and motion information of vulnerable people, resulting in insufficient accuracy in behavioral modeling and prediction of vulnerable people in complex and dynamic evacuation scenarios.
By obtaining the behavioral characteristics and movement trajectories of vulnerable people in complex dynamic evacuation scenarios, a knowledge graph of behavioral influencing factors is constructed, and behavior prediction is carried out based on this knowledge graph. Finally, the prediction results are combined with the macro evacuation strategy to build the optimal evacuation route.
The accuracy and adaptability of behavior prediction of vulnerable populations have been improved, the evacuation route has been optimized, and the safety and efficiency of evacuation have been improved.
Smart Images
Figure CN119476725A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of information processing technology, and in particular relates to a method and system for planning a disaster evacuation route for vulnerable groups. Background Art
[0002] During disaster evacuation, the behavioral characteristics and movement trajectories of vulnerable groups are significantly different from those of the general population. The traditional AlphaPose algorithm has difficulty in accurately capturing and analyzing the posture and motion information of vulnerable groups, resulting in insufficient accuracy in behavioral modeling and prediction of vulnerable groups in complex and dynamic evacuation scenarios. At the same time, the behavior of vulnerable groups is affected by multiple factors such as environmental factors and individual characteristics. How to comprehensively consider these factors and build a comprehensive and accurate micro-behavior model is a technical problem that needs to be solved urgently. In addition, in the actual evacuation route planning, how to combine the micro-behavior model of vulnerable groups with the macro-evacuation strategy to optimize the overall evacuation efficiency and safety also faces many technical challenges. Therefore, there is an urgent need for a method and system for disaster evacuation route planning for vulnerable groups. Summary of the invention
[0003] In order to solve the above technical problems, the present invention proposes a disaster evacuation route planning method and system for vulnerable groups, which is of great significance for improving the scientificity and effectiveness of disaster response and rescue.
[0004] The present invention provides a method for planning a disaster evacuation route for vulnerable groups, comprising:
[0005] Obtain the behavioral characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios;
[0006] Based on the behavior characteristics and movement trajectories, a knowledge graph of behavior influencing factors is constructed;
[0007] Based on the knowledge graph, predict the behavior of the vulnerable group and obtain prediction results;
[0008] The prediction results are combined with the macro evacuation strategy to construct the optimal evacuation route.
[0009] Optionally, based on the behavior characteristics and movement trajectory, constructing a behavior influencing factor knowledge graph includes:
[0010] Constructing a behavior prediction model, wherein the behavior prediction model is trained by a training set, and the training set is behavior features and movement trajectories;
[0011] Obtaining behavior analysis results of vulnerable groups through the behavior prediction model;
[0012] According to the behavior analysis results of the vulnerable groups, a knowledge graph of the behavior influencing factors is constructed.
[0013] Optionally, obtaining the training set includes:
[0014] Extracting and selecting the behavioral features to obtain a key feature subset that can reflect the behavioral patterns of vulnerable groups;
[0015] Based on the movement trajectory, obtaining a movement trajectory sequence;
[0016] The training set is obtained based on the key feature subset and the movement trajectory sequence.
[0017] Optionally, based on the behavior analysis results of the vulnerable groups, constructing the behavior influencing factor knowledge graph includes:
[0018] Based on the behavioral analysis results, obtain behavioral data of vulnerable groups;
[0019] Extracting features from the behavior data of the vulnerable group to obtain a behavior feature set;
[0020] According to the behavior feature set, an association rule mining algorithm is used to obtain an association rule library;
[0021] Based on the association rule base, a causal reasoning algorithm is used to construct a causal relationship graph;
[0022] Based on the cause-effect relationship diagram, a knowledge graph of behavior influencing factors is constructed.
[0023] Optionally, feature extraction is performed on the behavior data of the vulnerable group to obtain a behavior feature set including:
[0024] Extract behavioral features, environmental features and individual features to construct a multi-dimensional feature vector;
[0025] The principal component analysis method is used to reduce the dimension of the multi-dimensional feature vector to obtain the behavior feature set.
[0026] Optionally, based on the knowledge graph, predicting the behavior of the vulnerable group, obtaining the prediction result includes:
[0027] According to the knowledge graph of the behavior influencing factors, key nodes and relationships are extracted as prior knowledge for the behavior prediction model of vulnerable groups;
[0028] Dynamically adjust and optimize the structure and parameters of the behavior prediction model according to the key nodes and relationships in the behavior influencing factor knowledge graph;
[0029] Based on the optimized behavior prediction model, the prediction result is obtained.
