Isolated forest model-based shelter identification method and apparatus, and medium

Abstract

The invention discloses a shelter identification method and device based on an isolated forest model, and a medium, and the method comprises the steps: collecting Wi-Fi handshake data in a non-disaster period, and carrying out the data preprocessing, and obtaining a training data set; constructing an isolated forest model of the informal shelter according to the training data set; inputting the Wi-Fi handshake data of the informal shelter in the disaster period into the isolated forest model of the informal shelter to obtain an abnormal score of each data point; and identifying potential shelters according to the abnormal scores, and allocating rescue resources for the shelters. According to the method, the crowd gathering condition of an informal shelter during a disaster can be automatically identified, the monitoring capability of a refuge crowd gathering area is greatly improved, equipment connection and crowd activities during the disaster are monitored in real time by using Wi-Fi handshake data, the method does not depend on additional physical infrastructures, the equipment laying cost is saved, and the method is suitable for popularization and application. The application prospect is wide.

Description

Technical Field

[0001] The present invention relates to the field of public safety, and particularly to a method, device and medium for identifying a shelter based on an isolation forest model. Background Art

[0002] A shelter is a measure for resettling disaster victims in response to emergencies, and is also a safe shelter in modern big cities for the public to take shelter from major natural disasters such as earthquakes, fires, explosions, floods, etc. When a disaster occurs, many residents will choose to take shelter in informal shelters, such as shopping malls, parking lots, parks, etc. These locations are usually not within the scope of official monitoring and are difficult to detect. If the emergency rescue system cannot grasp the crowd gathering situation in these informal shelters in real time, it will lead to unreasonable resource allocation and response delays, increasing the complexity and difficulty of the rescue work. However, in actual emergency management, traditional disaster monitoring and emergency response systems mainly rely on the monitoring of physical infrastructure, such as cameras or on-site reports from personnel, and there are still problems such as monitoring blind spots and delayed responses. Existing methods for obtaining information on disaster rescue needs in real time mostly rely on satellite remote sensing images and on-site surveys by drones. The former is difficult to support the rescue needs of complex weather disasters accompanied by rainy weather, and the latter has limited spatial coverage and high costs.

[0003] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, device and medium for identifying a shelter based on an isolation forest model in view of the above-mentioned defects of the existing technology, aiming to solve the problems that the disaster detection equipment in the existing technology cannot meet the rescue needs of complex weather disasters, and has limited spatial coverage and high costs.

[0005] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0006] In a first aspect, the present invention provides a method for identifying a shelter based on an isolation forest model,

[0007] wherein the method includes:

[0008] Collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training data set;

[0009] Construct an isolation forest model for informal shelters according to the training data set, wherein the informal shelters are shelters excluding official shelters and regular residences;

[0010] Input the Wi-Fi handshake data of informal shelters during disaster periods into the isolation forest model of informal shelters to obtain the anomaly score of each data point;

[0011] Potential refuge sites are identified based on the anomaly scores, and rescue resources are allocated to the refuge sites.

[0012] In one implementation, the collecting of Wi-Fi handshake data during a non-disaster period and performing data preprocessing to obtain a training data set includes:

[0013] Collect Wi-Fi handshake data during non-disaster periods, where the Wi-Fi handshake data is four-way handshake request data between a terminal device and a Wi-Fi access point, including device ID, access point location, and connection time;

[0014] The Wi-Fi handshake data collected during the non-disaster period was cleaned and standardized to obtain standardized data;

[0015] Collecting POI data, and performing grid processing on the POI data to generate POI features of the grid;

[0016] Matching the standardized data with the POI features of the grid to obtain enhanced training data;

[0017] The enhanced training data is divided according to a preset time window to obtain the training data set.

[0018] In one implementation, the step of cleaning and standardizing the collected Wi-Fi handshake data during the non-disaster period to obtain standardized data includes:

[0019] Among the Wi-Fi handshake data in the non-disaster period, the Wi-Fi handshake data with access points located at regular residences and official shelters are identified and removed according to a geographic information system database to obtain removed data;

[0020] The eliminated data is standardized to obtain the standardized data.

[0021] In one implementation, constructing an isolation forest model of informal shelters based on the training data set includes:

[0022] In each time window, randomly selecting feature vectors and split points from the training data set, and constructing an isolation tree based on the feature vectors and split points, wherein the feature vectors include device connection frequency, device connection duration, device density, and access point location, and the split points are used to isolate data;

[0023] Repeat the steps of randomly selecting feature vectors and split points from the training data set in each time window, and constructing an isolated tree according to the feature vectors and split points, until a termination condition is reached, to obtain a plurality of isolated trees;

[0024] Based on the isolated trees, an isolation forest model of informal shelters was constructed.

