Tray dynamic tracking method and device based on Internet of Things

Abstract

The embodiment of the invention provides a tray dynamic tracking method and device based on the Internet of Things, and the method and device achieve the precise obtaining of positioning through signal fusion and confidence calculation. And a tracking mechanism is constructed, and a reliable trajectory generation strategy is established in combination with state classification and window adjustment. Trajectory optimization is introduced, and continuous improvement of tracking is ensured through breakpoint detection and probability filling. According to the method, the defects of the traditional technology in the aspects of data acquisition, state recognition, track processing and the like are effectively overcome, and technical guarantee is provided for tray tracking.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method and apparatus for dynamic tracking of pallets based on the Internet of Things. Background Technology

[0002] Existing pallet tracking methods have significant shortcomings. Traditional systems perform poorly in data acquisition and signal processing, failing to effectively and accurately locate the pallet and thus affecting tracking performance.

[0003] Furthermore, existing technologies suffer from bottlenecks in state recognition and trajectory generation. Most systems lack robust motion determination mechanisms and sampling adjustment strategies, resulting in suboptimal trajectory recording accuracy.

[0004] The existing system has technical shortcomings in trajectory processing. It lacks in-depth analysis of breakpoints, making it difficult to achieve efficient trajectory completion through probabilistic filling, thus affecting tracking accuracy. Solving these problems is crucial for improving pallet tracking capabilities. Summary of the Invention

[0005] To address the problems in the existing technology, this application provides a method and device for dynamic pallet tracking based on the Internet of Things, which can effectively solve the shortcomings of traditional technologies in data acquisition, status recognition and trajectory processing, and provide technical support for pallet tracking.

[0006] To solve at least one of the above problems, this application provides the following technical solution: In a first aspect, this application provides a pallet dynamic tracking method based on the Internet of Things, comprising: Edge nodes acquire tray sensing data from RFID readers and Bluetooth positioning base stations and aggregate them according to a preset message queue protocol to obtain a raw sensing data set. The raw sensing data set is then used to calculate confidence level based on signal strength threshold conditions and historical stability metrics to obtain confidence-bearing sensing data. The confidence-bearing sensing data of the same tray identifier is then weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector. Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, a motion state label is obtained by classifying the state according to a preset motion judgment threshold. The sampling window length is determined according to the motion state label according to a preset window adjustment rule to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to a weighted mean rule to obtain the perceived event. The cloud receives the perceived events and arranges them according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

[0007] Furthermore, it also includes: edge nodes acquiring tray identification and signal strength data through radio frequency identification readers and acquiring tray location coordinates and timestamp data through Bluetooth positioning base stations to generate multi-source acquisition data streams; and performing field extraction and format standardization processing on the multi-source acquisition data streams according to preset data parsing rules to generate standardized perception data sequences. The standardized sensing data sequence is accessed through a channel and written to a cache queue according to a preset message queue protocol to generate a data queue to be aggregated. The data queue to be aggregated is grouped and aggregated according to the tray identifier and sorted in ascending order according to the timestamp to generate the original sensing data set.

[0008] Furthermore, it also includes: comparing the signal strength of each data in the original sensing data set according to a preset signal strength threshold condition to generate a signal quality score; extracting a historical sensing data sliding window for each data according to the tray identifier and calculating the spatial consistency deviation between the current data and the data in the historical sensing data sliding window to generate a stability metric value; weighting and summing the signal quality score and the stability metric value according to a preset weight configuration to generate a confidence evaluation value and encapsulating it with the corresponding data to generate confidence-based sensing data; The confidence-sensing data with the same tray identifier is timestamped according to the standard time obtained by the network time protocol to generate a time-aligned sensing data sequence. Multiple data with similar timestamps in the time-aligned sensing data sequence are weighted and fused according to the fusion weight after normalization of their respective confidence evaluation values ​​to generate a spatiotemporal feature vector.

[0009] Furthermore, it also includes: calculating the position coordinate difference between adjacent vectors according to the timestamp order to generate a position difference sequence, and calculating the angle between the motion directions of adjacent vectors according to the timestamp order to generate a direction change amplitude sequence; The position difference sequence is subjected to low displacement determination within a continuous period according to a preset static determination threshold condition, and the direction change amplitude sequence is subjected to a sudden change determination according to a preset abrupt change threshold condition to generate motion feature determination results. The motion feature determination results are classified into static state, stable movement state and rapid change state according to a preset state classification rule to generate motion state labels.

