Trajectory tracking method and system for smart positioning of children's school bags

Abstract

The application relates to the technical field of data processing, and discloses a track tracking method and system for intelligently positioning a child's schoolbag. The method comprises the following steps: collecting and processing data of GPS, Bluetooth, Wi-Fi and acceleration sensors installed in the child's schoolbag to obtain multi-source positioning original data; inputting the original data into improved Kalman filtering algorithm fusion processing to obtain position information; constructing a space-time track sequence for the position information to form a historical activity track data set; and executing a DBSCAN clustering algorithm based on the data set to generate child activity route data. The positioning jitter elimination method based on the environment adaptive moving average algorithm effectively solves the positioning fluctuation problem of children in schools, parks and other places. Through the route extraction technology combining DBSCAN clustering and iterative projection fitting, the child's regular activity track and time rule can be accurately identified, and efficient and reliable abnormal behavior detection is realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for tracking the trajectory of children's schoolbags for intelligent positioning. Background Technology

[0002] With societal development, child safety has become an increasingly important concern for parents and society. Parents need to know their children's location in real time to ensure their safety, especially during school commutes and outdoor activities. Currently, various child positioning devices are available on the market, such as watches and wristbands, which typically use GPS and cell tower triangulation technologies to obtain location information. Meanwhile, some research institutions and companies have developed indoor positioning solutions based on technologies like Bluetooth and Wi-Fi, as well as trajectory tracking systems that combine big data analytics. These technologies have already achieved some success in commercial applications, providing basic location tracking functionality.

[0003] However, child tracking devices are easily detected and removed, reducing the effectiveness and sustainability of tracking. Secondly, in specific environments such as indoors and underground passages, signals are easily interfered with, leading to a decrease in positioning accuracy. Thirdly, existing technologies generally suffer from problems such as high power consumption, inaccurate positioning, and insecure data transmission. In particular, the jitter of positioning data is very obvious in complex environments, affecting the reliability of location information. Fourthly, existing devices lack effective technical solutions for analyzing children's activity trajectories and judging abnormal behavior, making it impossible to identify children's regular activity routes and patterns, and difficult to detect abnormal situations in a timely manner.

[0004] To address the issue of positioning jitter in existing technologies, a fixed-length sliding window is used, combined with time weighting and an environment-adaptive smoothing coefficient, to effectively eliminate positioning jitter under different environments. Simultaneously, to solve the problem of inaccurate conventional route recognition, iterative projection and fitting techniques are used to accurately extract representative routes from a large number of historical trajectory points. This, combined with temporal pattern analysis, constructs a children's activity route model, improving the accuracy of trajectory analysis and the reliability of abnormal behavior detection. Furthermore, integrating the positioning device into children's everyday school bags reduces the risk of the device being discovered and removed, improving the continuity and reliability of location tracking. Summary of the Invention

[0005] This application provides a trajectory tracking method and system for intelligent positioning of children's schoolbags. It employs multi-sensor fusion positioning technology, significantly improving positioning accuracy in different environments. A positioning jitter elimination method based on an environment-adaptive moving average algorithm effectively solves the problem of positioning fluctuations in children in places such as schools and parks. Through route extraction technology combining DBSCAN clustering and iterative projection fitting, it accurately identifies children's regular activity trajectories and time patterns, achieving efficient and reliable abnormal behavior detection.

[0006] In a first aspect, this application provides a trajectory tracking method for intelligently positioning children's schoolbags. The trajectory tracking method for intelligently positioning children's schoolbags includes: collecting and processing data from a GPS module, Bluetooth positioning module, Wi-Fi positioning module, and accelerometer installed in the children's schoolbag to obtain multi-source positioning raw data of the children's schoolbag; inputting the multi-source positioning raw data of the children's schoolbag into an improved Kalman filter algorithm for fusion processing to obtain location information; performing spatiotemporal trajectory sequence construction processing on the location information to obtain a historical activity trajectory dataset of the child carrying the schoolbag; and performing DBSCAN clustering algorithm processing on the historical activity trajectory dataset of the child carrying the schoolbag to obtain activity route data of the child carrying the schoolbag.

[0007] Optionally, the step of inputting the multi-source positioning raw data of the child's backpack into an improved Kalman filter algorithm for fusion processing to obtain location information includes:

[0008] The raw data from the GPS module is processed with pseudorange correction, ionospheric delay correction, and tropospheric delay correction to obtain a preliminary GPS position estimate.

[0009] The signal strength data of the Bluetooth positioning module and the Wi-Fi positioning module are processed by the trilateration method and the fingerprint matching algorithm, respectively, to obtain the Bluetooth location estimate and the Wi-Fi location estimate.

[0010] The data from the acceleration sensor is filtered and integrated to obtain the relative displacement change.

[0011] Construct a state vector containing three-dimensional coordinates and three-directional velocity components, and determine the system noise covariance matrix and measurement noise covariance matrix based on the accuracy of each sensor;

[0012] Based on the state vector, the system noise covariance matrix, and the measurement noise covariance matrix, Kalman filtering iterative calculations are performed, including state prediction, Kalman gain calculation, and state update, to obtain the state vector.

[0013] Based on the state vector, longitude, latitude, altitude, speed, and direction data are extracted, and a position accuracy estimate is calculated to generate the position information.

[0014] Optionally, the step of constructing a spatiotemporal trajectory sequence from the location information to obtain a dataset of the child's historical activity trajectory while carrying a schoolbag includes:

[0015] The location information is organized into a trajectory sequence in chronological order, where each location point includes longitude, latitude, timestamp, speed, and direction parameters;

[0016] Abnormal location points in the trajectory sequence are filtered out to remove location points that deviate from the reasonable range due to signal interference when the child is carrying a schoolbag, thus obtaining reliable trajectory data;

[0017] The reliable trajectory data is smoothed by applying a moving average algorithm to eliminate positioning jitter and obtain a smooth trajectory. The positioning jitter includes positioning jitter data caused by the child moving around in school or park.

[0018] The distance the child moves and the location where he stops while carrying his schoolbag are calculated based on the smooth trajectory, and the geographical coordinate range of the target activity area is identified, including the home and school.

[0019] Time analysis was performed on the locations where the children stayed to extract the time patterns of preset types of activities on school days and weekends, and the time probability distribution characteristics were obtained.

[0020] By combining the geographic coordinate range and the temporal probability distribution features, a dataset of historical activity trajectories of children carrying schoolbags is generated.