[0025] In one implementation, the Wi-Fi handshake data of informal shelters during the disaster period is input into the isolation forest model of informal shelters to obtain an abnormal score for each data point, including:

[0026] Obtain Wi-Fi handshake data during the disaster period. According to the geographic information system database, remove the Wi-Fi handshake data where the access point is located at a regular residence and official shelter, and obtain the Wi-Fi handshake data of informal shelters during the disaster period.

[0027] Inputting the Wi-Fi handshake data of the informal shelters during the disaster period into the isolation forest model of informal shelters, and calculating the average path length and expected path length from the Wi-Fi handshake data of the informal shelters during the disaster period to each isolated tree;

[0028] The anomaly score of each data point is calculated according to the ratio of the expected path length to the average path length.

[0029] In one implementation, the identifying a potential shelter according to the anomaly score and allocating rescue resources for the shelter includes:

[0030] Calculate the anomaly strength of each Wi-Fi hotspot according to the anomaly score of each data point, wherein the anomaly strength is the average of the anomaly scores of the data points connected to each Wi-Fi hotspot;

[0031] Comparing the abnormality intensity with a preset abnormality detection threshold, if the abnormality intensity is greater than or equal to the abnormality detection threshold, marking the geographical location corresponding to the Wi-Fi hotspot as a potential refuge place;

[0032] generating a geographic heat map based on the anomaly intensity of the potential refuge sites;

[0033] Rescue resources are deployed according to the geographic heat map.

[0034] In one implementation, generating a geographic heat map according to the abnormal intensity of the potential refuge places includes:

[0035] Creating a three-dimensional scene based on a geographic information system database, associating the coordinates of the potential refuge site with the coordinates in the three-dimensional scene to obtain a target location;

[0036] Combining the target location with the abnormal intensity of the corresponding potential shelter and mapping it into the three-dimensional scene to generate the geographical heat map.

[0037] In a second aspect, an embodiment of the present invention further provides a shelter identification device based on an isolation forest model. The device includes:

[0038] A training dataset acquisition module, configured to collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training dataset;

[0039] A model training module, configured to construct an isolation forest model of informal shelters according to the training dataset, where the informal shelters are shelters excluding official shelters and regular residences;

[0040] An abnormal score calculation module, configured to input the Wi-Fi handshake data of informal shelters during disaster periods into the isolation forest model of informal shelters to obtain the abnormal score of each data point;

[0041] A potential shelter identification module, configured to identify potential shelters according to the abnormal scores and allocate rescue resources to the shelters.

[0042] In a third aspect, an embodiment of the present invention further provides an intelligent terminal. The intelligent terminal includes a memory, a processor, and a shelter identification program based on an isolation forest model stored in the memory and executable on the processor. When the processor executes the shelter identification program based on the isolation forest model, the steps of the shelter identification method based on the isolation forest model as described in any one of the above are implemented.

[0043] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium. A shelter identification program based on an isolation forest model is stored on the computer-readable storage medium. When the shelter identification program based on the isolation forest model is executed by a processor, the steps of the shelter identification method based on the isolation forest model as described in any one of the above are implemented.

[0044] Beneficial effects: Compared with the prior art, the present invention provides a method for identifying shelters based on the Isolation Forest model. First, Wi-Fi handshake data during non-disaster periods is collected and preprocessed to obtain a training data set. By using Wi-Fi handshake data, device connections and crowd activities during disasters are monitored in real time, without relying on additional physical infrastructure, and can cover the dynamics of people in the city faster and more widely. Then, an Isolation Forest model for informal shelters is constructed based on the training data set, and by excluding the hotspots of residential areas and official shelters, monitoring blind spots can be discovered. Next, the Wi-Fi handshake data of informal shelters during disasters is input into the Isolation Forest model of informal shelters to obtain the anomaly score of each data point. The intelligent algorithm enables the system to quickly and accurately detect the hotspots where evacuating crowds gather, providing real-time decision support for emergency response. Finally, potential shelters are identified based on the anomaly scores, and rescue resources are allocated to these shelters to help the rescue team prioritize responses to high-risk areas and improve rescue efficiency. Description of the Drawings

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0046] Figure 1 It is a schematic flow chart of a method for identifying shelters based on the Isolation Forest model provided by an embodiment of the present invention.

[0047] Figure 2 It is a principle block diagram of a device for identifying shelters based on the Isolation Forest model provided by an embodiment of the present invention.

