An old people safety protection method based on telecommunication location data

Abstract

The application discloses an old people safety care method based on telecom location data, comprising the following steps: step 1, obtaining mobile terminal signaling data when the old people move and pre-processing the data; step 2, screening effective historical trajectory data of the old people in a preset time period as sample data, calculating the residence time of each residence point according to two time periods of day and night, and describing the time distribution of each residence point of the old people; step 3, using a point of interest recognition algorithm based on the combination of residence time weighting and density clustering algorithm to extract the old people's residence and daily activity data; and step 4, the old people safety care system carries out old people safety care based on the extracted old people's residence and daily activity data. The old people activity area can be mined based on the telecom location data, and the needs of the old people safety care can be met.

Description

Technical Field

[0001] This invention belongs to the field of big data analysis technology, specifically relating to a method for elderly safety care based on telecommunications location data. Background Technology

[0002] With the aging population becoming increasingly serious, young and middle-aged laborers are facing not only work pressure, but also problems such as the lack of care for their parents and children, especially the health and travel safety of the elderly.

[0003] With the development of artificial intelligence and big data analytics, intelligent elderly care is gradually coming into focus. Compared to the data types obtained by internet companies, operator data has advantages such as faster speed, larger scale, and richer variety. Mobile terminal signaling data is characterized by complex spatiotemporal location, large data volume, and high noise levels. Traditional OD methods for identifying points of interest in mobile terminal signaling data suffer from inaccurate results. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing a method for elderly safety care based on telecommunications location data. This method can simply and efficiently eliminate drift data and ping-pong data, and mine the activity area of ​​the elderly based on telecommunications location data, thereby meeting the needs of elderly safety care.

[0005] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for elderly safety care based on telecommunications location data includes:

[0007] Step 1: Acquire the mobile terminal signaling data of the elderly person while they are moving and preprocess the data;

[0008] Step 2: Select valid historical trajectory data of the elderly within a preset time period as sample data, calculate the duration of stay at each stop according to the two time periods of day and night, and depict the time distribution of the elderly at each stop.

[0009] Step 3: Use an interest point identification algorithm that combines dwell time weighting and density clustering to extract data on the elderly’s residence and daily activity locations;

[0010] Step 4: The elderly safety care system provides safety care based on the extracted data of the elderly's residence and daily activity locations.

[0011] To optimize the above technical solution, the specific measures also include:

[0012] The preprocessing described in step 1 above includes:

[0013] First, data cleaning is performed, including checking data consistency, removing invalid values, duplicate values, and missing values;

[0014] Then, by using speed threshold, distance threshold, and triangle circumcircle radius threshold control methods, long-distance drift points and short-distance jitter points are eliminated.

[0015] The specific implementation process for eliminating long-distance drift points and short-distance jitter points is as follows:

[0016] 1) The mobile terminal signaling data of the elderly within a preset time period is regarded as a time series, i.e., P(t) i )=(lon i ,lat i ,ts i ), P(t) i+1 )=(lon i+1 ,lat i+1 ,ts i+1 );

[0017] Where lon represents the longitude of the location of the mobile terminal when the elderly person is moving and interacting with the base station, lat represents the latitude of the location of the mobile terminal when the elderly person is moving and interacting with the base station, and ts represents the timestamp of the interaction between the mobile terminal and the base station.

[0018] 2) Eliminate long-distance drift points: Calculate the distance D between each record and the previous record one by one. i =Dis(P i -P i-1 ), D i+1 =Dis(P i+1 -P i ), and time difference T i =ts i -ts i-1 T i =ts i+1 -ts i The instantaneous velocity at the location point can be expressed as:

[0019] When D i D i+1 If both distance and instantaneous velocity exceed the distance threshold, the drift location point is removed.

[0020] 3) Eliminate short-distance jitter points:

[0021] Calculate the distance D between the i-th record and the (i-1)-th and (i+1)-th records respectively. i =Dis(P i -P i-1 ), Di+1 =Dis(P i+1 -P i );

[0022] The distance D between the (i-1)th record and the (i+1)th record i-1 =Dis(P i+1 -P i-1 );

[0023] Considering that the short-distance jitter points are not on a straight line, then {D i-1 D i D i+1 They can form a triangle;

[0024] Using the three sides of a triangle to represent the radius of its circumcircle, then {D} i-1 D i D i+1 The radius of the circumcircle of the triangle formed by} is expressed as in

[0025] If R > ThR, then remove the i-th record;

[0026] ThR is the radius threshold.