[0021] Optionally, the reliable trajectory data is smoothed using a moving average algorithm to eliminate positioning jitter and obtain a smooth trajectory. The positioning jitter includes positioning jitter data caused by the child moving around in school or park, including:

[0022] A sliding window with a fixed length of 5 consecutive position points is set for the reliable trajectory data, and each window is slid from the starting point of the trajectory to the ending point in sequence, and the position of each window is processed;

[0023] Calculate the time difference between each position point and the center point within the sliding window, and convert the time difference into a weighting coefficient, which is 1 minus the ratio of the time difference to 30 seconds;

[0024] Multiply the latitude and longitude coordinates of the position points within the sliding window by the corresponding weight coefficients, sum them, and then divide by the sum of the weight coefficients to obtain the smoothed latitude and longitude coordinates of the center point.

[0025] The current environment type is identified by the acceleration sensor data inside the child's backpack. When the acceleration fluctuation value is less than 0.5 meters per square second, it is determined to be an indoor environment, and when it is greater than 2 meters per square second, it is determined to be an outdoor activity environment.

[0026] Set the smoothing coefficient according to the identified environment type: 0.8 for indoor environment, 0.4 for outdoor open environment, and 0.6 for mixed environment.

[0027] The original coordinates and the smoothed coordinates are merged according to the smoothing coefficient ratio to obtain the smoothed trajectory.

[0028] Optionally, the step of processing the historical activity trajectory dataset of the child carrying the schoolbag using the DBSCAN clustering algorithm to obtain the activity route data of the child carrying the schoolbag includes:

[0029] The DBSCAN algorithm parameters were set for the historical activity trajectory dataset of children carrying schoolbags. The algorithm parameters included setting the distance threshold to 50 meters and the minimum number of points to 5, thus obtaining the initial clustering parameters.

[0030] Based on the initial clustering parameters, spatial clustering processing is performed on the historical activity trajectory data. Trajectory points with a distance of less than 50 meters and a number of points greater than 5 are divided into the same cluster to obtain preliminary clustering results.

[0031] For each cluster in the preliminary clustering results, a representative route is extracted by iterative projection and fitting to obtain the representative route of the cluster;

[0032] Based on the clustered representative routes, spatial features of children's activity areas are identified, and regular activity venues are identified according to the length of stay and frequency of visits, thus obtaining data on children's activity areas.

[0033] Based on the time pattern characteristics extracted from the children's activity area data, the children's travel time, route selection, and stay duration on weekdays and weekends were analyzed to obtain activity pattern data;

[0034] A children's activity route model is constructed by combining the cluster representative routes and the activity pattern data, including regular routes, time features and anomaly detection thresholds, and activity route data of children carrying school bags is generated.

[0035] Optionally, the step of extracting representative routes for each cluster in the preliminary clustering results through iterative projection and fitting to obtain representative routes for the clusters includes:

[0036] The set of trajectory points within each cluster in the preliminary clustering results is initialized as a polynomial curve, which is a cubic polynomial function, and serves as the initial representative route.

[0037] Calculate the vertical projection point from each trajectory point within the cluster to the initial representative route, record the position coordinates of the projection point and the time information of the corresponding original point, and obtain the set of projection points;

[0038] Based on the set of projection points, the representative route is refitted using the least squares method, and the coefficients of the polynomial curve are updated to obtain the iterated representative route.

[0039] The difference between the representative route after the iteration and the representative route of the previous iteration is calculated. It is determined whether the difference is less than a preset threshold of 0.5 meters or whether the number of iterations has reached 50, and the iteration termination judgment result is obtained.

[0040] Based on the iteration termination judgment result, determine whether to end the iteration. If the termination condition is not met, return to execute the projection step and fitting until the termination condition is met to obtain the representative route.

[0041] The representative route is uniformly sampled, and a point is selected every 10 meters along the route to generate a representative point sequence. The representative point sequence is used as the representative route for clustering.

[0042] Optionally, the step of identifying spatial features of children's activity areas based on the clustered representative routes, and identifying regular activity venues based on dwell time and visit frequency, to obtain children's activity area data includes:

[0043] Speed ​​analysis is performed on the representative routes of the cluster to identify locations where the speed is lower than a preset speed threshold and the duration exceeds a preset time threshold, and these locations are marked as stop point areas.

[0044] For each stop area, calculate the length of stay and the frequency of visits. The length of stay is the sum of the time spent at all stop points in that area, and the frequency of visits is the proportion of the number of days that the area was visited to the total number of observation days.

[0045] Threshold conditions are set based on the length of stay and frequency of visits. Areas where the stay time is longer than the morning teaching time and the frequency of visits on weekdays is higher than the regular school frequency are identified as schools. Areas where the stay time is longer than the night rest time and the daily frequency of visits is higher than the daily home return frequency are identified as families.

[0046] Geofencing is constructed for areas identified as home and school stops. The convex hull algorithm is used to convert the set of points within the area into a closed polygon boundary, which serves as the core activity area.

[0047] Based on the spatial relationship between the core activity area and the clustered representative routes, the main routes connecting home and school are extracted, and the secondary stops along the way are identified to construct an activity space network.

[0048] The geographic coordinates, boundary ranges, attribute labels, and spatial relationship data of the core activity area, main routes, and secondary stops are integrated to generate children's activity area data.

[0049] Secondly, this application provides a trajectory tracking system for intelligently locating children's schoolbags, the trajectory tracking system for intelligently locating children's schoolbags includes:

[0050] The data acquisition module is used to collect and process data from the GPS module, Bluetooth positioning module, Wi-Fi positioning module and accelerometer sensor installed in the child's backpack to obtain multi-source positioning raw data of the child's backpack.

[0051] The input module is used to input the multi-source positioning raw data of the child's schoolbag into the improved Kalman filter algorithm for fusion processing to obtain location information;

[0052] The construction module is used to perform spatiotemporal trajectory sequence construction processing on the location information to obtain a dataset of historical activity trajectories of children carrying schoolbags;

[0053] The processing module is used to perform DBSCAN clustering algorithm processing on the historical activity trajectory dataset of the children carrying schoolbags to obtain the activity route data of the children carrying schoolbags.

[0054] Thirdly, a trajectory tracking device for intelligently locating children's schoolbags is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the trajectory tracking device for intelligently locating children's schoolbags to execute the aforementioned trajectory tracking method for intelligently locating children's schoolbags.

[0055] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the above-described trajectory tracking method for intelligently locating children's schoolbags.