[0048] Figure 3 It is a principle block diagram of the internal structure of an intelligent terminal provided by an embodiment of the present invention. Detailed Embodiments

[0049] To make the objectives, technical solutions and effects of the present invention clearer and more definite, the following further describes the present invention in detail with reference to the accompanying drawings and by way of examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0050] Those skilled in the art can understand that, unless specifically stated, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present invention means the presence of the stated features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0051] Those skilled in the art can understand that, unless otherwise defined, all terms used herein (including technical terms and scientific terms) have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention pertains. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with their meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0052] In actual emergency management, traditional disaster monitoring and emergency response systems mainly rely on the monitoring of physical infrastructure, such as cameras or on-site reports by personnel, and there are problems such as monitoring blind spots and delayed responses. With the acceleration of the urbanization process, the disaster emergency system is facing increasingly complex challenges, and it is urgent to improve the emergency response efficiency by means of digital technology.

[0053] The mobile phone Wi-Fi (Wireless Fidelity, mobile hotspot) hotspot is a technology that converts the GPRS, 3G or 4G signals received by the mobile phone into Wi-Fi signals and sends them out, enabling portable devices such as mobile phones, tablets or laptops to access the Internet outdoors or in places without a network through a wireless network card or a WLAN module. During the Wi-Fi connection establishment process, four handshakes are required. The four-way handshake of Wi-Fi is a process in which the terminal device and the Wi-Fi base station exchange four messages. The four-way handshake is to generate a key for encrypting wireless data to ensure the establishment of a data link. By intelligently analyzing the Wi-Fi handshake protocol data, abnormal population activities during disasters can be effectively detected, and crowd gatherings at unofficial shelters can be identified in real time, and no additional hardware devices need to be installed.

[0054] Exemplary method

[0055] This embodiment provides a shelter identification based on the Isolation Forest model. As Figure 1As shown, the method includes the following steps:

[0056] Step S100: Collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training data set;

[0057] Specifically, a Wi-Fi hotspot is a wireless signal transmitting device that converts a wired network into a wireless network through wireless means, enabling electronic devices to connect to the Internet. During non-disaster periods, there is no abnormal aggregation of people. By collecting Wi-Fi handshake data during non-disaster periods and performing data cleaning and standardization, and removing invalid or redundant handshake data, the aggregation situation of people at specific locations during non-disaster periods can be grasped in real time.

[0058] In one implementation, step S100 of this embodiment includes the following steps:

[0059] Step S101: Collect Wi-Fi handshake data during non-disaster periods, where the Wi-Fi handshake data is the four-way handshake request data between a terminal device and a Wi-Fi access point, including device ID, access point location, and connection time;

[0060] Step S102: Perform data cleaning and standardization on the collected Wi-Fi handshake data during non-disaster periods to obtain standardized data;

[0061] Step S103: Collect POI data and perform grid processing on the POI data to generate POI features of the grid;

[0062] Step S104: Match the standardized data with the POI features of the grid to obtain enhanced training data;

[0063] Specifically, POI data (Point of Interest), i.e., points of interest, usually includes various information points that people may be interested in, such as geographical locations, commercial facilities, public facilities, transportation nodes, etc. These data usually contain location coordinates (latitude and longitude), names, addresses, categories, and other relevant information. POI data is widely used in fields such as navigation systems, map services, urban planning, and commercial site selection to help users quickly find specific services or facilities. POI data of various types is aggregated on a grid scale, and finally a value for each grid is obtained, which is POI gridification. In one embodiment, the POI gridification process first generates a circumscribed rectangle for the collected POI data and divides it, then counts the number of POIs and attaches an attribute table back to the grid. Specifically, UUIDGenerator is used to generate a unique ID for each grid, which serves as the key for attaching the subsequent attribute table; Clipper is used to clip the POIs within the range of each grid, and Aggregator is used to aggregate the total number of POIs within each grid, and then FeatureMerger is used to attach the generated attribute table back to the grid generated in the previous step using the unique ID.

[0064] In this embodiment, the POI data is aggregated into POI grids, and a 100*100 POI grid is used to count the standardized data traffic within each grid. Specifically, the standardized data is matched to the POI features of the grid to obtain enhanced training data, so as to more clearly understand the data differences in each grid. By further aggregating the standardized data to the cluster scale, more refined and accurate evaluation can be achieved, and the difficulty of judging outliers is reduced.

[0065] Step S105: Divide the enhanced training data according to a preset time window to obtain the training dataset.

