Intelligent rescue method for lost old people based on PTS algorithm and RF algorithm

Abstract

The invention discloses an intelligent rescue method for lost old people based on a PTS algorithm and an RF algorithm. The method comprises the steps that S1, family members of the lost old people rapidly record cases; s2, the command platform carries out task scheduling, carries out priority calculation on missing information of the old people according to a PTS algorithm, and sends rescue tasks to volunteers according to priorities; s3, the volunteer checks the current location, the position information of other volunteers and the thermodynamic diagram of people searching on the map according to the rescue system, and plans a people searching route, and S4, if a suspected lost person is found, the suspected lost person is photographed and uploaded, the suspected lost person is automatically compared with the lost person in the photo library, the information of the elder with the matching degree higher than 80% is synchronized to the family member end and the volunteer end, and the elder can search for the volunteer. After confirmation, the search and rescue task can be judged to be completed; s5, if abnormity occurs during search and rescue, the background can check task positioning and rescue details at any time, and rescue resources are manually adjusted and allocated; and S6, after the rescue task is finished, the system integrates, counts and stores related information, and subsequently assists prediction of the next search and rescue task through an RF machine learning algorithm.

Description

Technical Field

[0001] The present invention relates to the field of lost elderly rescue, and more particularly to an intelligent lost elderly rescue method based on a PTS algorithm and a RF algorithm. Background Art

[0002] In the case of an elderly person being lost and the rescue time being tight, it is difficult to guarantee the efficiency of finding the person by relying solely on the public security force, and it is easy to miss the best time for rescue after the person is lost. Therefore, independent and caring people in the private sector spontaneously organized a purely public welfare emergency rescue organization. However, at this stage, the rescue team has problems in dispatching and management, such as difficulty in task notification, difficulty in identifying the elderly, and uneven quality of volunteers. In order to realize the intelligentization and quality standardization of the rescue team, the present invention combines the existing data and invents this intelligent rescue solution to help the rescue team to achieve real-time update and sharing of rescue mission information, so as to grasp the best time for rescue. Summary of the invention

[0003] The present invention aims to overcome the above-mentioned shortcomings of the prior art and provide an intelligent rescue method for lost elderly people based on PTS algorithm and RF algorithm.

[0004] The present invention aims at the problem of lost elderly people, seizes the pain points of civilian rescue teams, and based on the PTS algorithm and the RF algorithm, combines other existing technologies to further improve and upgrade the current rescue plan, and provides a set of intelligent rescue plans suitable for the current situation of civilian rescue teams to solve the problems raised in the above background technology.

[0005] Different from other solutions for lost elderly people, the present invention is designed specifically for the problems of civilian rescue teams and creates a mature and reliable new rescue solution. Its features are face and body recognition, dynamic image intelligent collection of pictures, real-time comparison, and improved accuracy and effectiveness. Data retrieval and integration, call relevant resources at any time, and improve data sharing. Data visualization analysis and monitoring, realize mobile phone control of large screen, by combining data analysis technology with graphics technology, as a form of expression to clearly and effectively interpret and convey the analysis result information. Among them, based on the idea of ​​priority task scheduling algorithm in the operating system, the present invention designs and proposes the PTS (Priority Task Scheduling) algorithm. In fact, the order of task allocation is of key significance to the development of rescue operations. The more urgent the task, the more it needs to be solved as soon as possible. Whenever a new task enters the queue, the priority of each task in the task queue will be recalculated, and then it will be sorted from high to low based on the priority standard, and assigned to volunteer tasks according to the priority. According to the actual situation and data investigation, the present invention sets up reasonable thresholds and priority scheduling parameters to assist the allocation of rescue tasks in this scheme. Based on the RF (Random Forests) algorithm, combined with historical rescue data, reasonable suggestions are provided for ongoing rescue operations. Historical rescue data includes weather (temperature, precipitation, etc.), location, and the location where the elderly were finally found. The suggestions provided include recommending the number of volunteers needed for the rescue operation, predicting the success probability and completion time of the rescue operation, and providing action suggestions (for example, when the weather is too hot, the local lost elderly usually go indoors).

[0006] The present invention combines the characteristics of the above technologies to provide a set of intelligent rescue solutions for lost elderly people. The use of the present invention can effectively improve the efficiency of teamwork and the quality of rescue services, and promote the intelligentization and quality standardization of rescue teams.

[0007] The present invention provides an intelligent rescue method for lost elderly people based on PTS algorithm and RF algorithm, comprising:

[0008] S1: The family members of the missing elderly can quickly file a case. For the first report, they only need to upload some important information, and then upload the complete information of the missing elderly later;

[0009] S2: The command platform performs task scheduling and calculates the priority of the collected missing elderly information according to the PTS algorithm. Subsequently, a rescue task notification is sent to the number of volunteers within 5 kilometers of the missing place according to the priority of the task;

[0010] S3: The search and rescue volunteers check their current location and the location information of other volunteers who are carrying out the rescue mission together on the map according to the rescue system, as well as the heat map of the search, and plan the next search route.