[0056] The technical solution provided in this application integrates multiple sensors, including a GPS module, Bluetooth positioning module, Wi-Fi positioning module, and accelerometer, into a child's backpack to achieve multi-source positioning data acquisition. This effectively solves the limitations of single positioning technology in complex environments and ensures accurate location information in various scenarios of children's activities. The raw multi-source positioning data is fused using an improved Kalman filter algorithm, fully utilizing the advantages of each sensor and dynamically adjusting the weights of different data sources. This significantly improves the accuracy and reliability of positioning, maintaining good positioning performance, especially in indoor environments and areas with signal interference. This overcomes the problem of decreased positioning accuracy when children enter classrooms, libraries, and other similar locations in existing technologies. The location information is processed to construct a spatiotemporal trajectory sequence, forming a complete historical activity trajectory dataset. This dataset not only records spatial location changes but also includes multi-dimensional information such as time, speed, and direction, providing a rich data foundation for in-depth analysis of children's activity patterns. Based on the historical activity trajectory dataset, the DBSCAN clustering algorithm is applied to automatically identify and extract children's regular routes for daily activities, especially routes to and from school and extracurricular activities, from massive amounts of positioning data, establishing an accurate baseline model for abnormal behavior detection. This invention fully considers the contribution of algorithmic features to the solution when applying artificial intelligence algorithms in specific functional areas. The improved Kalman filter algorithm is optimized for children's activity characteristics, and improves the accuracy of fusion positioning by dynamically adjusting the weights of data from each sensor. The parameter settings of the DBSCAN clustering algorithm are specifically adjusted for the spatial distribution characteristics of children's activity trajectories. The selection of distance thresholds and minimum number of points considers both the accuracy of route recognition and computational efficiency. The moving average algorithm introduces an environmental adaptive mechanism, dynamically adjusting the smoothing coefficient according to different scene characteristics, effectively eliminating positioning jitter for children in places such as schools and parks. These algorithmic optimizations and innovations directly solve specific technical problems in children's location tracking, enabling this solution to provide more accurate, stable, and reliable trajectory tracking services, significantly improving the level of children's travel safety. Integrating the positioning function into children's daily essential backpacks, rather than easily detectable watches or pendants, reduces the risk of the device being discovered and removed, ensuring the continuous effectiveness of the positioning function. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a schematic diagram of an embodiment of a trajectory tracking method for intelligently locating children's schoolbags in this application.

[0059] Figure 2 This is a schematic diagram of one embodiment of a trajectory tracking system for intelligently locating children's schoolbags in this application.

[0060] Figure 3 This is a schematic block diagram of the trajectory tracking device for intelligent positioning of children's schoolbags in an embodiment of the present invention. Detailed Implementation

[0061] This application provides a method and system for tracking the trajectory of a child's backpack for intelligent positioning. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0062] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the trajectory tracking method for intelligently locating children's schoolbags in this application includes:

[0063] Step S101: Collect and process data from the GPS module, Bluetooth positioning module, Wi-Fi positioning module, and accelerometer installed in the child's backpack to obtain multi-source positioning raw data of the child's backpack;

[0064] Step S102: Input the multi-source positioning raw data of the child's schoolbag into the improved Kalman filter algorithm for fusion processing to obtain the location information;

[0065] Step S103: Perform spatiotemporal trajectory sequence construction processing on the location information to obtain a dataset of the historical activity trajectory of children carrying schoolbags;

[0066] Step S104: Based on the historical activity trajectory dataset of children carrying schoolbags, perform DBSCAN clustering algorithm processing to obtain the activity route data of children carrying schoolbags.

[0067] Specifically, the GPS module collects satellite navigation data at a frequency of 1Hz, including longitude, latitude, altitude, speed, and time information; the Bluetooth positioning module scans surrounding beacons every 2 seconds, recording beacon IDs and signal strength; the Wi-Fi positioning module scans visible access points every 5 seconds, recording MAC addresses and signal strength; and the accelerometer continuously collects motion status data. The collection frequencies of these sensors are dynamically adjusted according to the real-time environment. For example, when a strong GPS signal is detected, the Wi-Fi and Bluetooth scanning frequencies are reduced; when entering an indoor environment where the GPS signal weakens, the Wi-Fi and Bluetooth scanning frequencies are increased to ensure the continuity and accuracy of positioning. All collected data is timestamped and stored in a local cache, forming the multi-source positioning raw data for the children's backpack. Pseudorange correction, ionospheric delay correction, and tropospheric delay correction are applied to the raw GPS data to generate a preliminary GPS position estimate; trilateration and fingerprint matching algorithms are applied to the Bluetooth and Wi-Fi signal strength data to calculate the Bluetooth and Wi-Fi position estimates; and the accelerometer data is filtered and integrated to obtain the relative displacement change. Then, a state vector containing three-dimensional coordinates (x, y, z) and three directional velocity components (vx, vy, vz) is constructed, and the system noise covariance matrix and measurement noise covariance matrix are determined based on the accuracy characteristics of each sensor. Kalman filtering iterative calculations are performed based on these parameters, including state prediction, Kalman gain calculation, and state update. Finally, longitude, latitude, altitude, velocity, and direction data are extracted from the state vector, and a position accuracy estimate is calculated to form location information. In practical applications, when a child carrying a schoolbag enters a school building from outdoors, the system automatically switches from a GPS-based positioning mode to an indoor positioning mode based on Wi-Fi and Bluetooth.

[0068] Location information is organized into a trajectory sequence in chronological order, with each location point containing longitude, latitude, timestamp, velocity, and direction parameters. Abnormal location points in the trajectory sequence are then filtered out, eliminating those deviating from reasonable ranges due to signal interference, thus obtaining reliable trajectory data. Next, a moving average algorithm is applied to smooth the reliable trajectory data, eliminating positioning jitter and resulting in a smooth trajectory. In this process, a sliding window of five consecutive location points of fixed length is set, sliding sequentially from the trajectory's starting point to its ending point. The time difference between each location point within the window and the center point is calculated, and this time difference is converted into a weighting coefficient. Then, a weighted average is calculated for the coordinates of the location points within the window. Simultaneously, the system identifies the current environment type (indoor or outdoor) based on accelerometer data and adjusts the smoothing coefficient accordingly. The system calculates the movement distance and stopping position based on the smoothed trajectory, identifying the geographic coordinate range of the target activity location (e.g., home or school); performs time analysis on the stopping positions, extracting preset activity time patterns to obtain time probability distribution characteristics; finally, the geographic coordinate range and time probability distribution characteristics are combined to generate a historical activity trajectory dataset.