[0027] The specific implementation process of step 2 above for screening sample data is as follows:

[0028] Considering the behavioral patterns of the elderly, the daytime period is defined as 6:00-20:00, and the nighttime period is defined as 20:00-6:00 the next day;

[0029] The criteria for selecting sample data should simultaneously meet the following conditions:

[0030] A. During daytime hours, at least one mobile terminal signaling data record must be generated per hour;

[0031] B. During the nighttime period, more than two mobile terminal signaling data are generated;

[0032] C. More than 20 mobile terminal signaling data entries are generated in a single natural day;

[0033] The specific implementation process of step 2, which describes the distribution of each elderly person's accommodation point, is as follows:

[0034] Define the dwell time at each location as stayLen, the start time of the i-th record as the end time of the (i-1)-th record, and the end time of the i-th record as the appearance time of the i+1-th record, i.e., stayLen = ts i+1Start -ts i-1End ;

[0035] The cumulative duration and frequency of occurrence at each location during both daytime and nighttime periods were statistically analyzed to depict the time distribution of the elderly at each residence point.

[0036] The specific implementation process of the interest point identification algorithm based on the combination of dwell time weighting and density clustering algorithm described in step 3 above is as follows:

[0037] Based on the valid historical trajectory data of the elderly obtained in step 2, density clustering is performed according to daytime and nighttime time periods to generate target clusters;

[0038] For the generated target cluster, each point cluster is further decomposed to obtain all the location points in each point cluster and the dwell time of each location point in the point cluster;

[0039] The location point merging algorithm based on dwell time weighting is used to obtain the set of points of interest P for the daytime period. day The set of points of interest P during the nighttime period night ;

[0040] Define set P night The interest point ranked first in terms of length of stay is the place of residence, and a set P is defined. day The top three points of interest in terms of the length of stay are the places where the elderly usually go, which helps to identify the areas where the elderly usually go.

[0041] The process of generating the target cluster for the daytime sample dataset described above is as follows:

[0042] Input: Daytime time period sample dataset P = (p(t1), p(t2), p(t3), ..., p(t... i )),in

[0043] P(t i )=(lon i ,lat i ,ts i ), radius threshold ε=500, minimum number of points MinPts=2;

[0044] Output: The target cluster generated by clustering is P' = (p(t1), p(t2), p(t3), ..., p(t... j ));

[0045] The specific steps are as follows:

[0046] 1) For each point p in set P, mark the cluster index as the position index of point p in the set, mark the access status as unvisited, and mark the point type as undefined;

[0047] 2) Calculate the number of points Count in the ε-neighborhood of each point p in P. If Count is greater than MinPts, then mark point p as a core point and add it to the core point set P. c ;

[0048] 3) For the core point set P c For each point p in the set of core points, determine whether other points p' in the core point set are within the ε-neighborhood of p; if they are, add p' to the core point set P' of p.

[0049] 4) For the core point set P c For each point p in the set, if point p has not been visited, then take out point p' from the core point set P' of point p in turn. If p' has not been visited, then assign the clustering index of p to p' and mark it as visited.

[0050] 5) For each point p in P, if p is not a core point, then from the core point set P c Take points p' one by one and determine whether point p is within the ε-neighborhood of p'; if it is, mark p as an edge point and assign the clustering index of p' to p;

[0051] 6) Points in P that are not marked as core points or edge points are recorded as noise points.

[0052] The above-mentioned target cluster is further decomposed into all location points in each cluster and the dwell time of each location point in the cluster, including:

[0053] For the target cluster P', each point cluster is further decomposed as geo = [(lon1,lat1),(lon2,lat2),......(lon1,lat1),......(lon2,lat2 ... i ,lat i ] represents all points in each point cluster; staySum = (stayLen1, stayLen1, ..., stayLen) i ) indicates the duration of time each point in the point cluster resides.

[0054] The above implementation process of the location point merging algorithm based on dwell time weighting is as follows:

[0055] 1) Let length be the length of the set of dwell points (geo) and the set of dwell times (staySum), and define the weight coefficient for each location point as follows:

[0056] 2) If length = 1, return geo = (lon1, lat1);

[0057] 3) If length > 1 and i = 1, calculate x1 = cos(lat1) × cos(lon1) × fre1.