[0011] S4: If a suspected missing person is found, a photo will be uploaded and automatically compared with all the missing persons in the photo library. The information of the elderly with a matching degree higher than 80% will be synchronized to the family and volunteer terminals. After confirmation by the family and then confirmed by the command platform, the search and rescue mission can be considered completed;

[0012] S5: If an abnormality occurs during the search and rescue, the backend command platform can check the mission location and rescue details at any time, assist family members to modify the elderly person's information, and manually adjust the allocation of rescue resources according to the actual situation;

[0013] S6: After the rescue mission is completed, the system integrates and statistically stores relevant information, and then uses the RF machine learning algorithm to assist in the prediction of the approximate lost location for the next search and rescue mission.

[0014] Wherein, the PTS algorithm described in step S2 includes:

[0015] S21: Based on the improvement of the priority scheduling algorithm in the operating system, a weight relationship matrix is ​​first established according to the factors of missing time, age of the elderly, health status of the elderly, and climate conditions;

[0016] S22: Perform a weighted sum operation on the intra-group weight and inter-group weight of the task to obtain the priority of the task, establish a priority queue, and then push it to the corresponding volunteers for rescue.

[0017] Wherein, the setting of the weight relationship matrix described in step S21 is as follows:

[0018] The establishment of the weight relationship matrix takes into account the following factors: missing time, age of the elderly, health status of the elderly, climate conditions, and population density. The above factors are set as follows:

[0019] - If the missing time is between 0 and 3 hours, the weight within the group is 0.3 and the weight between groups is 0.3; if the missing time is between 3 and 24 hours, the weight within the group is 0.5 and the weight between groups is 0.3; if the missing time exceeds 24 hours, the weight within the group is 0.2 and the weight between groups is 0.3.

[0020] The weight is 0.3;

[0021] - Elderly people under 70 years old, with an intra-group weight of 0.3 and an inter-group weight of 0.2;

[0022] For the elderly aged between 70 and 80, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly aged over 80, the intra-group weight is 0.4 and the inter-group weight is 0.2;

[0023] - Elderly people with good health have a within-group weight of 0.2 and an between-group weight of 0.2;

[0024] For the elderly with average health status, the within-group weight is 0.3 and the between-group weight is 0.2;

[0025] For those with poor health, the within-group weight is 0.5 and the between-group weight is 0.2;

[0026] -For good weather conditions, the weight within the group is 0.2 and the weight between groups is 0.1; for average weather conditions, the weight within the group is 0.3 and the weight between groups is 0.1; for bad weather conditions, the weight within the group is 0.5 and the weight between groups is 0.1;

[0027] - For population density that is high, the weight within the group is 0.4 and the weight between groups is 0.2; for population density that is medium, the weight within the group is 0.3 and the weight between groups is 0.2; for population density that is low, the weight within the group is 0.3 and the weight between groups is 0.2;

[0028] The final priority calculation formula is:

[0029]

[0030] in:

[0031] P(i) is the final priority of the i-th task;

[0032] W i is the intra-group weight corresponding to the i-th task, which is calculated by multiplying the weight of the corresponding condition by the sum of the intra-group weights;

[0033] R ij It is the element in the i-th row and j-th column in the inter-group weight matrix, indicating the weight relationship between the i-th task and the j-th task.

[0034] Among them, some of the important information described in step S1 includes a recent photo of the elderly and the approximate location where they were lost.

[0035] Among them, the specific process of S6 includes:

[0036] S61: After each rescue mission, the existing rescue information will be combined with weather-related information, the data will be desensitized, integrated and stored in the database;

[0037] S62: Use this data to train a random forest model to provide auxiliary suggestions for the next rescue operation. According to the weather, age and other information of the new mission, the most likely place for the elderly to go is predicted. Volunteers then plan routes to search for the elderly based on their own experience and auxiliary suggestions.

[0038] Among them, the places where the elderly described in step S62 are most likely to go include: the place where the elderly are lost in hot weather is indoors, the elderly in their seventies are in a museum, and the elderly in their octogenarians appear in an elderly activity center.

[0039] The method of the present invention is integrated into the system and cannot be simply separated from the system. In other words, the algorithm proposed in the present invention is only a complete intelligent rescue solution that can solve the problems of the current civilian rescue team when it is matched with the corresponding intelligent rescue system. Therefore, when introducing the specific algorithm in the intelligent rescue solution, it is inseparable from first understanding the rescue system provided by this embodiment. The system includes: 1. Family members who report the case of missing elderly people through the applet and view the rescue details; 2. Volunteer end that includes task management and rescue activity information sharing through the app, and view recommended search locations; 3. Including the overall management of information such as the elderly, volunteers, and tasks through the web platform; 4. Staff who view the location of volunteers and the approximate coordinates of lost elderly people through the GIS geographic information system; 5. It also includes the WeChat public account maintenance end that issues new notifications and other messages from the rescue team 6. And a data visualization platform for mobile phone control of large screen and data visualization monitoring.