[0069] The DBSCAN algorithm parameters are set, including distance threshold and minimum number of points. Then, spatial clustering is performed on historical activity trajectory data, grouping trajectory points that meet the criteria into the same cluster to obtain preliminary clustering results. Next, representative routes are extracted for each cluster through iterative projection and fitting. Specifically, a polynomial curve is initialized as the initial representative route for the set of trajectory points within the cluster. The perpendicular projection point from each trajectory point to the initial representative route is calculated. Based on the set of projection points, the representative route is refitted using the least squares method. This process is repeated until the termination condition is met, ultimately yielding the clustered representative routes. Based on these clustered representative routes, spatial features of children's activity areas are identified. Regular activity locations, such as schools and homes, are identified based on dwell time and visit frequency, constructing an activity space network and ultimately generating activity route data for children carrying backpacks.

[0070] This embodiment addresses the problems of high power consumption, inaccurate positioning, insecure data transmission, and imprecise judgment of abnormal behavior inherent in existing child positioning devices. For example, when a child is on their way to school, the positioning system inside their backpack automatically identifies the usual route from home to school. If the child deviates from this route, the system will immediately issue a warning. Simultaneously, multi-sensor fusion technology ensures positioning accuracy in different environments, solving the problem of signal interference in specific environments found in existing technologies. Furthermore, integrating the positioning device into a child's everyday backpack reduces the risk of the device being discovered and removed, improves the reliability of location tracking, and thus effectively enhances the level of child safety during travel.

[0071] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0072] The raw data from the GPS module is processed with pseudorange correction, ionospheric delay correction and tropospheric delay correction to obtain the preliminary GPS position estimate;

[0073] The signal strength data of the Bluetooth positioning module and the Wi-Fi positioning module are processed by the trilateration method and the fingerprint matching algorithm respectively to obtain the Bluetooth location estimate and the Wi-Fi location estimate.

[0074] The data from the accelerometer is filtered and integrated to obtain the relative displacement change.

[0075] Construct a state vector containing three-dimensional coordinates and three-directional velocity components, and determine the system noise covariance matrix and measurement noise covariance matrix based on the accuracy of each sensor;

[0076] Kalman filtering iterative calculations are performed based on the state vector, system noise covariance matrix, and measurement noise covariance matrix, including state prediction, Kalman gain calculation, and state update, to obtain the state vector.

[0077] Based on the state vector, longitude, latitude, altitude, speed, and direction data are extracted, and position accuracy estimates are calculated to generate position information.

[0078] Specifically, three corrections are applied to the raw GPS data: pseudorange correction eliminates errors caused by satellite and receiver clock bias; ionospheric delay correction addresses the delay caused by the signal passing through the ionosphere, calculated using the Klobuchar model; and tropospheric delay correction corrects for signal offset caused by the atmosphere. These three corrections result in a more accurate preliminary GPS position estimate. Different methods are used to process Bluetooth and Wi-Fi signal strength data. Bluetooth positioning uses trilateration, calculating the relative distance using the signal strength of at least three beacons, and then determining the position using triangulation principles. Wi-Fi positioning applies a fingerprint matching algorithm, matching current signal characteristics against a pre-established database to find the most similar record as the position estimate.

[0079] Accelerometer data is low-pass filtered to remove high-frequency noise, then the gravitational acceleration component is subtracted, and finally double-integrated to obtain displacement. To prevent the accumulation of integration errors, a zero-velocity update is performed when the backpack is detected to be stationary, resetting the velocity value and effectively suppressing drift. A state vector X is constructed, containing position coordinates (x, y, z) and velocity components (vx, vy, vz). The system noise covariance matrix Q and measurement noise covariance matrix R are determined based on actual measurements. System noise mainly comes from the accuracy characteristics of the accelerometer, while measurement noise is dynamically adjusted according to the real-time accuracy of each positioning data source. Kalman filtering is performed, and state prediction calculates the state and covariance for the next moment; Kalman gain calculation determines the optimal weights; state update combines the predicted state and actual measurement values ​​to obtain the optimal estimate. The entire process achieves complementary advantages of each positioning data source, maintaining stable positioning accuracy in different environments. Longitude, latitude, altitude, velocity, and direction data are extracted from the state vector, and geographical location information is obtained through coordinate transformation. The estimated position accuracy is then calculated based on the state covariance matrix. This multi-source fusion method overcomes the limitations of single positioning technology, solves the problem of weak or unavailable GPS signals for children in indoor areas of schools, and ensures the continuity and reliability of positioning.

[0080] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0081] The location information is organized into a trajectory sequence in chronological order, where each location point contains longitude, latitude, timestamp, velocity, and direction parameters;

[0082] Abnormal location points in the trajectory sequence are filtered out to remove location points that deviate from the reasonable range due to signal interference when children are carrying backpacks, thus obtaining reliable trajectory data;

[0083] The reliable trajectory data is smoothed by applying a moving average algorithm to eliminate positioning jitter and obtain a smooth trajectory. Positioning jitter includes positioning jitter data caused by children moving around in school or park.

[0084] The movement distance and stopping position of children carrying backpacks are calculated based on the smooth trajectory, and the geographical coordinate range of the target activity places, including homes and schools, is identified.

[0085] By performing time analysis on the location of stay, the time patterns of children's preset types of activities on school days and weekends are extracted to obtain the time probability distribution characteristics;

[0086] By combining geographic coordinate range and temporal probability distribution features, a dataset of historical activity trajectories of children carrying schoolbags is generated.

[0087] Specifically, location information is organized into a trajectory sequence in chronological order, with each location point containing longitude, latitude, timestamp, speed, and direction parameters. This data is stored in a structured manner for easy subsequent processing and analysis. Filtering outlier locations in the trajectory sequence is a necessary step to ensure data reliability. Outliers are usually caused by signal interference or multipath effects, manifesting as abrupt changes in position relative to preceding and following points or speeds exceeding reasonable ranges. The filtering algorithm uses a speed threshold method, calculating the rate of change of speed between adjacent points. When the rate of change exceeds a preset threshold (e.g., 3 times the normal walking speed), it is identified as an outlier and removed. This method is suitable for detecting location anomalies in children on school grounds caused by building obstructions.