[0010] Furthermore, it also includes: matching and querying the motion state marker with the state window mapping table in the preset window adjustment rules to obtain the window length parameter of the corresponding state, and calculating the window start and end time of the window with the current timestamp to generate the sampling window configuration; The position coordinates of the spatiotemporal feature vectors within the sampling window configuration range are weighted and averaged according to the weights after normalization of the confidence evaluation values ​​of each vector to generate the event position. The event position is then encapsulated with the start and end timestamps configured in the sampling window, the motion state marker, and the average confidence within the window to generate a perception event.

[0011] Furthermore, it also includes: the cloud obtains the sensing events uploaded by each edge node through the message receiving interface and groups them according to the tray identifier to generate a tray event group set; the sensing events in each group in the tray event group set are arranged in ascending order of timestamp to generate the original trajectory sequence. The coordinate attribution of each sensing event in the original trajectory sequence is determined according to the preset semantic region boundary definition to generate a region attribution result. The region attribution result is converted into a semantic region code and a region level label is added. Then, the corresponding sensing event is encapsulated with field expansion to generate a semantic trajectory sequence.

[0012] Furthermore, it also includes: calculating the time interval of the timestamps of adjacent perceived events in the semantic trajectory sequence to generate a time interval sequence, performing over-limit detection on the time interval sequence according to a preset breakpoint judgment threshold condition, and extracting the semantic regions and timestamps of the events before and after the over-limit position to generate a trajectory breakpoint set; For each breakpoint in the trajectory breakpoint set, the trajectory inference model is called to calculate the probability of candidate paths based on the spatiotemporal feature vectors of events before and after the breakpoint and the historical flow path statistics. The path with the highest posterior probability is selected to generate a probability filling interval. The probability filling interval is inserted into the corresponding breakpoint position of the semantic trajectory sequence to generate a complete trajectory record. The complete trajectory record is written to persistent storage, and anomaly detection is performed on the complete trajectory record according to the preset dwell threshold conditions and standard flow path rules to generate alarm events.

[0013] Secondly, this application provides a pallet dynamic tracking device based on the Internet of Things, comprising: The tray sensing module is used by edge nodes to acquire tray sensing data through radio frequency identification readers and Bluetooth positioning base stations, and aggregate the data according to a preset message queue protocol to obtain a raw sensing data set. The raw sensing data set is then used to calculate confidence level according to signal strength threshold conditions and historical stability measures to obtain confidence-bearing sensing data. The confidence-bearing sensing data with the same tray identifier is then weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector. The motion labeling module is used to classify the state based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector according to the preset motion judgment threshold to obtain motion state labels, determine the sampling window length according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration, and aggregate the spatiotemporal feature vectors within the sampling window configuration range according to the weighted mean rule to obtain the perceived event. The trajectory tracking module is used to receive the perceived events in the cloud and arrange them according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is then mapped to a semantic trajectory sequence according to a preset semantic region boundary definition. The semantic trajectory sequence is then detected according to the breakpoint judgment threshold condition to obtain a trajectory breakpoint set. Based on the trajectory breakpoint set, a probability filling is performed according to the trajectory inference model to obtain a complete trajectory record. The complete trajectory record is then written to persistent storage and an alarm event is generated.

[0014] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the IoT-based tray dynamic tracking method.

[0015] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the IoT-based tray dynamic tracking method.

[0016] Fifthly, this application provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the IoT-based tray dynamic tracking method.

[0017] As can be seen from the above technical solution, this application provides a method and device for dynamic pallet tracking based on the Internet of Things (IoT). It achieves accurate location acquisition through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively solves the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the IoT-based dynamic pallet tracking method in the embodiments of this application. Figure 2 This is a structural diagram of the IoT-based pallet dynamic tracking device in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0020] Figure label: Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] The acquisition, storage, use, and processing of data in this application comply with relevant laws and regulations.

[0023] To address the problems existing in current technologies, this application provides an IoT-based method and apparatus for dynamic pallet tracking. Through signal fusion and confidence calculation, it achieves accurate location acquisition. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively solves the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking.