[0040] The family end of the lost elderly includes a quick case filing module, a clue supplement module, a rescue progress check module, an automatic task distribution module, and an automatic alarm prompt module. The quick case filing module simplifies the reporting process and improves the reporting efficiency. For the first report, it only needs to upload a recent photo of the lost elderly, select the approximate place of loss, fill in the name of the elderly, the name of the family and the contact information, and then complete the elderly information such as age, gender, ID number, clothing characteristics, facial features, etc. after the report is successful. Different from the cumbersome registration process of the traditional rescue scheme, this scheme takes into account the actual situation of the private rescue organization, simplifies the traditional cumbersome registration process, and prevents the rescue opportunity from being delayed due to registration. The core automatic task distribution module is based on the PTS algorithm proposed by the present invention. The priority of the task is calculated based on the time of disappearance of the elderly, the age of the elderly, the health status of the elderly, the climate conditions, and the population density collected by the task. The more urgent the task is, the more it needs to be solved as soon as possible. The command system automatically sends the OpenID of the volunteer within 5km of the lost place to the WeChat applet end according to the priority of the task, and the WeChat applet end sends the rescue task notification to the volunteer of the OpenID. If there are no qualified volunteers within the 5km range, the range will be expanded to 10km; if there are no volunteers within the 10km range, the WeChat mini program will notify the family to report again later. If the rescue request is reported for a certain period of time and the search is unsuccessful, the family mini program will automatically jump to the alarm dialing interface to prompt the police and assist the family to file a case at the police station.

[0041] The volunteer side includes a process record module for volunteer rescue, which fills in the current location, rescue route, and clues found, and realizes information sharing and communication among family members, command system, and volunteers, so that family members can understand the progress of the rescue and the command system can give important task instructions in time. Including a face recognition module (this solution uses the PyramidBox model technology provided by Baidu, which is not an innovation of this rescue solution). When a suspected missing person is found during the rescue process, a photo is uploaded to the system. The system automatically compares it with all the missing persons that have not been found in the photo library, and returns the information of the elderly and family members with a matching degree of more than 80% to the volunteer. The mini program automatically sends a confirmation message to the family. In addition, based on the GIS geographic information system and the data service toolkit LBS provided by the mini program side, the team members can view their current location on the map. At the same time, the location information of other volunteers who are performing the rescue mission together and the heat map of the search for people are introduced to facilitate the volunteer to plan the next search route, ensure that there will be no omission of important areas due to inadequate communication and multiple searches of the same area by different team members, and avoid delays in search and rescue or waste of manpower.

[0042] The backend command and management platform can jump from the task details to view the location of the task on the map and the personnel details within a radius of several thousand meters. It is convenient for the administrator's backend staff to assist family members in modifying the elderly's information and manually adjust the task details according to the actual situation. Combined with the automatic task dispatching mentioned above, the management platform is equivalent to giving staff the flexibility to manually participate in the task, which improves the rescue efficiency.

[0043] The data monitoring platform includes modules such as data visualization, mobile phone control large screen, and lost event summary. In order to help the rescue team and other ordinary personnel quickly and more intuitively understand the progress of the rescue team, the lost events in this period are graphically displayed, such as the age distribution of the lost person, the lost area, the number of rescue personnel involved, the distribution of volunteers across the country, the case resolution status, and other information. The core rescue suggestion module is based on the RF algorithm proposed by the present invention. Compared with other rescue schemes, this scheme provides reasonable suggestions for ongoing rescue operations based on these historical rescue data recorded statistically (for example, lost elderly people are usually prone to go indoors in hot weather). After receiving the suggestion information, the rescue personnel will carry out rescue according to the suggestion information, which can greatly improve the success rate and efficiency of the rescue. When a rescue mission is completed, the system will integrate and count the relevant information in the database, and use the weather interface to obtain necessary information, and then ask volunteers to confirm / supplement the relevant information of the mission to ensure the quality of the data.

[0044] From the perspective of concurrency and disaster tolerance, the intelligent rescue solution uses master-slave replication and sentinel mechanisms to build a cluster. The master server performs reading and writing, and the slave server performs replication query functions to prevent server crashes caused by high traffic. The sentinel mechanism uses a heartbeat polling algorithm to determine whether the master server is faulty, and notifies other servers to become the master server when a fault occurs. Not only that, the present invention also designs a dedicated 404 page that can be directly connected to other projects, enhance social influence, and call on more people to engage in this public welfare cause. Some interfaces are exposed to the outside world to facilitate other organizations to use data.

[0045] The advantages of the present invention are: the proposed scheme combines two algorithms and a rescue system, conducts analysis from multiple dimensions, utilizes data information in an all-round way, and performs task scheduling and allocation based on the PTS algorithm to carry out rescue; collects and utilizes data after the mission is completed, and provides guidance for the next rescue based on the RF algorithm; truly realizes end-to-end communication between family members, rescue teams, and command personnel, greatly improves the efficiency of rescuing lost elderly people, and solves the shortcomings of difficult task allocation and scheduling and low efficiency in rescue organized by volunteers of civilian rescue teams. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a specific schematic diagram of the process of the present invention;

[0047] Figure 2 is a schematic diagram of the intended users of the present invention;

[0048] Figure 3 It is a schematic diagram of the face recognition process of the present invention;

[0049] Figure 4 It is a schematic diagram of the architecture of the background command system of the present invention;

[0050] Figure 5 It is a detailed schematic diagram of the specific tasks GIS of the present invention;

[0051] Figure 6 It is a visualization diagram of the data platform of the present invention;

[0052] Figure 7 is a schematic diagram of the RF algorithm of the present invention;

[0053] Figure 8 It is a schematic diagram of the end-to-end relationship of users of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] Example 1

[0056] like Figure 1 — Figure 7 This embodiment relates to an intelligent rescue method for lost elderly people based on PTS algorithm and RF algorithm, including:

[0057] S1: The family members of the missing elderly can quickly file a case. For the first report, they only need to upload some important information, and then upload the complete information of the missing elderly later;

[0058] S2: The command platform performs task scheduling and calculates the priority of the collected missing elderly information according to the PTS algorithm. Subsequently, a rescue task notification is sent to the number of volunteers within 5 kilometers of the missing place according to the priority of the task;

[0059] S3: The search and rescue volunteers check their current location and the location information of other volunteers who are carrying out the rescue mission together on the map according to the rescue system, as well as the heat map of the search, and plan the next search route.