[0030] Optionally, dynamically adjusting and optimizing the structure and parameters of the behavior prediction model according to key nodes and relationships in the behavior influencing factor knowledge graph includes:
[0031] According to the key nodes and relationships in the behavior influencing factor knowledge graph, the initial structure and parameters of the prediction model are obtained;
[0032] Compare the prediction result with the actual action behavior data, calculate the prediction error, and trigger model optimization if the prediction error is greater than a preset threshold;
[0033] Obtain sample data with large prediction errors, adjust the model structure according to the knowledge graph node relationship corresponding to the sample data, and use the gradient descent algorithm to update the model parameters;
[0034] The optimized model structure and parameters are reapplied to the behavior prediction of vulnerable groups, and the above steps are repeated until the prediction error converges to the preset range.
[0035] Optionally, the prediction result is combined with a macro evacuation strategy to construct the optimal evacuation route;
[0036] Obtaining geographic information and population distribution of the evacuation area, and combining the macro evacuation strategy to build a multi-level and multi-granular evacuation route planning framework;
[0037] The optimal evacuation route is constructed based on the evacuation route planning framework and the prediction results.
[0038] The present invention also provides a disaster evacuation route planning system for vulnerable groups, including: a data collection module, a knowledge graph construction module, a behavior prediction module and a route planning module;
[0039] The data acquisition module is used to obtain the behavior characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios;
[0040] The knowledge graph construction module is used to construct an influencing factor knowledge graph according to the behavior characteristics and movement trajectory;
[0041] The behavior prediction module is used to perform behavior prediction based on the knowledge graph;
[0042] The route planning module is used to combine the prediction results with the macro evacuation strategy to construct an optimal evacuation route.
[0043] Compared with the prior art, the present invention has the following advantages and technical effects:
[0044] The present invention models the behavioral characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios through deep learning algorithms to construct a behavior prediction model. The improved AlphaPose algorithm is used to enhance the analysis of the posture and motion information of vulnerable groups, and a knowledge graph of behavioral influencing factors is constructed. Through incremental learning and transfer learning methods, the behavior prediction model is dynamically updated and optimized to improve its adaptability and prediction accuracy. The model is combined with the macro evacuation strategy to construct a multi-level and multi-granular evacuation route planning framework to achieve dynamic generation and real-time optimization of the route. The present invention integrates functions such as behavior analysis, route planning, simulation evaluation, etc., and provides decision support for rescue personnel through human-computer interaction and visualization technology, effectively improving the safety and efficiency of disaster evacuation of vulnerable groups. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0046] Figure 1 The present invention is a flowchart of a method for planning a disaster evacuation route for vulnerable groups according to an embodiment of the present invention. DETAILED DESCRIPTION
[0047] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0048] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0049] The present invention proposes a method for planning a disaster evacuation route for vulnerable groups, such as Figure 1 As shown, the specific steps include:
[0050] Obtain the behavioral characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios;
[0051] Based on behavioral characteristics and movement trajectories, a knowledge graph of behavioral influencing factors is constructed;
[0052] Based on the knowledge graph, predict the behavior of vulnerable groups and obtain prediction results;
[0053] Combine the prediction results with the macro-evacuation strategy to construct the optimal evacuation route.
[0054] Furthermore, based on the behavior characteristics and movement trajectories, the knowledge graph of behavior influencing factors is constructed, including:
[0055] Constructing a behavior prediction model, wherein the behavior prediction model is trained by a training set, and the training set is behavior features and movement trajectories;
[0056] Obtain behavioral analysis results of vulnerable groups through behavioral prediction models;
[0057] Based on the results of behavioral analysis of vulnerable groups, a knowledge graph of behavioral influencing factors is constructed.
[0058] Specifically, based on the historical behavior data of vulnerable groups, a behavior prediction model for vulnerable groups is constructed to predict the future behavior trends of vulnerable groups. Real-time video surveillance data of vulnerable groups is obtained, and the improved AlphaPose algorithm is used to estimate the posture of vulnerable groups in the video. The attention mechanism is introduced into the AlphaPose algorithm, and the key areas related to the behavior of vulnerable groups in the video frame are automatically learned through the attention module to enhance the algorithm's ability to capture the posture of vulnerable groups. Spatiotemporal context information is introduced into the AlphaPose algorithm, and the algorithm's ability to analyze the motion information of vulnerable groups is enhanced by analyzing the temporal continuity and spatial correlation of the posture of vulnerable groups in the video sequence. The posture and motion information of vulnerable groups obtained by the improved AlphaPose algorithm are integrated with the behavior prediction model of vulnerable groups to generate enhanced behavior analysis results of vulnerable groups. According to the enhanced behavior analysis results of vulnerable groups, it is judged whether the behavior of vulnerable groups is abnormal. If there is abnormal behavior, the early warning mechanism is triggered to notify the relevant departments to take intervention measures in time. The behavior prediction model of vulnerable groups and the AlphaPose algorithm are continuously optimized, and the system's ability to predict and analyze the behavior of vulnerable groups is continuously improved through incremental learning, so as to provide more accurate care services for vulnerable groups.