[0066] Specifically, when terminal devices such as mobile phones and pads connect to a Wi-Fi hotspot, four-way handshake data will be generated. The first handshake: The terminal sends a message to the base station, including a random number and an asymmetric encryption public key; the second handshake: The base station generates a random number and encrypts a message using the public key provided by the terminal, including the random number and the random number sent by the terminal before, and then sends it to the terminal; the third handshake: The terminal uses its own private key to decrypt the received message, verifies the random number, and generates a pairwise transient key (PTK); the fourth handshake: The terminal sends a message to the base station, including PTK and other necessary information, and the base station verifies PTK and completes the handshake process. During the four-way handshake process, the terminal device will exchange request data with the Wi-Fi hotspot, including device ID, connection frequency, access point location, access duration, etc., to establish a data link. Using the above request data, the number of people connected to the Wi-Fi hotspot can be obtained, thereby mastering the crowd gathering situation.

[0067] In this embodiment, the Wi-Fi handshake data in the non-disaster period is first cleaned and standardized, invalid or redundant handshake data is removed, and the data is divided according to the time window T. In one embodiment, the time window is divided every half an hour. Because when a disaster occurs, official shelters and conventional residences are both regular monitoring areas for disasters, with sufficient rescue manpower and timely measures, while unofficial shelters and unconventional residences often become monitoring blind spots, and it is difficult for people who take refuge there to receive timely and effective rescue. Therefore, this embodiment excludes the hot spots of residences and official shelters to obtain the gathering of people in unofficial or monitoring blind spots.

[0068] In one implementation, step S102 in this embodiment includes the following steps:

[0069] Step S1021: Among the Wi-Fi handshake data in the non-disaster period, according to the geographic information system database, identify and remove the Wi-Fi handshake data whose access point locations are regular residences and official shelters to obtain removed data;

[0070] Step S1022: normalize the eliminated data to obtain the standardized data.

[0071] Specifically, based on the Geographic Information System (GIS) database, Wi-Fi hotspots in regular residential areas and official shelters are identified and eliminated. The elimination rules are: filtered =D total -D residential -D official , where D filtered is the data after elimination, D total For all Wi-Fi handshake data, D residential and D official Represents the Wi-Fi hotspot data of the residence and the official shelter respectively. The data after elimination is standardized, and the data standardization method can be Min-max method to obtain standardized data.

[0072] Step S200, constructing an isolation forest model of informal shelters according to the training data set, wherein the informal shelters are shelters excluding official shelters and regular residences;

[0073] The Isolation Forest model is a fast anomaly detection method based on the integration of isolation trees (iTrees). The core idea of its anomaly detection is that anomaly points are outliers that are easily isolated. Therefore, the Isolation Forest randomly selects features and random thresholds for partitioning to generate multiple isolation trees until the isolation trees reach a certain height or until each leaf node contains only one point. Then, those outliers are easily partitioned early (i.e., the depth of the leaf node where they are located is relatively shallow). Since each isolation tree is independently generated by random sampling, there is a certain degree of independence between the isolation trees, and the integration of multiple isolation trees is the final Isolation Forest.

[0074] In this embodiment, an Isolation Forest model of informal shelters is constructed based on the training dataset. When a disaster occurs, by partitioning the outliers, it is possible to discover areas where people flow abnormally gather, thereby identifying potential unofficial shelters.

[0075] In one implementation, step S200 described in this embodiment includes the following steps:

[0076] Step S201: Within each time window, randomly select a feature vector and a split point from the training dataset, and construct an isolation tree based on the feature vector and the split point, where the feature vector includes device connection frequency, device connection duration, device density, and access point location, and the split point is used to isolate the data;

[0077] Specifically, a feature vector X = [x_1, x_2,..., x_n] is extracted within each time window. In this embodiment, n = 4, and the four feature vectors are: x_1 is the device connection frequency, x_2 is the device connection duration, x_3 is the device density, and x_4 is the access point location. The isolation tree isolates data points by randomly selecting features and split points.

[0078] Step S202: Repeat the step of randomly selecting a feature vector and a split point from the training dataset within each time window and constructing an isolation tree based on the feature vector and the split point until a termination condition is reached, obtaining a number of isolation trees;

[0079] In this embodiment, for each time window, an isolation tree is constructed. The process of constructing an isolation tree is recursive. At each node, a feature is randomly selected, and a split point is randomly selected between the maximum and minimum values of this feature. Then, the data is divided into the left subtree or the right subtree according to this split point. When the tree reaches the defined height, the number of samples in the node reaches a certain number, or the selected feature values of all samples are the same value, and after repeating the construction of a specific number of isolation trees, the collection is the Isolation Forest.

[0080] Step S203: construct an isolation forest model of informal shelters based on the isolated trees.