[0060] S4: If a suspected missing person is found, a photo will be uploaded and automatically compared with all the missing persons in the photo library. The information of the elderly with a matching degree higher than 80% will be synchronized to the family and volunteer terminals. After confirmation by the family and then confirmed by the command platform, the search and rescue mission can be considered completed;

[0061] S5: If an abnormality occurs during the search and rescue, the backend command platform can check the mission location and rescue details at any time, assist family members to modify the elderly person's information, and manually adjust the allocation of rescue resources according to the actual situation;

[0062] S6: After the rescue mission is completed, the system integrates and statistically stores relevant information, and then uses the RF machine learning algorithm to assist in the prediction of the approximate lost location for the next search and rescue mission.

[0063] Wherein, the PTS algorithm described in step S2 includes:

[0064] S21: Based on the improvement of the priority scheduling algorithm in the operating system, a weight relationship matrix is ​​first established according to the factors of missing time, age of the elderly, health status of the elderly, and climate conditions;

[0065] S22: Perform a weighted sum operation on the intra-group weight and inter-group weight of the task to obtain the priority of the task, establish a priority queue, and then push it to the corresponding volunteers for rescue.

[0066] Wherein, the setting of the weight relationship matrix described in step S21 is as follows:

[0067] The establishment of the weight relationship matrix takes into account the following factors: missing time, age of the elderly, health status of the elderly, climate conditions, and population density. The above factors are set as follows:

[0068] - If the missing time is between 0 and 3 hours, the weight within the group is 0.3 and the weight between groups is 0.3; if the missing time is between 3 and 24 hours, the weight within the group is 0.5 and the weight between groups is 0.3; if the missing time exceeds 24 hours, the weight within the group is 0.2 and the weight between groups is 0.3.

[0069] The weight is 0.3;

[0070] - Elderly people under 70 years old, with an intra-group weight of 0.3 and an inter-group weight of 0.2;

[0071] For the elderly aged between 70 and 80, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly aged over 80, the intra-group weight is 0.4 and the inter-group weight is 0.2;

[0072] - Elderly people with good health have a within-group weight of 0.2 and an between-group weight of 0.2;

[0073] For the elderly with average health status, the within-group weight is 0.3 and the between-group weight is 0.2;

[0074] For those with poor health, the within-group weight is 0.5 and the between-group weight is 0.2;

[0075] -For good weather conditions, the weight within the group is 0.2 and the weight between groups is 0.1; for average weather conditions, the weight within the group is 0.3 and the weight between groups is 0.1; for bad weather conditions, the weight within the group is 0.5 and the weight between groups is 0.1;

[0076] - For population density with high density, the weight within the group is 0.4 and the weight between groups is 0.2; for population density with medium density, the weight within the group is 0.3 and the weight between groups is 0.2; for population density with low density, the weight within the group is 0.3 and the weight between groups is 0.2

[0077] The final priority calculation formula is:

[0078]

[0079] in:

[0080] P(i) is the final priority of the i-th task;

[0081] W i is the intra-group weight corresponding to the i-th task, which is calculated by multiplying the weight of the corresponding condition by the sum of the intra-group weights;

[0082] R ij It is the element in the i-th row and j-th column in the inter-group weight matrix, indicating the weight relationship between the i-th task and the j-th task.

[0083] Among them, some of the important information described in step S1 includes a recent photo of the elderly and the approximate location where they were lost.

[0084] Among them, the specific process of S6 includes:

[0085] S61: After each rescue mission, the existing rescue information will be combined with weather-related information, the data will be desensitized, integrated and stored in the database;

[0086] S62: Use this data to train a random forest model to provide auxiliary suggestions for the next rescue operation. According to the weather, age and other information of the new mission, the most likely place for the elderly to go is predicted. Volunteers then plan routes to search for the elderly based on their own experience and auxiliary suggestions.

[0087] Among them, the places where the elderly described in step S62 are most likely to go include: the place where the elderly are lost in hot weather is indoors, the elderly in their seventies are in a museum, and the elderly in their octogenarians appear in an elderly activity center.