[0024] To effectively address the shortcomings of traditional technologies in data acquisition, status recognition, and trajectory processing, and to provide technical support for pallet tracking, this application provides an embodiment of an IoT-based dynamic pallet tracking method. See [link to embodiment]. Figure 1 The IoT-based tray dynamic tracking method specifically includes the following: Step S101: The edge node obtains the pallet sensing data from the RFID reader and Bluetooth positioning base station and aggregates it according to the preset message queue protocol to obtain the original sensing data set. The original sensing data set is used to calculate the confidence level according to the signal strength threshold condition and historical stability metric to obtain the confidence-bearing sensing data. The confidence-bearing sensing data of the same pallet identifier is weighted and fused according to the timestamp alignment rule to obtain the spatiotemporal feature vector. In this embodiment, the edge node scans and reads the tag chip on the tray carrier using an RFID reader to obtain the tray identifier and corresponding signal strength data. Simultaneously, the edge node calculates the coordinates of the tray's location using a Bluetooth positioning base station, obtaining the location coordinates and collection timestamp data. The data streams generated by these two acquisition channels are accessed through a preset message queue protocol to the edge node's local cache queue, and are then grouped and aggregated according to the tray identifier to generate the original sensing data set.

[0025] After the original sensing data set is generated, this embodiment calculates the confidence level of each data point. For the signal strength of each data point, it is compared with a preset signal strength threshold. Data with signal strength higher than the threshold receives a higher signal quality score, while data with signal strength lower than the threshold receives a lower signal quality score. For the spatial location of each data point, this embodiment extracts a sliding window of historical sensing data corresponding to the tray identifier and calculates the deviation distance between the current data's location coordinates and the average location of data within the historical sensing data sliding window. The smaller the deviation distance, the higher the stability metric value.

[0026] Accordingly, this embodiment calculates a confidence assessment value by comprehensively combining the signal quality score and the stability metric. The confidence assessment value is calculated using the formula C = α·G + β·H, where C represents the confidence assessment value, G represents the signal quality score, H represents the stability metric, and α and β represent the signal quality weighting coefficient and the stability weighting coefficient, respectively, with their sum being one. The confidence assessment value and the corresponding original sensing data are then encapsulated to generate sensing data with confidence.

[0027] After the confidence-sensing data is generated, this embodiment performs timestamp alignment processing on multiple data entries with the same tray identifier. Edge nodes obtain a standard time reference via a network time protocol, calibrating the local timestamps of each acquisition channel to a unified time coordinate system, generating a time-aligned sensing data sequence. Multiple data entries in the time-aligned sensing data sequence whose timestamp differences fall within a preset similarity judgment interval are filtered and grouped. Each group of data originates from different acquisition channels but points to the same tray and the same state at the same time.

[0028] Based on the aforementioned grouping results, this embodiment performs weighted fusion of the data within each group. The confidence assessment values ​​of each data point within a group are normalized to obtain a fusion weight. The position coordinates of each data point are then weighted and summed using these fusion weights to obtain fused position coordinates. The motion directions of each data point are then weighted and synthesized using the same fusion weights to obtain a fused motion direction. The fused position coordinates, fused motion direction, fused timestamp, and average confidence are then structurally encapsulated to generate a spatiotemporal feature vector. This spatiotemporal feature vector will be read in subsequent step S102 for determining motion state marking and configuring the adaptive sampling window.

[0029] Step S102: Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, state classification is performed according to the preset motion judgment threshold to obtain motion state labels. The sampling window length is determined according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to the weighted mean rule to obtain the perceived event. Based on the spatiotemporal feature vector generated in step S101, this embodiment calculates the position coordinate difference between adjacent vectors according to their timestamp order. Two adjacent vectors are read sequentially from the spatiotemporal feature vector sequence, their fused position coordinates are extracted, and the Euclidean distance between them is calculated. The distance values ​​of all adjacent vector pairs are arranged in chronological order to generate a position difference sequence. This position difference sequence reflects the spatial displacement change of the tray over a continuous time period.

[0030] After the position difference sequence is generated, this embodiment calculates the directional change amplitude of the spatiotemporal feature vector sequence. The fused motion direction is extracted from two adjacent vectors, the angle between the two direction vectors is calculated, and the angle values ​​of all adjacent vector pairs are arranged in chronological order to generate a directional change amplitude sequence. This directional change amplitude sequence reflects the degree of change in the pallet's motion direction over a continuous time period.

[0031] Accordingly, this embodiment classifies the position difference sequence and the direction change amplitude sequence according to a preset motion determination threshold. When the difference in multiple consecutive periods of the position difference sequence is lower than the preset stationary determination threshold, the motion state is marked as stationary. When the position difference sequence experiences a jump exceeding the preset motion determination threshold and the amplitude value of the corresponding period in the direction change amplitude sequence is lower than the preset steady motion threshold, the motion state is marked as steady movement. When the amplitude value in the direction change amplitude sequence exceeds the preset abrupt change threshold, the motion state is marked as rapid change.