[0081] Specifically, by constructing an isolation forest model of informal shelters, intelligent anomaly detection is performed on Wi-Fi handshake data to identify the intensity of abnormal activities, so that the system can quickly and accurately detect hot spots where evacuated people gather, providing real-time decision support for emergency response.

[0082] Step S300, input the Wi-Fi handshake data of informal shelters during the disaster period into the isolation forest model of informal shelters to obtain an anomaly score for each data point;

[0083] In one implementation, step S300 in this embodiment includes the following steps:

[0084] Step S301, obtaining Wi-Fi handshake data during the disaster period, and removing Wi-Fi handshake data where the access point is located at a regular residence and an official shelter according to a geographic information system database, to obtain Wi-Fi handshake data of informal shelters during the disaster period;

[0085] Step S302: input the Wi-Fi handshake data of the informal shelters during the disaster period into the isolation forest model of informal shelters, and calculate the average path length and expected path length from the Wi-Fi handshake data of the informal shelters during the disaster period to each isolated tree;

[0086] Specifically, for new test samples, namely Wi-Fi handshake data of informal shelters during disasters, their path lengths in each isolated tree are calculated, and the average path length is calculated. The expected path length is the average path length of the tree and is used for normalization.

[0087] Step S303: Calculate the abnormality score of each data point according to the ratio of the expected path length to the average path length.

[0088] In this embodiment, the isolation forest calculates the anomaly score of each data point. Based on the calculated anomaly score, a threshold can be set to determine which data points are abnormal. j Calculate the anomaly score S(X j )for:

[0089] Among them, n is the data size, X j is the data point, i.e., the jth terminal device connected to the Wi-Fi hotspot, E(h(X j ) is X jThe average path length, c(n) is the expected path length for a given data scale n and is a constant. The path length is X j Through the feature selection method in the isolation tree construction stage, the number of edges required to reach the node where the sample is isolated (the leaf node of the finally reached tree) from the root node of the tree, the average path length E(h(X j ) is this X j The average value of the path lengths of all trees in the isolation forest. When the score S(X j ) is close to 1, it indicates that the data point is abnormal.

[0090] Step S400: Identify potential shelters based on the abnormal score and allocate rescue resources to the shelters.

[0091] Specifically, when E(h(X j ) is approximately equal to c(n), the path length of the sample point is close to the average path length, and abnormality cannot be judged. When E(h(X j ) is closer to 0, the closer S(X j ) is to 1, indicating that the sample is isolated and may be abnormal. Based on the abnormal score to identify potential shelters, the rescue command system can allocate resources in real time and give priority to responding to high-abnormality areas.

[0092] In one implementation, step S400 of this embodiment includes the following steps:

[0093] Step S401: Calculate the abnormality intensity of each Wi-Fi hotspot according to the abnormal score of each data point, where the abnormality intensity is the average value of the abnormal scores of the data points connected to each Wi-Fi hotspot;

[0094] Specifically, according to the abnormal score of each data point, calculate the abnormality intensity S avg (AP i ) as:

[0095]

[0096] where m is the number of terminal devices connected to the Wi-Fi hotspot within the time window, S(X j ) is the abnormal score of device X j , X j is the terminal device that connects to the i-th Wi-Fi hotspot and generates handshake data, and AP i is the i-th Wi-Fi hotspot.

[0097] Step S402: Compare the abnormality intensity with a preset abnormality detection threshold. If the abnormality intensity is greater than or equal to the abnormality detection threshold, mark the geographical location corresponding to the Wi-Fi hotspot as a potential shelter;

[0098] Specifically, preset an anomaly detection threshold as S according to historical data threshold , if S avg (AP i ) ≥ S threshold , then mark the geographical location corresponding to the Wi-Fi hotspot as a potential shelter site.

[0099] Step S403: Generate a geographical heat map according to the anomaly intensity of the potential shelter sites;

[0100] Step S404: Allocate rescue resources according to the geographical heat map.

[0101] Specifically, by calculating the anomaly intensity of each Wi-Fi access point and generating a geographical heat map, the system can provide real-time and dynamic information on the distribution of anomaly hotspots for the rescue command center, helping the rescue team to respond to high-risk areas first. In this embodiment, the rules for allocating rescue resources are set as follows from four perspectives: resource type identification, resource priority determination, path planning, and real-time path adjustment: According to the geographical heat map, when the anomaly score of a certain area exceeds the set threshold and lasts for a long time, such as more than 30 minutes, the system will make a demand prediction based on historical data, including the types, quantities, and allocation priorities of the required resources; for different demands, the system will identify the corresponding resource requirements, such as medical support, rescue teams, supplies (such as water and food), and transportation vehicles, etc.; allocate rescue resources to high-risk areas first to ensure that the resources arrive within the shortest time. Combining multiple anomaly areas in the heat map, the command system determines the order of resource allocation; the system automatically generates the best path from the nearest resource point to the anomaly area, considering factors such as road conditions, traffic flow, and disaster impacts, to ensure the rapid arrival of resources; during the execution process, the path planning will be dynamically adjusted according to the real-time traffic conditions and disaster situations to avoid resource entrapment or delay.