[0058] y1=cos(lat1)×sin(lon1)×fre1, z1=sin(lat1)×fre1;

[0059] 4) Similarly, when i > 1, calculate x sequentially. i =x i-1 +cos(lat i )×cos(lon i )×fre i y i =y i-1 +cos(lat i )×sin(lon i )×fre i , z i =z i-1 +sin(lat i )×fre i ;

[0060] 5) The result obtained in (4) is further normalized to obtain

[0061] 6) Finally, the coordinates of the merge point are calculated using the arctangent function, as follows:

[0062] geo(lon,lat)=[degrees(atan(x,y)),degrees(atan2(z,sqrt(x 2 +y 2 )))];

[0063] 7) Following steps 1)-6), solve for the merge points in the interest point cluster P', and finally add the dwell time to the interest points, as shown below:

[0064] P'=[geo(lon1,lat1,staySum1),...,geo(lon i ,lat i ,staySum i )]

[0065] The elderly safety care system described in step 4 above is based on SOA architecture. The system is divided into different components or application services, supports distributed deployment and expansion, and achieves load balancing through the Nginx component. It is logically divided into: presentation layer, business layer and data / persistence layer.

[0066] The presentation layer is responsible for the interaction between the system and users or external systems; the business layer serves the presentation layer and implements business logic processing to meet the needs of the presentation layer; the data layer and persistence layer are responsible for storing system data.

[0067] The elderly safety care system described in step 4 above provides safety care for the elderly based on the extracted data of the elderly's residence and daily activity locations. The specific steps are as follows:

[0068] 1) Real-time location viewing:

[0069] By leveraging the operator's location capabilities, signaling data from the elderly's mobile terminals is collected every 15 minutes to display the area where the elderly's mobile terminals were last located.

[0070] 2) Guarding the activity area:

[0071] A. Set activity range thresholds, including the residence threshold Th home Threshold for daily activities (Th) activity ;

[0072] B. Collect signaling data from the elderly's mobile terminal every 15 minutes;

[0073] C. Calculate the elderly person's latest location and residence. home ;

[0074] D. Calculate the distance from the elderly person's latest location to their usual activity area. activity ;

[0075] E. If D is satisfied simultaneously home >Th home D activity >Th activity The system generates deviation alert information and sends the deviation information of the elderly person's abnormal activity state to their children's mobile terminals.

[0076] F. Set a hospital stay time threshold. hospital When the detection device detects that the elderly person's mobile terminal stays in the area around the hospital for more than Th hospital When necessary, a medical reminder message is generated, and the medical information of the elderly person's abnormal activity status is sent to their children's mobile terminals;

[0077] G. Set time thresholds for major transportation hubs such as train stations. station When the mobile device used to detect elderly people stays in areas around large transportation hubs such as train stations for more than 100 minutes... station When the elderly person's activity is abnormal, a travel reminder message is generated, and the travel information is sent to the children's mobile terminals.

[0078] H. Set time thresholds for dangerous locations such as rivers. dangerWhen the detection device detects that the elderly person's mobile terminal stays in areas around dangerous places such as rivers for more than Th danger At that time, a dangerous location alert message is generated, and the information about the dangerous location where the elderly person's activity is abnormal is sent to their children's mobile terminals;

[0079] 3) Location-based interest statistics:

[0080] An analysis of the elderly person's daily activity areas was conducted, and the areas where the elderly person frequently used their mobile phone over the past month were displayed.

[0081] The present invention has the following beneficial effects:

[0082] To address the challenges of complex spatiotemporal location, large data volume, and high noise levels in mobile terminal signaling data, a simple and efficient method using speed thresholds, distance thresholds, and the radius threshold of the circumcircle of a triangle can be employed to eliminate drift and ping-pong data. Furthermore, through telecommunications location and big data analytics, a Point of Interest (ROI) identification algorithm (RDW-DBSCAN) based on a combination of dwell time weighting and density clustering is proposed. This algorithm solves the problem of inaccurate ROI identification in mobile terminal signaling data using traditional OD methods, effectively identifying ROIs such as the residences and daily activity areas of elderly users.

[0083] Simultaneously addressing the issues of older users' inability to install apps and obtain GPS location information on their mobile phones, a seamless and unobtrusive elderly care system is designed. By fully leveraging telecommunications location information and combining it with data analysis of the elderly's mobile terminal location and behavior, abnormal situations can be accurately identified, and the system can determine the elderly's location, behavior, and other statuses, providing real-time intelligent alerts, including alerts for deviations from residence, deviations from daily activity locations, mobile terminal anomalies, and dangerous locations, ensuring the safety of the elderly. Attached Figure Description

[0084] Figure 1 This is a flowchart of the algorithm for identifying dwell points based on RDW-DBSCAN in this invention.

[0085] Figure 2 This is a pseudocode diagram of the coordinate point merging algorithm based on dwell time weighting of the present invention.

[0086] Figure 3 This is a service framework diagram of the elderly safety care system based on telecommunications location data according to the present invention.

[0087] Figure 4 This is a diagram illustrating the location viewing effect of the elderly safety care system based on telecommunications location data according to the present invention.