[0088] A moving average algorithm is applied to smooth the reliable trajectory data. A sliding window with a fixed length of 5 consecutive position points is set and slides from the starting point to the ending point of the trajectory. Each window position is processed. The time difference between each point in the window and the center point is calculated, and the time difference is converted into a weighting coefficient. A weighted average is then calculated for the coordinates of the position points in the window. At the same time, the current environment type is identified based on the accelerometer data, and the smoothing coefficient is dynamically adjusted. For example, a larger smoothing coefficient (0.8) is used in a school classroom, while a smaller smoothing coefficient (0.4) is used in open areas such as a playground, effectively eliminating positioning jitter for children in different scenarios.

[0089] The movement distance and stopping positions of children carrying schoolbags are calculated based on smooth trajectories to identify target activity locations. Movement distance is obtained by summing the cumulative distances between adjacent trajectory points, while stopping positions are identified using a combination of speed and time criteria: locations with speeds below 0.5 m / s and durations exceeding 60 seconds are marked as stopping points. Cluster analysis of the spatiotemporal distribution characteristics of these stopping points identifies the geographic coordinate ranges of regular activity locations such as homes and schools. Temporal analysis is performed on stopping positions to extract temporal patterns. The frequency of visits to each identified location at different time periods (e.g., morning, afternoon, evening) and different date types (weekdays, weekends) is statistically analyzed to form temporal probability distribution characteristics. For example, the probability of visiting the school area from 7:30-8:00 AM on school days, and the probability of visiting the home area from 5:00-5:30 PM on school days, constitutes the temporal pattern of children's daily activities. Finally, the geographic coordinate ranges and temporal probability distribution characteristics are combined to generate a historical activity trajectory dataset. This dataset not only contains spatial information but also incorporates temporal patterns, comprehensively reflecting children's activity patterns while carrying schoolbags.

[0090] In one specific embodiment, the process of applying a moving average algorithm to smooth the reliable trajectory data can specifically include the following steps:

[0091] Set a sliding window with a fixed length of 5 consecutive position points for reliable trajectory data, and slide each window sequentially from the starting point to the end point of the trajectory, and process the position of each window;

[0092] Calculate the time difference between each position point and the center point within the sliding window, and convert the time difference into a weighting coefficient. The weighting coefficient is 1 minus the ratio of the time difference to 30 seconds.

[0093] Multiply the latitude and longitude coordinates of the position points within the sliding window by the corresponding weight coefficients, sum them, and then divide by the sum of the weight coefficients to obtain the smoothed latitude and longitude coordinates of the center point.

[0094] The current environment type is identified by the acceleration sensor data inside the child's backpack. When the acceleration fluctuation value is less than 0.5 meters per square second, it is determined to be an indoor environment, and when it is greater than 2 meters per square second, it is determined to be an outdoor activity environment.

[0095] Set the smoothing coefficient according to the identified environment type: 0.8 for indoor environment, 0.4 for outdoor open environment, and 0.6 for mixed environment.

[0096] The original coordinates and the smoothed coordinates are merged according to the smoothing coefficient ratio to obtain the smoothed trajectory.

[0097] Specifically, a sliding window with a fixed length of 5 consecutive location points is set for reliable trajectory data. Five points are chosen because this length effectively eliminates short-term noise without overly smoothing the actual trajectory changes. The operation involves moving this window, containing 5 location points, one point forward each time from the trajectory's starting point until the trajectory's ending point, processing the data within the window one by one. This design ensures that every point on the trajectory (except for the two points at the beginning and end) is processed as a center point. Calculating the time difference between each location point within the sliding window and the center point is the basis for weighted calculation. For any location point within the window, the absolute difference between its timestamp and the center point's timestamp is calculated, and this time difference is then converted into a weighting coefficient. The formula for calculating the weighting coefficient is: Weighting coefficient = 1 - (Time difference / 30 seconds). When the time difference is greater than or equal to 30 seconds, the weighting coefficient is 0. This design gives higher weights to data points closer to the center point in time, thus preserving the temporally continuous trajectory characteristics during smoothing and reducing the influence of data points with large temporal intervals.

[0098] The core step in generating smooth coordinates is calculating a weighted average of the latitude and longitude coordinates of the points within the sliding window. The latitude and longitude coordinates of each point within the window are multiplied by the corresponding weight coefficient obtained in the previous step. Then, all the weighted latitude and longitude values ​​are summed, and finally divided by the sum of all weight coefficients to obtain the smoothed latitude and longitude coordinates of the center point. This weighted averaging method considers the information of all points within the window and emphasizes the importance of points that are more relevant in time through weight coefficients, effectively reducing the impact of outliers. Identifying the current environment type from accelerometer data inside a child's backpack is an innovative aspect of the smoothing process. Acceleration fluctuation value refers to the standard deviation of acceleration data over a period of time (usually 10 seconds). A small fluctuation value indicates a stable motion state, while a large fluctuation value indicates a drastic change in motion state. When the acceleration fluctuation value is less than 0.5 m / s², it is determined to be an indoor environment, because indoor activities are usually relatively calm; when the fluctuation value is greater than 2 m / s², it is determined to be an outdoor activity environment, reflecting the diversity and activity level of outdoor activities; and values ​​in between are determined to be a mixed environment.

[0099] The key to adaptive smoothing is setting different smoothing coefficients according to the identified environment type. In indoor environments (such as classrooms), GPS signals are weak and easily affected by reflections, resulting in higher noise in the positioning data. Therefore, a higher smoothing coefficient (0.8) is set to enhance the smoothing effect. In open outdoor environments (such as playgrounds), GPS signals are good, and the positioning data is more accurate. A lower smoothing coefficient (0.4) is set to retain more original trajectory details. In mixed environments (such as semi-open corridors), a medium smoothing coefficient (0.6) is set. Finally, the original coordinates and the smoothed coordinates are merged according to the smoothing coefficient ratio to obtain the final smoothed trajectory. The specific calculation is: final coordinates = original coordinates × (1 - smoothing coefficient) + smoothed coordinates × smoothing coefficient. This design achieves adaptive smoothing effects in different environments, effectively eliminating jitter in noisy environments while preserving true trajectory characteristics in environments with good signals.

[0100] For example, a primary school student carries a smart positioning backpack and moves around the school, from the classroom to the playground and back to the classroom. The raw trajectory data is first filtered for outliers to obtain reliable trajectory data. When applying the moving average algorithm, for a trajectory segment in the classroom, the time interval between adjacent points is 3 seconds, and the time differences between the center point and the other four points are 6 seconds, 3 seconds, 0 seconds, 3 seconds, and 6 seconds, respectively, which are converted into weighting coefficients of 0.8, 0.9, 1.0, 0.9, and 0.8. The acceleration fluctuation value is 0.3 m / s², which is considered an indoor environment, and a smoothing coefficient of 0.8 is set. Through weighted averaging and coefficient fusion calculation, positioning jitter in the classroom is effectively eliminated. When the student is active on the playground, the acceleration fluctuation value increases to 2.5 m / s², and the smoothing coefficient is automatically adjusted to 0.4, preserving the true movement trajectory on the playground. The entire process adaptively processes trajectory data in different environments, solving the positioning jitter problem for children active in different scenarios.