[0088] Example 2

[0089] Combination Figure 1 and Figure 2, this embodiment relates to an intelligent rescue system for lost elderly people based on PTS algorithm and RF algorithm, which is used to implement an intelligent rescue method for lost elderly people based on PTS algorithm and RF algorithm described in Example 1. Different from other solutions for lost elderly people, the system of this embodiment is specially built for civilian rescue teams, with simple process, efficient task allocation, and rapid information synchronization. Family members report cases through the mini program end, and the background command system performs information management and task judgment, and automatically dispatches tasks for personnel scheduling. Based on the idea of ​​priority task scheduling algorithm in the operating system, the present invention designs and proposes PTS (Priority Task Scheduling) algorithm. In fact, the order of task allocation is of key significance to the development of rescue operations. The more urgent the task, the more it needs to be solved as soon as possible. The present invention obtains the priority of the task through the PTS algorithm, and then sorts it from high to low based on the priority standard, and dispatches it to volunteers to perform the task. During the volunteer search, based on the RF (random forest) algorithm and combined with the historical rescue data of the "Homecoming Operation" system, reasonable suggestions are provided for the ongoing rescue operations. The historical rescue data includes weather (temperature, precipitation, etc.), location, and the place where the elderly were finally found. The suggestions provided include recommending the number of volunteers needed for the rescue operation, predicting the success probability and completion time of the rescue operation, and providing action suggestions (such as: the weather is too hot, and local lost elderly people usually go indoors), etc. The task level will still change according to changes in time and background commands, and further task personnel will be dispatched.

[0090] In summary, this technical solution can be used for different roles (see Figure 2 ) is divided into family members, volunteers, command system, data platform and third parties for detailed description:

[0091] 1. Family Members

[0092] The family side of the lost elderly includes a quick case filing module, a clue supplement module, a rescue progress check module, an automatic task assignment module, and an automatic alarm prompt module. The quick case filing module simplifies the reporting process and improves the reporting efficiency. For the first report, you only need to upload a recent photo of the lost elderly, select the approximate location of the lost elderly, fill in the elderly’s name, family name and contact information, and complete the elderly’s information such as age, gender, ID number, clothing characteristics, facial features, etc. after the report is successful.

[0093] This plan simplifies the cumbersome registration process in traditional rescue plans to prevent delays in rescue due to registration. Family members can file a case simply by filling in some important fields through the WeChat mini-program authorization. In order to better find the elderly and continue to provide additional clues, this plan also takes into account the special needs of family members. If an older person reports the case, they can also contact the back-end management staff to report the case and improve the efficiency of the rescue.

[0094] After the case was reported, the PTS algorithm (Priority Task Scheduling) was proposed based on the idea of ​​task scheduling algorithm in the operating system. The priority in this algorithm is mainly based on the following factors: time of disappearance, age of the elderly, health status of the elderly, climate conditions, and population density. The weight matrix of the above factors is set as follows based on relevant data:

[0095] - If the missing time is between 0 and 3 hours, the intra-group weight is 0.3 and the inter-group weight is 0.3; if the missing time is between 3 and 24 hours, the intra-group weight is 0.5 and the inter-group weight is 0.3; if the missing time exceeds 24 hours, the intra-group weight is 0.2 and the inter-group weight is 0.3;

[0096] - For the elderly under 70 years old, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly between 70 and 80 years old, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly over 80 years old, the intra-group weight is 0.4 and the inter-group weight is 0.2;

[0097] - For the elderly with good health, the intra-group weight is 0.2 and the inter-group weight is 0.2; for the elderly with average health, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly with poor health, the intra-group weight is 0.5 and the inter-group weight is 0.2;

[0098] -For good weather conditions, the weight within the group is 0.2 and the weight between groups is 0.1; for average weather conditions, the weight within the group is 0.3 and the weight between groups is 0.1; for bad weather conditions, the weight within the group is 0.5 and the weight between groups is 0.1;

[0099] - For population density that is high, the weight within the group is 0.4 and the weight between groups is 0.2; for population density that is medium, the weight within the group is 0.3 and the weight between groups is 0.2; for population density that is low, the weight within the group is 0.3 and the weight between groups is 0.2;

[0100] The final priority calculation formula is:

[0101]

[0102] in:

[0103] P(i) is the final priority of the i-th task.

[0104] W i is the intra-group weight corresponding to the i-th task, which is calculated by multiplying the weight of the corresponding condition by the sum of the intra-group weights.

[0105] R ijis the element in the i-th row and j-th column of the inter-group weight matrix, indicating the weight relationship between the i-th task and the j-th task. This formula combines the intra-group weight and the inter-group weight of each task to give the final task priority. In fact, the more urgent the task is, the sooner it needs to be solved. According to the priority queue, the command system will automatically send the OpenID of the volunteer within 5km of the lost place to the WeChat applet according to the priority of the task, and the WeChat applet will send a rescue task notification to the volunteer with the OpenID. If there are no qualified volunteers within 5km, the range will be expanded to 10km; if there are no volunteers within 10km, the WeChat applet will notify the family to report again later. If the rescue request is reported for a certain period of time and the search is unsuccessful, the family applet will automatically jump to the alarm dialing interface to prompt the alarm and assist the family to file a case at the police station.

[0106] At the same time, taking into account some emergencies in the system, a manual mechanism was introduced on the basis of automatically calculating the priority of task assignment. Staff can manually supervise the priority and rationality of tasks in the command system to ensure the normal notification and progress of tasks, and avoid data deviations caused by abnormalities that affect rescue operations.

[0107] 2. Volunteers

[0108] Team members set notification methods in mobile applications or mini-programs. When the command platform releases a rescue mission, it will issue a rescue mission notification based on the method selected by the team members (such as SMS notification, WeChat mini-program notification, and phone notification). Team members will respond based on their own circumstances as to whether to participate in this rescue mission. During the rescue process, when volunteers find clues related to missing persons, they can submit them to the process recording module, fill in the current location, rescue route, clues found, etc., to achieve end-to-end sharing and communication of information among family members, the command platform, and volunteers, so that family members can follow up on the progress of the rescue and the command platform can give important task instructions in a timely manner.