[0088] Figure 5 This is a diagram illustrating the activity area protection effect of the elderly safety care system based on telecommunications location data according to the present invention. Detailed Implementation

[0089] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0090] Although the steps in this invention are arranged by reference numerals, this is not intended to limit the order of the steps. Unless the order of the steps is explicitly stated or the execution of a step requires other steps as a basis, the relative order of the steps can be adjusted. It is understood that the term "and / or" as used herein refers to and covers any and all possible combinations of one or more of the associated listed items.

[0091] like Figure 1-5 As shown, the present invention provides a method for elderly safety care based on telecommunications location data, characterized in that it includes:

[0092] Step 1: Acquire the mobile terminal signaling data of the elderly person while they are moving and preprocess the data;

[0093] The preprocessing described in step 1 includes:

[0094] First, data cleaning is performed, including checking data consistency, removing invalid values, duplicate values, and missing values;

[0095] Then, by using speed threshold, distance threshold, and triangle circumcircle radius threshold control methods, long-distance drift points and short-distance jitter points are eliminated.

[0096] The specific implementation process for eliminating long-distance drift points and short-distance jitter points is as follows:

[0097] 1) Consider the mobile terminal signaling data of an elderly person over a certain period of time as a time series, i.e., P(t) i )=(lon i ,lat i ,ts i ), P(t) i+1 )=(lon i+1 ,lat i+1 ,ts i+1 );

[0098] Where lon represents the longitude of the location of the mobile terminal when the elderly person is moving and interacting with the base station, lat represents the latitude of the location of the mobile terminal when the elderly person is moving and interacting with the base station, and ts represents the timestamp of the interaction between the mobile terminal and the base station.

[0099] 2) Eliminate long-distance drift points: Calculate the distance D between each record and the previous record one by one.i =Dis(P i -P i-1 ), D i+1 =Dis(P i+1 -P i ), and time difference T i =ts i -ts i-1 T i =ts i+1 -ts i The instantaneous velocity at the location point can be expressed as:

[0100] When D i D i+1 If the distance exceeds 3km and the instantaneous speed exceeds 90km / h, the drift location point is removed (Note: The distance threshold is set to 3km and the instantaneous speed threshold is set to 90km / h, taking into account the base station density and the elderly's travel patterns. These thresholds can be dynamically adjusted according to the area where the elderly live).

[0101] 3) Eliminate short-distance jitter points:

[0102] Calculate the distance D between the i-th record and the (i-1)-th and (i+1)-th records respectively. i =Dis(P i -P i-1 ), D i+1 =Dis(P i+1 -P i );

[0103] The distance D between the (i-1)th record and the (i+1)th record i-1 =Dis(P i+1 -P i-1 Considering that the short-distance jitter points are not on a straight line, then {D} i-1 D i D i+1 They can form a triangle;

[0104] Using the three sides of a triangle to represent the radius of its circumcircle, then {D} i-1 D i D i+1 The radius of the circumcircle of the triangle formed by} is expressed as in

[0105] Depending on the base station density distribution, the radius threshold ThR can be set between 500 meters and 3 kilometers.

[0106] If R > ThR, then remove the i-th record.

[0107] Step 2: By preprocessing the mobile terminal signaling data during movement, select the valid historical trajectory data of the elderly over the past month as sample data, and calculate the duration of stay at each stop according to the two time periods of day and night to depict the time distribution of the elderly at each stop.

[0108] The specific implementation process of step 2, which involves screening valid sample data and characterizing the distribution of each elderly person's residence location, is as follows:

[0109] 1) Considering the behavioral patterns of the elderly, the daytime period is defined as 6:00-20:00, and the nighttime period is defined as 20:00-6:00 the next day;

[0110] The criteria for selecting valid sample data should simultaneously meet the following conditions:

[0111] A. During daytime hours, at least one mobile terminal signaling data record must be generated per hour;

[0112] B. During the nighttime period, more than two mobile terminal signaling data are generated;

[0113] C. More than 20 mobile terminal signaling data entries are generated in a single natural day;

[0114] 2) Define the dwell time of each location point as stayLen, the start time of the i-th record as the end time of the (i-1)-th record, and the end time of the i-th record as the appearance time of the i+1-th record, i.e., stayLen = ts i+1Start -ts i-1End ;

[0115] 3) Calculate the cumulative duration and frequency of occurrence of each location during the daytime and nighttime periods to depict the time distribution of the elderly at each residence point.

[0116] Step 3: Use the Point of Interest Recognition (RDW-DBSCAN) algorithm, which combines dwell time weighting and density clustering algorithms, to extract data on the elderly’s residence and daily activity locations.