[0059] Furthermore, obtaining a training set includes:
[0060] Extract and select behavioral features to obtain a subset of key features that can reflect the behavioral patterns of vulnerable groups;
[0061] Based on the moving trajectory, a moving trajectory sequence is obtained;
[0062] Based on the key feature subset and the movement trajectory sequence, a training set is obtained.
[0063] Furthermore, based on the analysis results of the behavior of vulnerable groups, a knowledge graph of behavior influencing factors is constructed, including:
[0064] Based on the results of behavioral analysis, obtain behavioral data of vulnerable groups;
[0065] Extract features from the behavior data of vulnerable groups to obtain a set of behavioral features;
[0066] According to the behavioral feature set, an association rule mining algorithm is used to obtain an association rule base;
[0067] Based on the association rule base, the causal reasoning algorithm is used to construct a causal relationship graph;
[0068] Based on the cause-effect relationship diagram, a knowledge graph of behavior influencing factors is constructed.
[0069] Specifically, the behavior data of vulnerable groups are obtained, and the behavior data are preprocessed to remove noise data, fill missing values, and correct outliers to obtain a cleaned vulnerable group behavior data set. Feature engineering is performed on the cleaned vulnerable group behavior data set to extract behavioral features, environmental features, and individual features, construct a multi-dimensional feature vector, and use principal component analysis to reduce the dimension to obtain a vulnerable group behavior feature set after dimensionality reduction. Based on the vulnerable group behavior feature set, an association rule mining algorithm is used to discover the association rules between different behavior patterns and obtain a vulnerable group behavior association rule base. Based on the vulnerable group behavior association rule base, a causal reasoning algorithm is used to infer the causal relationship between different influencing factors, construct a causal relationship graph, and reveal the mechanism of action of the influencing factors. The vulnerable group behavior association rules and the causal relationship graph are integrated to construct a knowledge graph of the influencing factors of the vulnerable group behavior. A graph embedding algorithm is used to learn the low-dimensional vector representation of nodes and edges in the knowledge graph to obtain a knowledge graph embedding vector. Based on the knowledge graph embedding vector, a graph neural network algorithm is used to perform representation learning and relational reasoning on the influencing factors of the vulnerable group behavior to reveal the interaction pattern of multiple influencing factors. Based on the reasoning results of graph neural networks, an analysis report on factors affecting the behavior of vulnerable groups is generated, which explains the impact mechanism of different factors on the behavior of vulnerable groups and provides a decision-making basis for relevant departments to formulate policies to support vulnerable groups.
[0070] Furthermore, feature extraction is performed on the behavior data of vulnerable groups, and the behavior feature set obtained includes:
[0071] Extract behavioral features, environmental features and individual features to construct a multi-dimensional feature vector;
[0072] The principal component analysis method is used to reduce the dimension of the multi-dimensional feature vector and obtain the behavioral feature set.
[0073] Furthermore, based on the knowledge graph, the behavior of vulnerable groups is predicted, and the prediction results include:
[0074] Based on the knowledge graph of behavioral influencing factors, key nodes and relationships are extracted as prior knowledge for the behavior prediction model of vulnerable groups;
[0075] Dynamically adjust and optimize the structure and parameters of the behavior prediction model based on the key nodes and relationships in the knowledge graph of behavior influencing factors;
[0076] Based on the optimized behavior prediction model, the prediction results are obtained.
[0077] Specifically, the behavior data and influencing factor data of vulnerable groups at different evacuation stages are obtained to construct a vulnerable group behavior data set and an influencing factor data set. According to the knowledge graph of influencing factors of vulnerable group behavior, key nodes and relationships are extracted as prior knowledge of the vulnerable group behavior prediction model. The incremental learning method is used to dynamically add the newly acquired vulnerable group behavior data and influencing factor data to the training data set, and the vulnerable group behavior prediction model is incrementally trained and optimized. The transfer learning method is used to transfer the knowledge of the vulnerable group behavior prediction model in similar scenarios to the current model to improve the generalization ability and prediction accuracy of the model. According to the key nodes and relationships in the knowledge graph of influencing factors of vulnerable group behavior, the structure and parameters of the vulnerable group behavior prediction model are dynamically adjusted and optimized. Through continuous learning and optimization of the vulnerable group behavior prediction model, the adaptability and prediction accuracy of the model to the changes in vulnerable group behavior are continuously improved. The optimized vulnerable group behavior prediction model is applied to the actual vulnerable group evacuation scenario to provide personalized evacuation guidance and services for vulnerable groups and improve the safety level of vulnerable groups.