[0032] After the motion state marker is generated, this embodiment determines the sampling window length according to a preset window adjustment rule. The preset window adjustment rule maintains a state window mapping table, and the motion state marker is matched against this table. When the motion state marker is in a stationary state, the upper limit of the stationary window is used as the window length parameter. When the motion state marker is in a smooth moving state, the baseline value of the moving window is used as the window length parameter. When the motion state marker is in a rapidly changing state, the lower limit of the rapidly changing window is used as the window length parameter. The window length parameter is then used to calculate the window start and end times using the current timestamp to generate the sampling window configuration.

[0033] Based on the aforementioned sampling window configuration, this embodiment performs aggregation processing on the spatiotemporal feature vectors within its range. All vectors whose timestamps fall within the start and end intervals of the sampling window configuration are selected from the spatiotemporal feature vector sequence. The confidence evaluation values ​​of each vector are extracted and normalized to obtain aggregation weights. The weighted average of the fused position coordinates of each vector is calculated using the aggregation weights to generate an event location. The event location is encapsulated with the start and end timestamps of the sampling window configuration, the motion state marker, the number of vectors within the window, and the average confidence score to generate a perceived event. This perceived event will be uploaded to the cloud in subsequent step S201 for the construction of the original trajectory sequence and semantic region mapping.

[0034] Step S103: The cloud receives the perceived event and arranges it according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

[0035] The cloud acquires sensing events uploaded by each edge node through a message receiving interface. These sensing events are the aggregation results generated in step S102. The cloud groups all received sensing events according to tray identifiers, grouping sensing events corresponding to the same tray identifier into the same group to generate a tray event group set. The sensing events within each group of the tray event group set are arranged in ascending order of timestamps to generate the original trajectory sequence.

[0036] After the original trajectory sequence is generated, this embodiment performs semantic region mapping processing on it. A preset semantic region boundary definition is maintained in the cloud, which includes the spatial boundary coordinates and hierarchical relationships of each semantic region. Each perceived event in the original trajectory sequence is traversed, and the event location coordinates of each event are extracted and compared with the boundary coordinates of each region in the preset semantic region boundary definition to determine their affiliation. When the event location coordinates fall within the boundary range of a certain semantic region, the semantic region code and region hierarchy label of that region are appended to the corresponding perceived event, generating a semantic trajectory sequence.

[0037] Accordingly, this embodiment performs trajectory breakpoint detection on the semantic trajectory sequence. It iterates through adjacent perceived events in the semantic trajectory sequence, calculates the timestamp interval between adjacent events, and arranges all interval values ​​sequentially to generate a time interval sequence. Each interval value in the time interval sequence is compared with a preset breakpoint determination threshold. When an interval value exceeds the preset breakpoint determination threshold, the position is marked as a trajectory breakpoint, and the semantic region codes and timestamps of the two perceived events before and after the breakpoint are extracted. All breakpoint positions and their preceding and following event information are collected to generate a trajectory breakpoint set.

[0038] After the set of trajectory breakpoints is generated, this embodiment calls the trajectory inference model to perform probability filling for each breakpoint. The input of the trajectory inference model includes the spatiotemporal feature vector of the event before the breakpoint, the spatiotemporal feature vector of the event after the breakpoint, the time span of the breakpoint, and the historical flow path statistics of the tray identifier. The trajectory inference model generates a set of candidate paths and calculates the posterior probability for each candidate path based on the typical traversed area sequence and dwell time distribution between the same start and end semantic region pairs in the historical flow path statistics. The candidate path with the highest posterior probability and satisfying the time constraint is selected as the inference result. The virtual event sequence generated by the inference is labeled with an inference tag and a confidence evaluation value and inserted into the corresponding breakpoint position of the semantic trajectory sequence to generate the probability filling interval.

[0039] Based on the aforementioned insertion results of probability-filled intervals, this embodiment performs integrity repair on the semantic trajectory sequence. All probability-filled intervals and the original perceived events are rearranged in timestamp order to generate a complete trajectory record. The cloud calculates a trajectory integrity index for the complete trajectory record, which represents the effective event coverage rate. The complete trajectory record, along with a unique transaction identifier and a write timestamp, is then written to persistent storage.