[0101] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0102] Set DBSCAN algorithm parameters for a dataset of children’s historical activity trajectories with backpacks. The algorithm parameters include setting the distance threshold to 50 meters and the minimum number of points to 5, to obtain the initial clustering parameters.

[0103] Based on the initial clustering parameters, spatial clustering processing is performed on historical activity trajectory data. Trajectory points with a distance of less than 50 meters and a number of points greater than 5 are divided into the same cluster to obtain preliminary clustering results.

[0104] For each cluster in the preliminary clustering results, representative routes are extracted through iterative projection and fitting to obtain representative routes for each cluster.

[0105] Specifically, the trajectory point set within each cluster in the preliminary clustering results is initialized as a polynomial curve, which is a cubic polynomial function, serving as the initial representative route. The vertical projection point from each trajectory point within the cluster to the initial representative route is calculated, and the position coordinates of the projection points and the corresponding time information of the original points are recorded to obtain a projection point set. Based on the projection point set, the representative route is refitted using the least squares method, updating the coefficients of the polynomial curve to obtain the iterated representative route. The difference between the iterated representative route and the previous iteration's representative route is calculated, and it is determined whether the difference is less than a preset threshold of 0.5 meters or whether the number of iterations has reached 50, resulting in an iteration termination judgment. The iteration termination judgment result determines whether to end the iteration; if the termination condition is not met, the projection step and fitting are returned until the termination condition is met, resulting in the representative route. The representative route is uniformly sampled, with a point selected every 10 meters along the route to generate a representative point sequence, which is used as the cluster representative route.

[0106] Based on clustering representative routes, spatial features of children's activity areas are identified, and regular activity venues are identified based on dwell time and visit frequency to obtain children's activity area data;

[0107] Speed ​​analysis was performed on representative routes in the cluster to identify locations where the speed was below a preset speed threshold and the duration exceeded a preset time threshold, marking these as stop-point areas. For each stop-point area, the dwell time and visit frequency were calculated. The dwell time was the sum of the times spent at all stop-points within the area, and the visit frequency was the proportion of days the area was visited out of the total observation days. Threshold conditions were set based on dwell time and visit frequency: areas where the dwell time exceeded morning teaching hours and the weekday visit frequency was higher than the regular school attendance frequency were identified as schools; areas where the dwell time exceeded nighttime rest hours and the daily visit frequency was higher than the daily home attendance frequency were identified as families. Geofencing was constructed for stop-point areas identified as families and schools, using a convex hull algorithm to convert the set of points within the area into closed polygon boundaries, serving as core activity areas. Based on the spatial relationship between the core activity areas and representative routes in the cluster, the main routes connecting families and schools were extracted, and secondary stop-points along the routes were identified, constructing an activity space network. The geographic coordinates, boundary ranges, attribute labels, and spatial relationship data of the core activity areas, main routes, and secondary stop-points were integrated to generate children's activity area data.

[0108] Based on the time pattern characteristics extracted from children's activity area data, we analyzed children's travel time, route selection, and stay duration on weekdays and weekends to obtain activity pattern data;

[0109] A children's activity route model is constructed by combining cluster representative route and activity pattern data, including regular routes, time features and anomaly detection thresholds, to generate activity route data of children carrying school bags.

[0110] Specifically, the DBSCAN algorithm parameters are set, including a distance threshold of 50 meters and a minimum number of points of 5. The distance threshold defines the maximum distance between two trajectory points considered as neighbors. Setting it to 50 meters takes into account the spatial scale of children's daily activities, avoiding both overly broad approaches that could lead to confusion between different routes and overly strict approaches that could cause the same route to be scattered. Setting the minimum number of points to 5 ensures that there are enough points in the cluster to form meaningful routes, avoiding misclassifying accidentally passed areas as regular routes. Based on these initial parameters, spatial clustering is performed on historical activity trajectory data. The DBSCAN algorithm starts from any unvisited point, finds all points in its neighborhood (points less than 50 meters away), and if the number of neighborhood points is greater than 5, a cluster is formed and continuously expanded until no new points can be added. For points outside the core area, if they are in the neighborhood of multiple core points, they are assigned to the corresponding cluster; otherwise, they are marked as noise points. This process is particularly suitable for children's activity trajectory analysis because children often have fixed activity routes (such as the route to school) as well as irregular and occasional movements. The algorithm clusters children’s regular routes into clusters by density partitioning, while identifying occasional activities as noise, thus obtaining preliminary clustering results.

[0111] Extracting representative routes for each cluster through iterative projection and fitting is a crucial step in transforming a discrete set of points into a continuous path. First, the point set within each cluster is initialized as a cubic polynomial curve, in the form f(t) = at³ + bt² + ct + d, serving as the initial representative route. The parameters a, b, c, and d are initially fitted from the point set using the least squares method. A cubic polynomial is chosen because it effectively represents the curvature of the route without overfitting and causing curve oscillations. Next, the perpendicular projection point from each trajectory point to the curve is calculated. This requires solving the problem of finding the shortest distance from each point to the curve. Newton's iteration method is used to find the point on the curve closest to the original point, and the coordinates of the projected point and the corresponding time information of the original point are recorded to form a set of projected points. Based on this set of projected points, the representative route is refitted using the least squares method. The least squares method determines the optimal polynomial coefficients by minimizing the sum of squared distances between the actual points and the fitted curve. This step essentially maps discrete trajectory points onto a continuous curve, updates the polynomial curve parameters, and obtains the iteratively derived representative route. The difference between the old and new representative routes is then calculated by uniformly sampling several points along the curve and calculating the average distance between corresponding points. If the difference is less than 0.5 meters or the number of iterations reaches 50, the iteration terminates; otherwise, the projection and fitting steps are repeated until the termination condition is met. This convergence mechanism ensures the accuracy and computational efficiency of route extraction. The obtained representative routes undergo uniform sampling, selecting a point every 10 meters along the route to generate a representative point sequence, thus realizing the transformation from high-density raw data to a concise representative route. This technique is particularly important in tracking children's backpack trajectories because it is necessary to extract route patterns of children's regular activities from a large number of historical location points.