[0078] Furthermore, according to the key nodes and relationships in the knowledge graph of behavior influencing factors, the structure and parameters of the behavior prediction model are dynamically adjusted and optimized, including:
[0079] Obtain the initial structure and parameters of the prediction model based on the key nodes and relationships in the knowledge graph of behavior influencing factors;
[0080] Compare the prediction results with the actual action behavior data and calculate the prediction error. If the prediction error is greater than the preset threshold, the model optimization is triggered.
[0081] Obtain sample data with large prediction errors, adjust the model structure according to the knowledge graph node relationship corresponding to the sample data, and use the gradient descent algorithm to update the model parameters;
[0082] The optimized model structure and parameters are reapplied to the behavior prediction of vulnerable groups, and the above steps are repeated until the prediction error converges to the preset range.
[0083] Specifically, obtain the key nodes and relationship data in the knowledge graph of factors affecting the behavior of vulnerable groups, and construct the initial structure and parameters of the vulnerable group behavior prediction model based on the node relationship data; use the prediction model to predict the behavior data of vulnerable groups and obtain the prediction results; compare the prediction results with the actual behavior data, calculate the prediction error, and trigger model optimization if the prediction error is greater than the preset threshold; obtain sample data with large prediction errors, adjust the structure of the model according to the knowledge graph node relationship corresponding to the sample data, and use the gradient descent algorithm to update the parameters of the model; reapply the optimized model structure and parameters to the vulnerable group behavior prediction, and repeat the above steps until the prediction error converges to an acceptable range; for the optimized model, use the cross-validation method to evaluate the generalization performance of the model and determine the optimal structure and parameter combination of the model; apply the optimal model to the actual vulnerable group behavior prediction scenario, continuously monitor the prediction effect of the model, and regularly trigger the re-optimization of the model according to the update of the knowledge graph.
[0084] Furthermore, the prediction results are combined with the macro evacuation strategy to construct the optimal evacuation route;
[0085] Obtain geographic information and population distribution of the evacuation area, and build a multi-level and multi-granular evacuation route planning framework based on macro-evacuation strategies;
[0086] Based on the evacuation route planning framework and the prediction results, the optimal evacuation route is constructed.
[0087] Specifically, based on the historical behavior data of vulnerable groups, a machine learning algorithm is used to establish a behavior prediction model for vulnerable groups, and the model is used to predict the behavioral characteristics and movement trajectory of vulnerable groups during the evacuation process. The geographic information, population distribution and other data of the evacuation area are obtained, and combined with the macro evacuation strategy, a multi-level and multi-granular evacuation route planning framework is constructed. The framework includes regional division at the macro level and individual motion planning at the micro level. At the micro level, according to the behavior prediction model of vulnerable groups, the intelligent agent modeling method is used to simulate the individual's motion trajectory to obtain the micro evacuation route. At the macro level, the graph theory algorithm is used to optimize the road network in the region to obtain the macro evacuation route, and it is used as the boundary condition of the micro simulation. Through the interactive iteration of micro simulation and macro optimization algorithm, the evacuation route is continuously updated and optimized. In each iteration, the macro optimization target is adjusted according to the micro simulation results, and the optimization results are fed back to the micro simulation to form a closed-loop control. During the evacuation process, the location and status information of vulnerable groups is obtained in real time to determine whether they deviate from the preset route. If they deviate, the dynamic adjustment mechanism of the route is triggered to replan the evacuation route according to the current situation. Continuously monitor the evacuation process and obtain real-time data on traffic flow, congestion, etc. If the evacuation efficiency is low or there are safety hazards, the emergency optimization mechanism will be activated to adjust the evacuation route in real time to ensure that the evacuation work is safe and efficient.
[0088] The present invention also provides a disaster evacuation route planning system for vulnerable groups, including: a data collection module, a knowledge graph construction module, a behavior prediction module and a route planning module;
[0089] Data collection module, used to obtain the behavioral characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios;
[0090] The knowledge graph construction module is used to construct a knowledge graph of influencing factors based on behavioral characteristics and movement trajectories;
[0091] Behavior prediction module, used to predict behavior based on knowledge graph;
[0092] The route planning module is used to combine the prediction results with the macro evacuation strategy to construct the optimal evacuation route.