[0040] After the complete trajectory record is written, this embodiment performs anomaly detection and alarm generation. The complete trajectory record is segmented according to semantic regions, and the dwell time of the pallet in each semantic region is calculated. The dwell time is compared with the preset normal dwell threshold condition for that region. When the dwell time exceeds the threshold condition, an abnormal dwell alarm is generated. At the same time, the semantic region transfer sequence of the complete trajectory record is extracted and pattern matched with preset standard flow path rules. When an unregistered area or sequence deviation occurs, a path deviation alarm is generated. The abnormal dwell alarm and path deviation alarm are sent to the warehouse management system and the mobile terminals of relevant responsible persons through a message push interface. The alarm message includes pallet identifier, alarm type, trigger location, trigger time, and summary information of associated trajectory segments. The complete trajectory record and alarm events will serve as the final output of pallet dynamic tracking, supporting subsequent inventory status updates and anomaly handling closed loop.

[0041] As described above, the IoT-based dynamic pallet tracking method provided in this application can achieve accurate location acquisition through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively addresses the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking.

[0042] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S201: The edge node obtains the tray identification and signal strength data through the radio frequency identification reader and obtains the tray location coordinates and timestamp data through the Bluetooth positioning base station to generate a multi-source acquisition data stream. The multi-source acquisition data stream is processed by field extraction and format standardization according to the preset data parsing rules to generate a standardized perception data sequence. Step S202: The standardized sensing data sequence is accessed through a channel and written to a cache queue according to a preset message queue protocol to generate a data queue to be aggregated. The data queue to be aggregated is grouped and aggregated according to the tray identifier and sorted in ascending order according to the timestamp to generate the original sensing data set.

[0043] Edge nodes periodically scan the tag chips on the tray carrier using RFID readers to obtain the tray identification and corresponding signal strength data. In each scanning cycle, the RFID reader sends a query command to the tag chips within its coverage area. Upon response, the tag chips return the stored tray identification code. Simultaneously, the reader records the received signal strength value as a signal quality reference for that acquisition.

[0044] While acquiring the tray identification and signal strength data, the edge node calculates the coordinates of the tray's location via a Bluetooth positioning base station. The Bluetooth positioning base station receives broadcast signals emitted by Bluetooth beacons on the tray carrier, calculates the tray's position coordinates based on the signal arrival time difference and signal strength attenuation characteristics, and adds a timestamp to each location data point. The tray identification and signal strength data streams generated by the RFID reader and the location coordinate and timestamp data streams generated by the Bluetooth positioning base station are merged to generate a multi-source acquired data stream.

[0045] Accordingly, this embodiment processes the multi-source acquired data streams according to preset data parsing rules. These preset data parsing rules define the field structure and field extraction methods for each acquired data source. For data generated by the RFID reader, the tray identifier field and signal strength field are extracted. For data generated by the Bluetooth positioning base station, the location coordinate field and timestamp field are extracted. The extracted fields are standardized and encapsulated according to a unified data format, with each field using a fixed byte length and encoding method to generate a standardized sensing data sequence. Each record in the standardized sensing data sequence contains four fields: tray identifier, signal strength, location coordinates, and timestamp.

[0046] Based on the standardized sensing data sequence generated in step S201, this embodiment performs channel access processing according to a preset message queue protocol. Edge nodes deploy a message queue service locally, and the preset message queue protocol defines the communication format and acknowledgment mechanism between the data producer and the message queue. Each record in the standardized sensing data sequence is encapsulated into a message format specified by the message queue protocol and written to a local cache queue through the channel access interface, generating a data queue to be aggregated. The data queue to be aggregated adopts a first-in, first-out (FIFO) storage mechanism to ensure that data is processed in the order of arrival.

[0047] After the data queue to be aggregated is generated, this embodiment performs grouping, aggregation, and sorting processing on it. Data records are read in batches from the data queue to be aggregated, and grouped according to the pallet identifier field, with records having the same pallet identifier grouped into the same group. Records within each group are sorted in ascending order according to the timestamp field, so that the sensing data of the same pallet is arranged in the order of collection time. The sorting results of all groups are aggregated to generate the original sensing data set. The original sensing data set will be read in the subsequent step S301 for confidence calculation and generation of sensing data with confidence.