[0112] Stop-point areas are identified through speed analysis. Specifically, a speed threshold (e.g., 0.5 m / s) and a time threshold (e.g., 60 seconds) are set. When the speed is below the threshold and the duration exceeds the time threshold, the area is marked as a stop-point. Two key indicators are calculated for each stop-point area: stop duration (the total time spent at all stop-points within the area) and visit frequency (the proportion of days the area was visited out of the total observed days). Based on these indicators, discrimination criteria are set; for example, areas with long stop times and high weekday visit frequency are identified as schools, and areas with long stop times and high daily visit frequency are identified as homes. Geofencing is constructed for these core areas, and the convex hull algorithm is used to convert the point set into closed polygon boundaries, extracting the main routes connecting homes and schools, and constructing an activity space network. Temporal pattern features are extracted from the identified activity area data to analyze children's activity patterns at different times. Finally, spatial information and temporal patterns are combined to construct a children's activity route model, including regular routes, temporal features, and anomaly detection thresholds, providing a foundation for abnormal behavior detection.

[0113] For example, after collecting two weeks of data through a smart positioning system inside a schoolbag, cluster analysis was performed using DBSCAN parameters. The algorithm identified three main routes: home to school, school to extracurricular activity center, and weekend leisure route. For the 4300 trajectory points clustered on the home-school route, representative routes were extracted through iterative projection and fitting. After 12 iterations, the difference decreased to 0.4 meters, below the threshold, and the iteration was terminated. Uniform sampling yielded a route consisting of 142 representative points. Speed ​​analysis revealed three main stopping points on this route: the front door (average stay 8 minutes), the intersection (average stay 90 seconds), and the school gate (average stay 5 minutes). Time analysis indicated that elementary school students typically pass through this route around 7:30 AM. Integrating these spatial and temporal features to form an activity route model not only accurately describes children's routine activity patterns but also lays the foundation for abnormal behavior detection, solving the problem of inaccurate analysis of children's activity trajectories in existing technologies.

[0114] The above describes the trajectory tracking method for intelligently locating children's schoolbags in the embodiments of this application. The following describes the trajectory tracking system for intelligently locating children's schoolbags in the embodiments of this application. Please refer to [link / reference]. Figure 2 One embodiment of the trajectory tracking system for intelligently locating children's schoolbags in this application includes:

[0115] The data acquisition module 201 is used to acquire and process data from the GPS module, Bluetooth positioning module, Wi-Fi positioning module and accelerometer installed in the child's backpack to obtain multi-source positioning raw data of the child's backpack.

[0116] Input module 202 is used to input the multi-source positioning raw data of the child's schoolbag into the improved Kalman filter algorithm for fusion processing to obtain location information;

[0117] Construction module 203 is used to perform spatiotemporal trajectory sequence construction processing on the location information to obtain a dataset of historical activity trajectories of children carrying schoolbags;

[0118] The processing module 204 is used to perform DBSCAN clustering algorithm processing on the historical activity trajectory dataset of the children carrying schoolbags to obtain the activity route data of the children carrying schoolbags.

[0119] above Figure 2 The trajectory tracking system for intelligent positioning of children's schoolbags in this embodiment of the invention will be described in detail from the perspective of modular functional entities. The trajectory tracking device for intelligent positioning of children's schoolbags in this embodiment of the invention will be described in detail from the perspective of hardware processing.

[0120] Figure 3 This is a schematic diagram of the structure of a trajectory tracking device for intelligent positioning of children's backpacks provided in an embodiment of the present invention. The trajectory tracking device 300 for intelligent positioning of children's backpacks can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the trajectory tracking device 300 for intelligent positioning of children's backpacks. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the trajectory tracking device 300 for intelligent positioning of children's backpacks to implement the steps of the trajectory tracking method for intelligent positioning of children's backpacks described above.

[0121] The trajectory tracking device 300 for intelligent positioning of children's backpacks may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3The illustrated track tracking device structure for intelligent positioning of children's backpacks does not constitute a limitation on the track tracking device for intelligent positioning of children's backpacks provided by the present invention. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0122] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the trajectory tracking method for intelligently locating children's backpacks.

[0123] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0124] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a trajectory tracking device (which can be a personal computer, server, or network device, etc.) for intelligently locating children's backpacks to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0125] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A trajectory tracking method for intelligent positioning of children's schoolbags, characterized in that, include: Data acquisition and processing were performed on the GPS module, Bluetooth positioning module, Wi-Fi positioning module, and accelerometer installed in the child's backpack to obtain multi-source positioning raw data of the child's backpack; The original multi-source positioning data of the children's schoolbags is input into an improved Kalman filter algorithm for fusion processing to obtain location information; Location information is organized into a trajectory sequence in chronological order, where each location point includes longitude, latitude, timestamp, speed, and direction parameters. Abnormal location points in the trajectory sequence are filtered out, removing those that deviate from the reasonable range due to signal interference when the child is carrying a backpack, thus obtaining reliable trajectory data. A moving average algorithm is applied to the reliable trajectory data to smooth it, eliminating positioning jitter and obtaining a smooth trajectory. Positioning jitter includes data caused by the child moving around at school or in a park. Based on the smooth trajectory, the distance the child travels with the backpack and their stopping position are calculated, identifying the geographical coordinate range of the target activity area, which includes home and school. By performing time analysis on the location of stay, the time patterns of children's preset types of activities on school days and weekends are extracted to obtain the time probability distribution characteristics; By combining geographic coordinate range and temporal probability distribution features, a dataset of historical activity trajectories of children carrying schoolbags is generated. DBSCAN algorithm parameters were set for the historical activity trajectory dataset of children carrying schoolbags; spatial clustering was performed on the historical activity trajectory data based on the initial clustering parameters, and trajectory points that meet the conditions were divided into the same cluster to obtain preliminary clustering results; For each cluster in the preliminary clustering results, representative routes are extracted through iterative projection and fitting to obtain representative routes for each cluster. Spatial features of children's activity areas are identified based on representative cluster routes. Regular activity locations are identified based on dwell time and visit frequency to obtain children's activity area data. Temporal pattern features are extracted from the children's activity area data to analyze children's travel time, route selection, and dwell time on weekdays and weekends to obtain activity pattern data. A children's activity route model is constructed by combining the representative cluster routes and activity pattern data, including regular routes, temporal features, and anomaly detection thresholds, to generate activity route data for children carrying backpacks. For each cluster in the preliminary clustering results, representative routes are extracted through iterative projection and fitting to obtain representative routes for the clusters. This includes: initializing the set of trajectory points within each cluster in the preliminary clustering results into a polynomial curve (a cubic polynomial function) as the initial representative route; calculating the perpendicular projection point from each trajectory point within the cluster to the initial representative route, recording the position coordinates of the projection points and the time information of the corresponding original points to obtain a set of projection points; refitting the representative route using the least squares method based on the set of projection points, updating the coefficients of the polynomial curve to obtain the iterated representative route; calculating the difference between the iterated representative route and the representative route of the previous iteration, determining whether the difference is less than a preset threshold or the number of iterations, and obtaining the iteration termination judgment result; determining whether to end the iteration based on the iteration termination judgment result, and if the termination condition is not met, returning to the projection step and fitting until the termination condition is met to obtain the representative route; uniformly sampling the representative route to generate a representative point sequence, which is used as the cluster representative route.