[0109] When a suspected missing person is found during the rescue process, a photo is taken and uploaded to the system. The system automatically compares the photo with all the missing persons that have not been found in the photo library, and returns the information of the elderly and their families with a matching degree of more than 80% to the volunteers (this system uses the PyramidBox model technology provided by Baidu). The mini program automatically sends a confirmation message to the family, such as Figure 3This is a flowchart of face recognition for this solution. When the family members confirm that the elderly person is the missing person they are looking for, the volunteer confirms that the elderly person has been found on the mobile app or mini program, and the rescue mission is over. In addition, based on the GIS geographic information system and the LBS data service toolkit provided by the mini program, the team members can view their current location on the map. At the same time, the location information of other volunteers who are performing the rescue mission together and the heat map of the search for people are introduced to facilitate the volunteer to plan the next search route, ensure that there will be no omissions in important areas due to inadequate communication and multiple searches of the same area by different team members, and avoid delays in search and rescue or waste of manpower.

[0110] Every search and rescue operation for lost elderly people will be affected by actual conditions such as weather and location. During the search and rescue process, volunteers often need to make judgments and actions based on their own search and rescue experience. However, due to differences in search and rescue experience among volunteers, the execution of rescue activities is difficult to predict, and volunteers often find it difficult to make judgments based on limited search and rescue experience. Therefore, if the information of previous rescue events can be integrated and utilized to provide guidance and suggestions for ongoing rescue events, it will help the rescue. Based on this, every time a task is completed, the system will integrate and count the relevant information in the database, and use the weather interface to obtain necessary information, and then ask volunteers to confirm / supplement the relevant information of the task to ensure the quality of the data. At the same time, the system desensitizes personal information and deletes fields involving privacy. Ultimately, these data will be stored in a non-relational database. Through the RF machine learning algorithm, suggestions are given for the next rescue operation. For example, lost elderly people are usually prone to go indoors in hot weather. If the amount of data is large enough, it can also be predicted that this indoor location is most likely a museum or a nursing home. RF algorithms such as Figure 7 As shown, the implementation details of the RF algorithm in this intelligent rescue solution are introduced below:

[0111] 1. Data Collection:

[0112] Get the rescue data of the elderly, including the elderly's age X 年龄 、Degree of Alzheimer's disease X 程度 、Number of times lost in the past X 走失次数 and other information, as well as geographic locations related to the rescue mission X 地理位置 、Weather conditionsX 天气 At the same time, define the target variable y to indicate whether the task is completed.

[0113] 2. Sample selection:

[0114] A certain proportion of samples are randomly selected with replacement to form the training sample set D for each decision tree. i . It is expressed as:

[0115]

[0116] Where D i is the training sample set of the i-th decision tree, containing N samples.

[0117] 3. Feature Selection:

[0118] For each training sample set, at each node split, a portion of features are randomly selected for training the decision tree. The process of selecting features can be expressed as:

[0119]

[0120] Where m is the number of selected features and p is the total number of features.

[0121] 4. Training decision tree:

[0122] Use the selected features and corresponding labels to train a decision tree. For each decision tree, use a randomly sampled data set and randomly selected features to grow the tree. During the growth process, information entropy is used to measure the impurity of the node. For each node, calculate the information gain of each feature, and select the feature with the largest information gain for partitioning. Where N is the total number of samples of the parent node, N + is the total number of samples of child node i, and k is the number of feature values. The growth of the decision tree will be divided according to the feature with the largest information gain, so that each division can more effectively reduce uncertainty.

[0123] Information entropy is an indicator used to measure the uncertainty or disorder of data. For a probability distribution, the calculation formula for its information entropy is:

[0124]

[0125] Where n is the number of categories, p i is the probability of a certain category.

[0126] The higher the information entropy, the greater the uncertainty of the data. In the context of decision trees, information entropy is used to measure the purity of a node, that is, the samples contained in the node tend to belong to the same category more and more.

[0127] 5. Repeat steps 2-4:

[0128] Repeat the above process, extract samples, select features, train decision trees multiple times, and generate multiple independent decision trees to form a random forest.

[0129] 6. Voting choice (classification question):

[0130] For classification problems, the classification results of each decision tree are voted, and the category with the most votes is selected as the final classification result.

[0131] 7. Average Prediction (Regression Problem):

[0132] For regression problems, the average of the regression results of each decision tree is taken as the final regression result.

[0133] 8. Model output:

[0134] The resulting random forest model can output specific rescue suggestions, such as recommending the number of volunteers, predicting the probability of rescue success and completion time, and providing action suggestions.

[0135] Some of the concepts and formulas involved here include information entropy, information gain, Gini impurity, etc., which are important indicators for node segmentation and evaluation in decision trees. In specific implementations, appropriate evaluation indicators can be selected according to the characteristics of different problems.