[0117] The specific implementation process of the interest point identification (RDW-DBSCAN) algorithm based on the combination of dwell time weighting and density clustering algorithms described in step 3 is as follows:

[0118] Based on the effective historical trajectory dataset of the elderly obtained in step 2, density clustering (DBSCAN) is performed according to the daytime period (6:00-20:00) and the nighttime period (20:00-6:00) to generate target clusters;

[0119] Take the daytime sample dataset as an example.

[0120] Input: Daytime time period sample dataset P = (p(t1), p(t2), p(t3), ..., p(t... i )),in

[0121] P(t i )=(lon i ,lat i ,ts i ), radius threshold ε=500, minimum number of points MinPts=2.

[0122] Output: The target cluster generated by clustering is P' = (p(t1), p(t2), p(t3), ..., p(t... j )).

[0123] 1) For each point p in set P, mark the cluster index as the position index of point p in the set, mark the access status as unvisited, and mark the point type as undefined.

[0124] 2) Calculate the number of points Count in the ε-neighborhood of each point p in P. If Count is greater than MinPts, then mark point p as a core point and add it to the core point set P. c ;

[0125] 3) For the core point set P c For each point p in the set of core points, determine whether other points p' in the core point set are within the ε-neighborhood of p; if they are, add p' to the core point set P' of p.

[0126] 4) For the core point set P c For each point p in the set, if point p has not been visited, then point p' in the core point set P' of point p is taken in turn. If p' has not been visited, then the clustering index of p is assigned to p' and marked as visited;

[0127] 5) For each point p in P, if p is not a core point, then from the core point set P c Take points p' sequentially. Determine if point p is within the ε-neighborhood of p'; if so, mark p as an edge point and assign the clustering index of p' to p.

[0128] 6) Points in P that are not marked as core points or edge points are recorded as noise points.

[0129] Generally, k-distance is used when selecting parameters for the DBSCAN algorithm. In this scheme, spherical distance is used, and the formula is as follows:

[0130]

[0131] Where α = lat1 - lat2 represents the difference in latitude between the two points, and β = lon1 - lon2 represents the difference in longitude between the two points.

[0132] For the interest point clusters generated above, i.e., the target cluster P', each point cluster is further decomposed as geo = [(lon1,lat1),(lon2,lat2),......(lon1,lat1),......(lon2,lat2 ... i ,lat i ] represents all points in each point cluster; staySum = (stayLen1, stayLen1, ..., stayLen) i ) indicates the duration of time each point in the point cluster resides.

[0133] Next, a location point merging algorithm based on dwell time weighting is designed as follows:

[0134] 1) Let length be the length of the set of dwell points (geo) and the set of dwell times (staySum), and define the weight coefficient for each location point as follows:

[0135] 2) If length = 1, return geo = (lon1, lat1);

[0136] 3) If length > 1 and i = 1, calculate x1 = cos(lat1) × cos(lon1) × fre1.

[0137] y1=cos(lat1)×sin(lon1)×fre1, z1=sin(lat1)×fre1;

[0138] 4) Similarly, when i > 1, calculate x sequentially. i =x i-1 +cos(lat i )×cos(lon i )×fre i y i =y i-1 +cos(lat i )×sin(lon i )×fre i , z i =z i-1 +sin(lat i )×fre i ;

[0139] 5) The result obtained in (4) is further normalized to obtain

[0140] 6) Finally, the coordinates of the merge point are calculated using the arctangent function, as follows:

[0141] geo(lon,lat)=[degrees(atan(x,y)),degrees(atan2(z,sqrt(x 2 +y 2 )))];

[0142] 7) Following steps 1)-6), solve for the merge points in the interest point cluster P', and finally add the dwell time to the interest points, as shown below:

[0143] P'=[geo(lon1,lat1,staySum1),...,geo(lon i ,lat i ,staySum i )];

[0144] Using the interest point identification algorithm (RDW-DBSCAN) described above, which combines dwell time weighting and density clustering, the interest point set P for the daytime period is obtained. day The set of points of interest P during the nighttime period night .

[0145] Based on the behavioral patterns and activity characteristics of the elderly, P is defined. night In the set, the interest point ranked first in terms of stay duration in staySum is the place of residence, and P is defined as... day The top 3 points of interest with the longest stay in the collection were the places where the elderly usually went, thus revealing the areas where the elderly usually went.

[0146] Step 4: The elderly safety care system provides safety care based on the extracted data of the elderly's residence and daily activity locations.

[0147] This invention designs an elderly safety care system based on telecommunications location data, using the elderly's residential and daily activity locations as the specific steps:

[0148] 1) Real-time location viewing

[0149] By leveraging the operator's location capabilities, signaling data from the elderly person's mobile phone terminal is collected every 15 minutes, and the location of the elderly person's mobile phone terminal at the most recent time is displayed.