[0102] In one implementation, step S403 of this embodiment includes the following steps:

[0103] Step S4031: Create a three-dimensional scene according to the geographical information system database, associate the coordinates of the potential shelter sites with the coordinates in the three-dimensional scene to obtain the target positions;

[0104] Step S4032: Combine the target positions with the anomaly intensities of the corresponding potential shelter sites, map them into the three-dimensional scene, and generate the geographical heat map.

[0105] Specifically, a Geographic Heat Map is a heat map made based on a map base map. It is used to display the distribution rules of various data in space, such as population density, housing price distribution, traffic flow, etc. It can more accurately display the spatial distribution and trend of data, thereby helping people better understand and apply the data.

[0106] In this embodiment, the abnormal intensity heat map H(x, y) of the target position is calculated as follows:

[0107]

[0108] where (x, y) are the coordinates of the target position in the three-dimensional scene, and δ(x - x i , y - y i ) represents the correlation between the position coordinates of the Wi-Fi hotspot AP i and the coordinates of the target position. The geographic heat map generated through the abnormal intensity heat map of the target position can visually display potential shelter locations in the three-dimensional scene.

[0109] Exemplary device

[0110] As Figure 2 shown in

[0111] This embodiment also provides a shelter identification device based on an isolation forest model. The device includes:

[0112] A training dataset acquisition module 10 for collecting Wi-Fi handshake data during non-disaster periods and performing data preprocessing to obtain a training dataset;

[0113] A model training module 20 for constructing an isolation forest model of informal shelters based on the training dataset, where the informal shelters are shelters excluding official shelters and regular residences;

[0114] An abnormal score calculation module 30 for inputting the Wi-Fi handshake data of informal shelters during disaster periods into the isolation forest model of informal shelters to obtain the abnormal score of each data point;

[0115] A potential shelter identification module 40 for identifying potential shelters based on the abnormal scores and allocating rescue resources to the shelters.

[0116] In one implementation, the training dataset acquisition module 10 includes: A data collection unit for collecting Wi-Fi handshake data during non-disaster periods, where the Wi-Fi handshake data is the four-way handshake request data between a terminal device and a Wi-Fi access point, including device ID, access point location, and connection time;

[0117] A data processing unit for cleaning and standardizing the collected Wi-Fi handshake data during non-disaster periods to obtain standardized data;

[0118] A POI feature generation unit for collecting POI data and performing grid processing on the POI data to generate POI features of grids;

[0119] A feature matching unit for matching the standardized data with the POI features of the grids to obtain enhanced training data;

[0120] A training dataset acquisition unit for dividing the enhanced training data according to a preset time window to obtain the training dataset.

[0121] In one implementation, the data processing unit includes:

[0122] A data elimination subunit for identifying and eliminating Wi-Fi handshake data with access point locations being regular residences and official shelters in the Wi-Fi handshake data during non-disaster periods to obtain eliminated data;

[0123] A data standardization subunit for performing standardization processing on the eliminated data to obtain the standardized data.

[0124] In one implementation, the model training module 20 includes:

[0125] An isolation tree construction unit for randomly selecting feature vectors and split points from the training dataset within each time window and constructing isolation trees according to the feature vectors and split points, where the feature vectors include device connection frequency, device connection duration, device density, and access point location, and the split points are used to isolate data;

[0126] An iteration unit for repeatedly executing the step of randomly selecting feature vectors and split points from the training dataset within each time window and constructing isolation trees according to the feature vectors and split points until a termination condition is reached, to obtain a number of isolation trees;

[0127] A model establishment unit for constructing an isolation forest model of informal shelters according to the isolation trees.

[0128] In one implementation, the abnormal score calculation module 30 includes:

[0129] A data cleaning unit, configured to obtain Wi-Fi handshake data during a disaster period, and eliminate the Wi-Fi handshake data with the access point locations being regular residences and official shelters according to the geographic information system database, so as to obtain the Wi-Fi handshake data of informal shelters during the disaster period;

[0130] A length calculation unit, configured to input the Wi-Fi handshake data of informal shelters during the disaster period into the isolation forest model of informal shelters, and calculate the average path length and expected path length from the Wi-Fi handshake data of informal shelters during the disaster period to each isolation tree;

[0131] An anomaly score calculation unit, configured to calculate the anomaly score of each data point according to the ratio of the expected path length to the average path length.