[0136] In summary, traditional rescue solutions usually stop processing data after a mission is completed, and do not make good use of the information in the mission data. Compared with traditional rescue solutions, this intelligent rescue solution is based on the RF algorithm and uses historical data to provide auxiliary suggestions for new rescue operations for the volunteers of the rescue team. It predicts the most likely location for the elderly based on the weather, age and other information of this mission, and recommends it to the volunteers. The volunteers plan the route based on their own experience and suggestions. On the premise of effectively promoting the rescue, the rescue system also introduces the location sharing of volunteers under the same mission and the heat map display of the search path. Combining this algorithm and the rescue system can increase the probability of finding the lost elderly, proving the efficiency of this intelligent rescue solution.

[0137] 3. Command System

[0138] like Figure 4 The following is a schematic diagram of the backend command system. The command system is divided into two modules: automation and administrator. The automation module provides system automation support. Its main functions include automatic alarm, automatic task dispatching, automatic task upgrade and downgrade, timely response to face recognition and body recognition; and supports administrators in information management and task management. Figure 5 The figure shows a detailed GIS diagram of a specific task. The administrator can jump from the task details to view the location of the task on the map and the personnel details within a radius of several kilometers, which is convenient for the administrator's back-end staff to assist the family members to modify the elderly's information and manually adjust the task details according to the actual situation. Combined with the previous automatic task dispatching, the management platform is equivalent to providing a platform for staff to manually participate in tasks, which improves the flexibility and efficiency of rescue.

[0139] 4. Data Platform and Third Party

[0140] In order to help the rescue team and other ordinary personnel quickly and more intuitively understand the progress of the rescue team and understand the meaning or changes of the data, the platform displays the relevant information of the lost events during the period. Summarize and count the lost events that occurred in the specified time period, including the age of the lost person, the lost area, the number of rescue personnel and other information, and display them in a graphical manner. Display multi-dimensional and abstract data in front of people in the form of two-dimensional or three-dimensional graphics to achieve interaction between people and information data. Support direct operation on multiple screens to achieve multi-dimensional interaction with the screen. Support multiple interaction methods such as global keyboard and mouse, mouse, and mobile phone control of large screen. Such as Figure 6 The following is a visualization diagram of the data platform for this solution.

[0141] When a rescue mission is completed, the system will integrate and count the relevant information in the database, and use the weather interface to obtain necessary information, and then ask volunteers to confirm / supplement the relevant information of the mission to ensure the quality of the data.

[0142] This system has designed a dedicated 404 page. After the website introduces the interface file of this page, when a 404 error occurs during access, the information of the missing elderly that has not been found will be randomly loaded. Other websites are allowed to access and use it, making use of idle network space to exert public welfare power, allowing more people to help find the lost elderly, and further increasing the data utilization of this solution. Figure 7 Shown is a schematic diagram of the exclusive 404 page designed for this solution.

[0143] Different from other rescue software for lost elderly people, this system has the following features:

[0144] (1) Rapid case filing mechanism: Compared with the cumbersome registration process of traditional rescue solutions, the family members can first report part of the information for case filing, and then supplement it later to prevent delays in rescue due to registration information. In addition, the present invention can also assist special groups in case reporting through back-end management personnel.

[0145] (2) All-round and multi-angle data utilization. Compared with the prior art, the intelligent rescue solution provided by the present invention analyzes from multiple dimensions, dispatches tasks based on the PST algorithm to carry out rescue, collects and utilizes data after the task is completed, and provides some guidance for the next rescue based on the RF algorithm. It also exposes some information data to the outside through the 404 page, and uses the network space to increase the utilization rate of data.

[0146] (3) The end-to-end connection of roles is realized, which greatly improves the success rate, efficiency and flexibility of rescue. For family members: After reporting a case, the case progress can be tracked in real time, supplemented and adjusted according to the information searched by volunteers, and the information is continuously updated in the clue module to improve the success rate of rescue. For volunteers: The platform realizes the automatic dispatching and scheduling of tasks. Volunteers can decide to accept tasks based on their own situation. During the task, they can view the location information of teammates and the heat map of the search route (for example, which sections of the road have been walked by teammates), which is convenient for the volunteer to plan the next search route, ensuring that important areas are not missed due to inadequate communication and that different team members search the same area multiple times, avoiding delays in search and rescue or wasting manpower, and improving the efficiency of rescue. For the command platform: managers can intervene in unexpected or sudden situations and adjust the progress of tasks according to actual conditions, which improves the stability and flexibility of the entire system. The end-to-end connection between family members, rescue teams, and command personnel is truly realized, greatly improving the success rate of rescuing lost elderly people.

[0147] The specific case of this intelligent rescue solution is described as follows:

[0148] Xiao Ming found that his 70-year-old father was missing, so he reported the case on the WeChat applet. He only needed to upload a recent photo of the missing elderly, select the approximate place where he was lost, and fill in the elderly’s name, family members’ names and contact information to successfully report the case. The reported case information is uploaded to the system, and the system calculates the existing clue information in combination with the PTS algorithm, adds the task to the priority queue for sorting, and then the command platform dispatches personnel according to the priority of the task and assigns the task to volunteers in the relevant area. Volunteer A is still performing the previous task, but now there is a more urgent task to search for Xiao Ming’s father. He can choose to accept or reject the task based on his own situation. After that, volunteer B took on the task. The command platform recommended a search location to him through the RF algorithm - the library. Volunteer B planned the search and rescue route based on the GIS geographic information system and the data service toolkit LBS provided by the mini program. At the same time, he also saw the location information of other volunteer C who performed the rescue mission together and the heat map of the path C walked. He learned that C had checked the nearby Alpha Library, so he chose to go to the Beta Library. Finally, he found someone who looked like Xiao Ming's father in the Beta Library. After communicating and taking pictures, he confirmed that the old man was Xiao Ming's missing father (Alzheimer's patients usually do not lose their ability to communicate, they may just forget the specific location of their home), and asked the command platform to complete the task. After the task is completed, the system collects and records the detailed data information of the task and transmits it to the data platform for storage, which will be used for subsequent model training and data display.