2. The trajectory tracking method for intelligent positioning of children's schoolbags according to claim 1, characterized in that, The raw data from multi-source positioning of children's backpacks is fused using an improved Kalman filter algorithm to obtain location information. This includes: performing pseudorange correction, ionospheric delay correction, and tropospheric delay correction on the raw data from the GPS module to obtain a preliminary GPS position estimate; applying trilateration and fingerprint matching algorithms to the signal strength data from the Bluetooth and Wi-Fi positioning modules to obtain Bluetooth and Wi-Fi position estimates, respectively; filtering and integrating the data from the accelerometer to obtain the relative displacement change; constructing a state vector containing three-dimensional coordinates and three-directional velocity components, and determining the system noise covariance matrix and measurement noise covariance matrix based on the accuracy of each sensor; performing iterative Kalman filter calculations based on the state vector, system noise covariance matrix, and measurement noise covariance matrix, including state prediction, Kalman gain calculation, and state update, to obtain the state vector; extracting longitude, latitude, altitude, velocity, and direction data from the state vector, calculating the position accuracy estimate, and generating location information.

3. The trajectory tracking method for intelligent positioning of children's schoolbags according to claim 1, characterized in that, The reliable trajectory data is smoothed using a moving average algorithm to eliminate positioning jitter and obtain a smooth trajectory. Positioning jitter includes data caused by the child moving around in school or park. The process involves: setting a fixed-length sliding window of 5 consecutive location points for the reliable trajectory data, and sliding each window sequentially from the trajectory start point to the end point, processing the position of each window; calculating the time difference between each location point and the center point within the sliding window, converting the time difference into a weighting coefficient (1 minus the ratio of the time difference to 30 seconds); and analyzing the latitude and longitude coordinates of the location points within the sliding window. Multiply by the corresponding weighting coefficients, sum the results, and then divide by the sum of the weighting coefficients to obtain the smoothed latitude and longitude coordinates of the center point. Identify the current environment type from the accelerometer data inside the child's backpack. If the acceleration fluctuation value is less than 0.5 m / s², it is determined to be an indoor environment; if it is greater than 2 m / s², it is determined to be an outdoor activity environment. Set the smoothing coefficient according to the identified environment type: 0.8 for indoor environment, 0.4 for outdoor open environment, and 0.6 for mixed environment. Merge the original coordinates and the smoothed coordinates according to the smoothing coefficient ratio to obtain the smoothed trajectory.

4. The trajectory tracking method for intelligent positioning of children's schoolbags according to claim 1, characterized in that, The algorithm parameters include a distance threshold of 50 meters, a minimum number of points of 5, and the classification of trajectory points with a distance of less than 50 meters and a number of points greater than 5 into the same cluster.

5. The trajectory tracking method for intelligent positioning of children's schoolbags according to claim 4, characterized in that, The preset threshold is 0.5 meters or the number of iterations reaches 50; the representative route is uniformly sampled, and a point is selected every 10 meters along the route to generate a representative point sequence.

6. The trajectory tracking method for intelligent positioning of children's schoolbags according to claim 5, characterized in that, Spatial feature identification of children's activity areas is performed based on representative routes formed by clustering. Regular activity locations are identified based on dwell time and visit frequency, resulting in children's activity area data. This includes: speed analysis of representative routes formed by clustering to identify locations where the speed is below a preset speed threshold but the duration exceeds a preset time threshold, marking these as dwell point areas; calculating the dwell time and visit frequency for each dwell point area, where the dwell time is the sum of the times spent at all dwell points within the area, and the visit frequency is the proportion of days the area was visited out of the total number of observation days; and setting threshold conditions based on dwell time and visit frequency, where dwell time exceeds morning teaching hours and weekday visit frequency is higher than [a certain threshold]. Areas with regular school attendance are identified as schools, while areas where children spend more time at school than during nighttime rest and whose daily visit frequency is higher than their daily home return frequency are identified as homes. Geofencing is constructed for these home and school areas, and the set of points within each area is transformed into closed polygon boundaries using a convex hull algorithm, serving as the core activity area. Based on the spatial relationship between the core activity area and representative cluster routes, the main routes connecting homes and schools are extracted, and secondary stops along these routes are identified to construct an activity space network. The geographic coordinates, boundary ranges, attribute labels, and spatial relationship data of the core activity area, main routes, and secondary stops are integrated to generate children's activity area data.

7. A trajectory tracking system for intelligently locating children's schoolbags, characterized in that, For implementing the trajectory tracking method for intelligent positioning of children's schoolbags as described in any one of claims 1-6, the trajectory tracking system for intelligent positioning of children's schoolbags includes: The data acquisition module is used to collect and process data from the GPS module, Bluetooth positioning module, Wi-Fi positioning module and accelerometer sensor installed in the child's backpack to obtain multi-source positioning raw data of the child's backpack. The input module is used to input the multi-source positioning raw data of the child's backpack into the improved Kalman filter algorithm for fusion processing to obtain the location information; The module is used to construct spatiotemporal trajectory sequences from location information to obtain a dataset of historical activity trajectories of children carrying schoolbags. The processing module is used to perform DBSCAN clustering algorithm processing on the historical activity trajectory dataset of children carrying schoolbags to obtain the activity route data of children carrying schoolbags.

8. A trajectory tracking device for intelligently locating children's schoolbags, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the trajectory tracking method for intelligent positioning of children's backpacks as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When a computer program is run by a processor, it causes the processor to execute the trajectory tracking method for intelligently locating a child's backpack as described in any one of claims 1 to 6.

Images