[0132] In one implementation, the potential shelter identification module 40 includes:

[0133] An anomaly intensity calculation unit, configured to calculate the anomaly intensity of each Wi-Fi hotspot according to the anomaly score of each data point, where the anomaly intensity is the average value of the anomaly scores of the data points connected to each Wi-Fi hotspot;

[0134] A comparison unit, configured to compare the anomaly intensity with a preset anomaly detection threshold. If the anomaly intensity is greater than or equal to the anomaly detection threshold, mark the geographical location corresponding to the Wi-Fi hotspot as a potential shelter;

[0135] A geographic heat map generation unit, configured to generate a geographic heat map according to the anomaly intensity of the potential shelter;

[0136] An allocation unit, configured to allocate rescue resources according to the geographic heat map.

[0137] In one implementation, the geographic heat map generation unit includes:

[0138] A target location acquisition subunit, configured to create a three-dimensional scene according to the geographic information system database, associate the coordinates of the potential shelter with the coordinates in the three-dimensional scene, and obtain the target location;

[0139] A location mapping subunit, configured to combine the target location with the corresponding anomaly intensity of the potential shelter, map it into the three-dimensional scene, and generate the geographic heat map.

[0140] Based on the above embodiments, the present invention further provides an intelligent terminal, and its principle block diagram can be as Figure 3As shown in the figure. The intelligent terminal includes a processor, a memory, a network interface, a display screen, and a temperature sensor connected through a system bus. Among them, the processor of the intelligent terminal is used to provide computing and control capabilities. The memory of the intelligent terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the intelligent terminal is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it realizes a method for identifying a shelter based on an isolation forest model. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen. The temperature sensor of the intelligent terminal is pre-set inside the intelligent terminal and is used to detect the operating temperature of internal devices.

[0141] Those skilled in the art can understand that Figure 3 the block diagram of the principle shown in the figure is only the block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the intelligent terminal to which the solution of the present invention is applied. The specific intelligent terminal may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0142] In one embodiment, an intelligent terminal is provided. The intelligent terminal includes a memory, a processor, and a shelter identification program based on an isolation forest model stored in the memory and operable on the processor. When the processor executes the shelter identification program based on the isolation forest model, the following operation instructions are realized:

[0143] Collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training data set;

[0144] Construct an isolation forest model of informal shelters according to the training data set, where the informal shelters are shelters excluding official shelters and regular residences;

[0145] Input the Wi-Fi handshake data of informal shelters during disaster periods into the isolation forest model of informal shelters to obtain the anomaly score of each data point;

[0146] Identify potential shelters according to the anomaly score and allocate rescue resources to the shelters.

[0147] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, operational database, or other medium used in the embodiments provided by the present invention can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0148] In summary, the present invention discloses a method for identifying a shelter based on an isolation forest model. The method includes: collecting Wi-Fi handshake data during non-disaster periods and performing data preprocessing to obtain a training data set; constructing an isolation forest model for informal shelters based on the training data set; inputting the Wi-Fi handshake data of informal shelters during disaster periods into the isolation forest model of informal shelters to obtain the anomaly score of each data point; identifying potential shelters according to the anomaly score, and allocating rescue resources for the shelters. The present invention can automatically identify the crowd gathering situation in informal shelters during disasters, greatly improving the monitoring ability of crowded areas of shelter-seeking people. By using Wi-Fi handshake data to monitor device connections and crowd activities in real time during disasters, it does not rely on additional physical infrastructure, saving the cost of equipment installation, and has broad application prospects.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying shelters based on an isolation forest model, characterized in that: The method comprises: Collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training data set; Constructing an isolation forest model of informal shelters according to the training data set, wherein the informal shelters are shelters excluding official shelters and regular residences; The Wi-Fi handshake data of informal shelters during the disaster period was input into the Isolation Forest Model of informal shelters to obtain the anomaly score of each data point; Potential refuge sites are identified based on the anomaly scores, and rescue resources are allocated to the refuge sites.