[0149] The above description is a specific embodiment of the present invention and the technical principle used. If the changes made according to the concept of the present invention do not exceed the spirit covered by the description and drawings, they should still fall within the scope of protection of the present invention.

Claims

1. An intelligent rescue method for lost elderly people based on PTS algorithm and RF algorithm, comprising the following steps: S1: The family members of the missing elderly can quickly file a case. For the first report, they only need to upload some important information, and then upload the complete information of the missing elderly later; S2: The command platform performs task scheduling and calculates the priority of the collected missing elderly information according to the PTS algorithm. Subsequently, a rescue task notification is sent to the number of volunteers within 5 kilometers of the missing place according to the priority of the task; S3: The search and rescue volunteers check their current location and the location information of other volunteers who are carrying out the rescue mission together on the map according to the rescue system, as well as the heat map of the search, and plan the next search route. S4: If a suspected missing person is found, a photo will be uploaded and automatically compared with all the missing persons in the photo library. The information of the elderly with a matching degree higher than 80% will be synchronized to the family and volunteer terminals. After confirmation by the family and then confirmed by the command platform, the search and rescue mission can be considered completed; S5: If an abnormality occurs during the search and rescue, the backend command platform can check the mission location and rescue details at any time, assist family members to modify the elderly person's information, and manually adjust the allocation of rescue resources according to the actual situation; S6: After the rescue mission is completed, the system integrates and statistically stores relevant information, and then uses the RF machine learning algorithm to assist in the prediction of the approximate lost location for the next search and rescue mission.

2. The method according to claim 1, characterized in that The PTS algorithm described in step S2 includes: S21: Based on the improvement of the priority scheduling algorithm in the operating system, a weight relationship matrix is ​​first established according to the factors of missing time, age of the elderly, health status of the elderly, and climate conditions; S22: Perform a weighted sum operation on the intra-group weight and inter-group weight of the task to obtain the priority of the task, establish a priority queue, and then push it to the corresponding volunteers for rescue.

3. The method according to claim 2, characterized in that The weight relationship matrix described in step S21 is set as follows: The establishment of the weight relationship matrix takes into account the following factors: missing time, age of the elderly, health status of the elderly, climate conditions, and population density. The above factors are set as follows: - For those whose missing time is between 0 and 3 hours, the within-group weight is 0.3 and the between-group weight is 0.3; If the missing time is between 3 and 24 hours, the intra-group weight is 0.5 and the inter-group weight is 0.3; if the missing time exceeds 24 hours, the intra-group weight is 0.2 and the inter-group weight is 0.3; - Elderly people under 70 years old, with a within-group weight of 0.3 and a between-group weight of 0.2; For the elderly aged between 70 and 80, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly aged over 80, the intra-group weight is 0.4 and the inter-group weight is 0.2; - For the elderly with good health, the intra-group weight is 0.2 and the inter-group weight is 0.2; for the elderly with average health, the intra-group weight is 0.3 and the inter-group weight is 0.2; for the elderly with poor health, the intra-group weight is 0.5 and the inter-group weight is 0.2; -For good climate conditions, the intra-group weight is 0.2 and the inter-group weight is 0.1; for average climate conditions, The weight within the group is 0.3, and the weight between groups is 0.1; for severe climatic conditions, the weight within the group is 0.5, and the weight between groups is 0.1; -For high population density, the within-group weight is 0.4 and the between-group weight is 0.2; for medium population density, The weight within the group is 0.3, and the weight between groups is 0.2; for low population density, the weight within the group is 0.3, and the weight between groups is 0.2 The final priority calculation formula is: in: P(i) is the final priority of the i-th task; W i is the intra-group weight corresponding to the i-th task, which is calculated by multiplying the weight of the corresponding condition by the sum of the intra-group weights; R ij It is the element in the i-th row and j-th column in the inter-group weight matrix, indicating the weight relationship between the i-th task and the j-th task.

4. The method according to claim 1, characterized in that Part of the important information described in step S1 includes a recent photo of the elderly and the approximate location where he / she was lost.

5. The method according to claim 1, characterized in that The specific process of S6 includes: S61: After each rescue mission, the existing rescue information will be combined with weather-related information, the data will be desensitized, integrated and stored in the database; S62: Use this data to train a random forest model to provide auxiliary suggestions for the next rescue operation. According to the weather, age and other information of the new mission, the most likely place for the elderly to go is predicted. Volunteers then plan routes to search for the elderly based on their own experience and auxiliary suggestions.

6. The method according to claim 1, characterized in that The places that the elderly described in step S62 are most likely to go include: the place where the elderly was lost in hot weather is indoors, the elderly in their seventies are in a museum, and the elderly in their octogenarians appear in an elderly activity center.

Images