[0093] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A method for planning a disaster evacuation route for vulnerable groups, characterized in that: include: Obtain the behavioral characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios; Based on the behavior characteristics and movement trajectories, a knowledge graph of behavior influencing factors is constructed; Based on the knowledge graph, predict the behavior of the vulnerable group and obtain prediction results; The prediction results are combined with the macro evacuation strategy to construct the optimal evacuation route.
2. A method for planning a disaster evacuation route for vulnerable groups according to claim 1, characterized in that: Based on the behavior characteristics and movement trajectories, a knowledge graph of behavior influencing factors is constructed, including: Constructing a behavior prediction model, wherein the behavior prediction model is trained by a training set, and the training set is behavior features and movement trajectories; Obtaining behavior analysis results of vulnerable groups through the behavior prediction model; According to the behavior analysis results of the vulnerable groups, a knowledge graph of the behavior influencing factors is constructed.
3. A method for planning a disaster evacuation route for vulnerable groups according to claim 2, characterized in that: Acquiring the training set includes: Extracting and selecting the behavioral features to obtain a key feature subset that can reflect the behavioral patterns of vulnerable groups; Based on the movement trajectory, obtaining a movement trajectory sequence; The training set is obtained based on the key feature subset and the movement trajectory sequence.
4. A method for planning a disaster evacuation route for vulnerable groups according to claim 2, characterized in that: According to the behavior analysis results of the vulnerable groups, constructing the knowledge graph of the behavior influencing factors includes: Based on the behavioral analysis results, obtain behavioral data of vulnerable groups; Extracting features from the behavior data of the vulnerable group to obtain a behavior feature set; According to the behavior feature set, an association rule mining algorithm is used to obtain an association rule library; Based on the association rule base, a causal reasoning algorithm is used to construct a causal relationship graph; Based on the cause-effect relationship diagram, a knowledge graph of behavior influencing factors is constructed.
5. A method for planning a disaster evacuation route for vulnerable groups according to claim 4, characterized in that: Feature extraction is performed on the behavior data of the vulnerable group to obtain a behavior feature set including: Extract behavioral features, environmental features and individual features to construct a multi-dimensional feature vector; The principal component analysis method is used to reduce the dimension of the multi-dimensional feature vector to obtain the behavior feature set.
6. A method for planning a disaster evacuation route for vulnerable groups according to claim 2, characterized in that: Based on the knowledge graph, the behavior of the vulnerable group is predicted, and the prediction results obtained include: According to the knowledge graph of the behavior influencing factors, key nodes and relationships are extracted as prior knowledge for the behavior prediction model of vulnerable groups; Dynamically adjust and optimize the structure and parameters of the behavior prediction model according to the key nodes and relationships in the behavior influencing factor knowledge graph; Based on the optimized behavior prediction model, the prediction result is obtained.
7. A method for planning a disaster evacuation route for vulnerable groups according to claim 6, characterized in that: According to the key nodes and relationships in the behavior influencing factor knowledge graph, dynamically adjusting and optimizing the structure and parameters of the behavior prediction model includes: S1. Obtaining the initial structure and parameters of the prediction model according to the key nodes and relationships in the behavior influencing factor knowledge graph; S2. Compare the prediction result with the actual action behavior data and calculate the prediction error. If the prediction error is greater than a preset threshold, model optimization is triggered; S3, obtaining sample data with large prediction errors, adjusting the model structure according to the knowledge graph node relationship corresponding to the sample data, and updating the model parameters using the gradient descent algorithm; S4. Reapply the optimized model structure and parameters to the prediction of vulnerable population behavior, and repeat the above steps S1-S5 until the prediction error converges to the preset range.
8. A method for planning a disaster evacuation route for vulnerable groups according to claim 6, characterized in that: Combining the prediction results with the macro evacuation strategy to construct the optimal evacuation route; Obtaining geographic information and population distribution of the evacuation area, and combining the macro evacuation strategy to build a multi-level and multi-granular evacuation route planning framework; The optimal evacuation route is constructed based on the evacuation route planning framework and the prediction results.
9. A disaster evacuation route planning system for vulnerable groups, characterized in that: include: Data collection module, knowledge graph construction module, behavior prediction module and route planning module; The data acquisition module is used to obtain the behavior characteristics and movement trajectories of vulnerable groups in complex dynamic evacuation scenarios; The knowledge graph construction module is used to construct an influencing factor knowledge graph according to the behavior characteristics and movement trajectory; The behavior prediction module is used to perform behavior prediction based on the knowledge graph; The route planning module is used to combine the prediction results with the macro evacuation strategy to construct the optimal evacuation route.