2. The method for identifying shelters based on the isolation forest model according to claim 1, characterized in that: The Wi-Fi handshake data during the non-disaster period is collected and preprocessed to obtain a training data set, including: Collect Wi-Fi handshake data during non-disaster periods, where the Wi-Fi handshake data is four-way handshake request data between a terminal device and a Wi-Fi access point, including device ID, access point location, and connection time; The Wi-Fi handshake data collected during the non-disaster period was cleaned and standardized to obtain standardized data; Collecting POI data, and performing grid processing on the POI data to generate POI features of the grid; Matching the standardized data with the POI features of the grid to obtain enhanced training data; The enhanced training data is divided according to a preset time window to obtain the training data set.

3. The method for identifying shelters based on the isolation forest model according to claim 2, characterized in that: The data cleaning and standardization of the collected Wi-Fi handshake data during the non-disaster period to obtain standardized data includes: Among the Wi-Fi handshake data in the non-disaster period, the Wi-Fi handshake data with access points located at regular residences and official shelters are identified and removed according to a geographic information system database to obtain removed data; The eliminated data is standardized to obtain the standardized data.

4. The method for identifying shelters based on the isolation forest model according to claim 1, characterized in that: The step of constructing an isolation forest model of informal shelters according to the training data set includes: In each time window, randomly selecting feature vectors and split points from the training data set, and constructing an isolation tree based on the feature vectors and split points, wherein the feature vectors include device connection frequency, device connection duration, device density, and access point location, and the split points are used to isolate data; Repeat the steps of randomly selecting feature vectors and split points from the training data set in each time window, and constructing an isolated tree according to the feature vectors and split points, until a termination condition is reached, to obtain a plurality of isolated trees; Based on the isolated trees, an isolation forest model of informal shelters was constructed.

5. The method for identifying shelters based on the isolation forest model according to claim 1, characterized in that: The Wi-Fi handshake data of informal shelters during the disaster period is input into the isolation forest model of informal shelters to obtain the anomaly score of each data point, including: Obtain Wi-Fi handshake data during the disaster period. According to the geographic information system database, remove the Wi-Fi handshake data where the access point is located at a regular residence and official shelter, and obtain the Wi-Fi handshake data of informal shelters during the disaster period. Inputting the Wi-Fi handshake data of the informal shelters during the disaster period into the isolation forest model of informal shelters, and calculating the average path length and expected path length from the Wi-Fi handshake data of the informal shelters during the disaster period to each isolated tree; The anomaly score of each data point is calculated according to the ratio of the expected path length to the average path length.

6. The method for identifying shelters based on the isolation forest model according to claim 5, characterized in that: The step of identifying a potential refuge site according to the abnormal score and allocating rescue resources to the refuge site includes: Calculate the anomaly strength of each Wi-Fi hotspot according to the anomaly score of each data point, wherein the anomaly strength is the average of the anomaly scores of the data points connected to each Wi-Fi hotspot; Comparing the abnormality intensity with a preset abnormality detection threshold, if the abnormality intensity is greater than or equal to the abnormality detection threshold, marking the geographical location corresponding to the Wi-Fi hotspot as a potential refuge place; generating a geographic heat map based on the anomaly intensity of the potential refuge sites; Rescue resources are deployed according to the geographic heat map.

7. The method for identifying shelters based on the isolation forest model according to claim 6, characterized in that: The generating of a geographic heat map according to the abnormal intensity of the potential refuge places comprises: Creating a three-dimensional scene based on a geographic information system database, associating the coordinates of the potential refuge site with the coordinates in the three-dimensional scene to obtain a target location; The target location is combined with the abnormal intensity of the corresponding potential refuge site and mapped into the three-dimensional scene to generate the geographic heat map.

8. A shelter identification device based on an isolation forest model, characterized in that: The device comprises: The training data set acquisition module is used to collect Wi-Fi handshake data during non-disaster periods and perform data preprocessing to obtain a training data set; A model training module, used to construct an isolation forest model of informal shelters according to the training data set, wherein the informal shelters are shelters excluding official shelters and regular residences; The anomaly score calculation module is used to input the Wi-Fi handshake data of informal shelters during the disaster period into the isolation forest model of informal shelters to obtain the anomaly score of each data point; The potential refuge site identification module is used to identify potential refuge sites according to the abnormality scores and allocate rescue resources to the refuge sites.

9. An intelligent terminal, characterized in that: The intelligent terminal includes a memory, a processor, and a shelter identification program based on an isolation forest model stored in the memory and executable on the processor. When the processor executes the shelter identification program based on the isolation forest model, the steps of the shelter identification method based on the isolation forest model as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a shelter identification program based on the isolation forest model. When the shelter identification program based on the isolation forest model is executed by the processor, the steps of the shelter identification method based on the isolation forest model as described in any one of claims 1-7 are implemented.

Images