[0150] 2) Guarding the activity area

[0151] A. Set activity range thresholds, including the residence threshold Th home Threshold for daily activities (Th) activity ;

[0152] B. Collect signaling data from the elderly's mobile terminal every 15 minutes;

[0153] C. Calculate the elderly person's latest location and residence. home ;

[0154] D. Calculate the distance from the elderly person's latest location to their usual activity area. activity ;

[0155] E. If D is satisfied simultaneously home >Th home D activity >Th activity The system generates deviation alert information and sends the deviation information of the elderly person's abnormal activity status to their children's mobile terminals.

[0156] F. Set a hospital stay time threshold. hospital When the detection device detects that the elderly person's mobile terminal stays in the area around the hospital for more than Th hospital When necessary, a medical appointment reminder message is generated, and the medical appointment information of the elderly person with abnormal activity status is sent to their children's mobile terminals.

[0157] G. Set time thresholds for major transportation hubs such as train stations. station When the mobile device used to detect elderly people stays in areas around large transportation hubs such as train stations for more than 100 minutes... station When the elderly person's activity is abnormal, a travel reminder message is generated, and the travel information is sent to the children's mobile terminals.

[0158] H. Set time thresholds for dangerous locations such as rivers. danger When the detection device detects that the elderly person's mobile terminal stays in areas around dangerous places such as rivers for more than Th danger At that time, a dangerous location alert message is generated, and the information about dangerous locations where the elderly person's activity is abnormal is sent to their children's mobile terminals.

[0159] 3) Location Interest Statistics

[0160] Using the point of interest identification algorithm (RDW-DBSCAN) described above, which combines dwell time weighting and density clustering algorithms, the daily activity areas of the elderly are analyzed, and the areas where the elderly frequently use their mobile phones in the past month are displayed.

[0161] The elderly safety and protection system is designed based on SOA architecture. The system is divided into different components or application services, supports distributed deployment and expansion, and uses Nginx to achieve load balancing. Logically, it is divided into: a presentation layer, a business layer, and a data / persistence layer. The presentation layer is responsible for the system's interaction with users or external systems; the business layer serves the presentation layer, mainly implementing business logic processing to meet the presentation layer's needs; the data / persistence layer is responsible for storing system data.

[0162] In summary, the present invention:

[0163] 1. An interest point identification algorithm (RDW-DBSCAN) based on a combination of dwell time weighting and density clustering is proposed. According to the activity patterns of the elderly, it can effectively discover interest points such as the elderly's residence and daily activity locations.

[0164] 2. A method for eliminating short-range jitter points is proposed. Considering the characteristics of complex spatiotemporal location, large data volume, and high noise in mobile terminal signaling data, the method of threshold control of the radius of the circumcircle of a triangle can simply and efficiently eliminate short-range jitter points.

[0165] 3. Design an elderly safety care system based on telecommunications location data. Leveraging the advantages of telecommunications big data, the system provides services such as real-time location viewing, alerts for deviation from activity areas, medical appointment reminders, travel reminders, danger zone alerts, and location interest statistics through intelligent analysis, thereby meeting the needs of children for monitoring and caring for their elderly parents.

[0166] This invention is primarily applied to the safety and protection of the elderly, meeting the needs of children caring for their elders. The advantages of this invention are:

[0167] 1. Compared to mature solutions based on wearable hardware devices, this solution leverages the advantages of telecom location big data to achieve end-to-end processing, accurately identifying abnormal situations of the elderly and meeting the needs of children for monitoring and caring for their elderly parents.

[0168] 2. Compared with the traditional OD method for identifying points of interest in mobile terminal signaling data, we propose an interest point identification algorithm (RDW-DBSCAN) that combines dwell time weighting and density clustering algorithms. This algorithm can quickly and effectively mine points of interest such as the elderly’s residence and daily activity locations, and can achieve a certain level of accuracy.

[0169] 3. In view of the characteristics of mobile terminal signaling data, such as complex spatiotemporal location, large data volume and high noise, a simple and efficient method for eliminating drift data and ping-pong data is designed, which does not require too much time in the data preprocessing stage.

[0170] 4. This solution is simple to deploy and highly portable. Users only need to install the APP (lightweight) to use the functional services.

[0171] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0172] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for safety care of the elderly based on telecommunication location data, characterized by, Comprise: Step 1: Obtain the mobile terminal signaling data when the old people move and preprocess the data, the preprocessing including: Firstly, data cleaning is carried out, including checking data consistency, eliminating invalid format values, repeated values and missing values; Then, long-distance drift position points are eliminated by using speed threshold and distance threshold, and short-distance jitter position points are eliminated by using triangle circumscribed circle radius threshold control method; Step 2: Screen the effective historical trajectory data of the old people in the preset time period as sample data, calculate the residence time of each residence point according to the day and night time period, and describe the time distribution of each residence point of the old people; The specific implementation process of screening sample data is as follows: Considering the behavior rule of the old people, the day time period is defined as: 6:00-20:00, and the night time period is defined as 20:00-6:00 the next day; The sample data screening standard should meet the following conditions at the same time: A. In the day time period, at least 1 mobile terminal signaling data is generated per hour; B. In the night time period, more than 2 mobile terminal signaling data is generated; C. More than 20 mobile terminal signaling data is generated in a natural day; The specific implementation process of describing the residence point distribution of each old person is as follows: The stay duration of each location point is defined as stayLen, the start time of the ith record is the end time of the ith-1 record, the end time of the ith record is the occurrence time of the ith+1 record, i.e. stayLen = ts i+1Start -ts i-1End The cumulative duration and occurrence frequency of each location point in the daytime period and the nighttime period are counted respectively to depict the time distribution of each stay point of the old person. Step 3: An interest point recognition algorithm based on residence time weighted and density clustering algorithm is used to extract the residence and daily activity data of the old people, and the specific implementation process of the interest point recognition algorithm based on residence time weighted and density clustering algorithm is as follows: Based on the effective historical trajectory data of the old people obtained in step 2, density clustering is carried out according to the day time period and the night time period, and target clusters are generated; For the generated target clusters, each point cluster is further decomposed to obtain all position points in each point cluster and the residence time of each position point in the point cluster; The position point merging algorithm based on the length of residence time weighting is used to obtain the interest point set P day in the daytime period and the interest point set P night in the nighttime period, respectively. The set P is defined night The interest point with the first longest residence time in the community is the residence, and the set P is defined day The interest point with the third longest residence time in the community is the daily activity place, and the daily activity area of the old people is further mined. Step 4: The old people safety care system extracts the residence and daily activity data of the old people, and carries out safety care for the old people.

2. The method for the safety protection of the old people based on the telecom location data according to claim 1, characterized in that, The specific implementation process of eliminating long-distance drift position points and short-distance jitter position points is as follows: 1) The mobile terminal signaling data of the old person's movement in a preset time period is regarded as a time sequence, i.e. P(t i ) =(lon i ,lat i ,ts i ), P(t i+1 ) =(lon i+1 ,lat i+1 ,ts i+1 ); Wherein, lon represents the longitude of the position where the mobile terminal interacts with the base station when the old people move, lat represents the latitude of the position where the mobile terminal interacts with the base station when the old people move, and ts represents the time stamp when the mobile terminal interacts with the base station when the old people move; 2) Eliminate long distance drift position points: Calculate the distance D between each record and the previous record one by one i = Dis (P i - P i-1 ), D i+1 = Dis (P i+1 - P i ), and the time difference T i = ts i - ts i-1 , T i = ts i+1 - ts i , then the instantaneous velocity of the position point can be expressed as When D i , D i+1 both exceed the distance threshold, and the instantaneous speed exceeds the instantaneous speed threshold, the drift location point is removed 3) Eliminate short-distance jitter position points: Calculate the distance D of the ith record and the ith-1, ith+1 record respectively i = Dis(P i -P i-1 ), D i+1 = Dis(P i+1 -P i ); Distance D between the (i-1)th record and the (i+1)th record i-1 = Dis(P i+1 - P i-1 ); Considering that the short distance jitter position points are not on a straight line, then {D i-1 , D i , D i+1} can form a triangle; The radius of the circumscribed circle of a triangle is expressed by using three sides of the triangle as i-1 , D i , D i+1} and the radius of the circumscribed circle of a triangle is expressed as where If R>ThR, the i-th record is eliminated; ThR is the radius threshold.

3. The method for the safety protection of the old people based on the telecom location data according to claim 1, characterized in that, For the sample data set of the day time period, the process of generating target clusters is as follows: Input: Daytime period sample data set P = (p(t1), p(t2), p(t3),..., p(t i )), where P(t i ) = (lon i , lat i , ts i ), radius threshold ε = 500, minimum point number MinPts = 2; Output: The target cluster P' generated by clustering = (p(t1), p(t2), p(t3),..., p(tn) ) j )) The specific steps are as follows: 1) For each point p in set P, mark the cluster index as the position index of point p in the set, mark the access state as unvisited, and mark the point type as undefined; 2) Calculate the number of points Count within the ε-neighborhood of each point p in P. If Count is greater than MinPts, then mark point p as a core point and add it to the core point set P c ; 3) For each point p in the core point set P c determine whether other points p' in the core point set are within the ε neighborhood of p. If p' is in P', then p' is added to the core point set P' of p; 4) For each point p in the core point set P c P' of p, if p' has not been visited, assign the cluster index of p to p', and mark p' as visited. 5) For each point p in P, if p is not a core point, then take out points p' from the core point set P c in turn, and judge whether point p is in the ε neighborhood of p'. If p' is in P', then p is marked as an edge point, and the cluster index of p' is assigned to p; 6) The points in P that are not marked as core points or edge points are noise points.

4. The method for the safety protection of the old people based on the telecom location data according to claim 3, characterized in that, The specific implementation process of eliminating long-distance drift position points and short-distance jitter position points is as follows: For target cluster P', each point cluster is further decomposed, geo = [(lon1, lat1), (lon2, lat2),..., (lon i , lat i )] represents all location points in each point cluster; staySum = (stayLen1, stayLen2,..., stayLen i ) represents the duration of each location point in the point cluster.

5. The method for the safety protection of the old people based on the telecom location data according to claim 4, characterized in that, The implementation process of the location point merging algorithm based on the residence time length weighting is as follows: 1) Let the length of the set of residence points geo and the set of residence time lengths staySum be length, define the weight coefficient of each location point as 2) If length = 1, return geo = (lon1, lat1); 3) If length > 1, i = 1, calculate x1 = cos(lat1) x cos(lon1) x fre1, y1 = cos(lat1) x sin(lon1) x fre1, and z1 = sin(lat1) x fre1; 6) Finally, the merging point coordinates are calculated by using the arctangent function as follows: 4) Similarly, when i > 1, calculate x i = x i-1 + cos(lat i ) x cos(lon i ) x fre i , y i = y i-1 + cos(lat i ) x sin(lon i ) x fre i , z i = z i-1 + sin(lat i ) x fre i ; 5) The results obtained in (4) are further normalized to obtain 7) According to steps 1) to 6), the merging points in the point-of-interest cluster P' are solved respectively, and finally the residence time length is added to the point of interest as shown below: geo(lon, lat) = [degrees(atan(x, y)), degrees(atan2(z, sqrt(x 2 +y 2 )))] ; The old person safety care system in step 4 is designed based on the SOA architecture, the whole system is divided into different components or application services, supports distributed deployment and expansion, and realizes load balancing through the Nginx component; according to logical association, it is divided into: presentation layer, business layer and data, persistent layer; P' = [geo(lon1, lat1, staySum1),..., geo(lon i , lat i , staySum i )].

6. The method for the safety protection of the old people based on the telecom location data according to claim 1, characterized in that, Among them, the presentation layer is responsible for the interaction between the system and the user or external system; the business layer serves the presentation layer and realizes the business logic processing to meet the needs of the presentation layer; the data layer and the persistent layer are responsible for the storage of system data. The old person safety care system in step 4 is based on the extracted old person residence and daily activity data, and the specific steps are as follows:

7. The method for the safety protection of the old people based on the telecom location data according to claim 1, characterized in that, 1) Real-time position checking: Through the location capability of the operator, the old person's mobile terminal signaling data is collected every 15 minutes, and the area where the old person's mobile terminal is located recently is displayed; 2) Activity area guardianship: B, collect the old person's mobile terminal signaling data every 15 minutes; A. setting a range of activity threshold, including a residence threshold Th home , a daily activity threshold Th activity ; 3) Position interest statistics: C. Calculate the latest location and residence D of the old person home ; D, calculate the distance D of the old person's latest position daily activity place activity ; E, if D home Th home , D activity Th activity , generates a deviation alert message and sends the deviation message of the old person's abnormal activity state to the child mobile terminal; F, setting a hospital residence time threshold Th hospital when the elderly mobile terminal is detected to reside in the hospital surrounding area for a time exceeding Th hospital , generating a hospital visit reminding information, and sending the hospital visit information of the elderly activity abnormal state to the child mobile terminal; G, setting a time threshold Th for a large transportation hub such as a railway station station When it is detected that the old person's mobile terminal stays in the area around the large transportation hub such as the railway station for more than Th station , a travel reminder information is generated, and the travel information of the old person's abnormal activity state is transmitted to the child's mobile terminal H, setting a time threshold Th for dangerous places such as rivers danger When the time that the old person's mobile terminal stays in the area around the dangerous place such as a river exceeds Th, a dangerous place warning message is generated, and the dangerous place information of the old person's abnormal activity state is sent to the child's mobile terminal. danger ​ The old person's daily activity area is analyzed, and the area where the old person's mobile terminal frequently moves in the past month is displayed. ​

Images