[0048] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S301: The signal strength of each data in the original sensing data set is compared and judged according to the preset signal strength threshold condition to generate a signal quality score. The historical sensing data sliding window is extracted for each data according to the tray identifier, and the spatial consistency deviation between the current data and the data in the historical sensing data sliding window is calculated to generate a stability metric value. The signal quality score and the stability metric value are weighted and summed according to the preset weight configuration to generate a confidence evaluation value and encapsulated with the corresponding data to generate a confidence sensing data. Step S302: For the confidence-sensing data with the same pallet identifier, timestamp calibration is performed according to the standard time obtained by the network time protocol to generate a time-aligned sensing data sequence. For multiple data with similar timestamps in the time-aligned sensing data sequence, the position coordinates and motion direction are weighted and fused according to the fusion weight after normalization of their respective confidence evaluation values ​​to generate a spatiotemporal feature vector.

[0049] Based on the original sensing data set generated in step S202 above, this embodiment calculates a signal quality score for each data record. All data records in the original sensing data set are traversed, and the signal strength field value of each record is extracted and compared with a preset signal strength threshold. When the signal strength is higher than the preset signal strength threshold, a higher signal quality score is assigned proportionally to the magnitude of the exceedance. When the signal strength is lower than the preset signal strength threshold, a lower signal quality score is assigned proportionally to the magnitude of the fall below the threshold.

[0050] After the signal quality score is generated, this embodiment calculates a stability metric for each data record. For each data record, historical sensing data for that pallet is retrieved according to its pallet identifier field. Several records with the most recent timestamps are extracted from the historical data to form a historical sensing data sliding window. The average position coordinates of all records within the historical sensing data sliding window are calculated as a historical position reference point. The Euclidean distance between the current data record's position coordinates and the historical position reference point is calculated as the spatial consistency deviation. The spatial consistency deviation is converted into a stability metric value according to an inverse proportional mapping rule; the smaller the deviation, the higher the stability metric value.

[0051] Accordingly, this embodiment performs a comprehensive calculation of the signal quality score and the stability metric. The signal quality weight coefficient and stability weight coefficient are read according to a preset weight configuration. The signal quality score is multiplied by the signal quality weight coefficient, and the stability metric is multiplied by the stability weight coefficient, and then summed to generate a confidence assessment value. This confidence assessment value is then encapsulated as a new field along with the corresponding original data record to generate confidence-aware data. Each record in the confidence-aware data includes five fields: tray identifier, signal strength, location coordinates, timestamp, and confidence assessment value.

[0052] Based on the confidence-aware data generated in step S301 above, this embodiment performs timestamp calibration processing on data with the same tray identifier. Edge nodes synchronize their clocks with a standard time server via a network time protocol to obtain a standard time reference. For each record in the confidence-aware data, the deviation between its original timestamp and the standard time reference is calculated. A deviation correction is added to the original timestamp to generate a calibrated timestamp. The calibrated timestamp replaces the original timestamp field, generating a time-aligned awareness data sequence.

[0053] After the time-aligned sensing data sequence is generated, this embodiment performs weighted fusion on multiple data points with similar timestamps. The time-aligned sensing data sequence is traversed, and multiple data points whose calibration timestamp differences fall within a preset similarity judgment interval are grouped into the same fusion group. The confidence assessment value of each data point within each fusion group is normalized, and the fusion weight is obtained by dividing each confidence assessment value by the sum of the confidence assessment values ​​within the group. The fusion weight is used to weight and sum the position coordinates of each data point within the group to obtain the fused position coordinates, and the fusion weight is used to weight and synthesize the movement directions of each data point within the group to obtain the fused movement direction. The fused position coordinates, fused movement direction, fused timestamp, and average confidence within the group are structurally encapsulated to generate a spatiotemporal feature vector. This spatiotemporal feature vector will be read in subsequent step S401 for calculating the position difference sequence and the direction change amplitude sequence.

[0054] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S401: Calculate the position coordinate difference of adjacent vectors according to the timestamp order to generate a position difference sequence; calculate the angle between the motion directions of adjacent vectors according to the timestamp order to generate a direction change amplitude sequence. Step S402: The position difference sequence is subjected to low displacement determination within a continuous period according to the preset static determination threshold condition, and the direction change amplitude sequence is subjected to direction change determination according to the preset abrupt change threshold condition to generate motion feature determination results. The motion feature determination results are classified into static state, stable movement state and rapid change state according to the preset state classification rules to generate motion state labels.

[0055] Based on the spatiotemporal feature vectors generated in step S302 above, this embodiment calculates the position coordinate differences according to the timestamp order. The spatiotemporal feature vectors are arranged in ascending order according to the fused timestamp, and adjacent vectors are read sequentially, with their respective fused position coordinates extracted. For each pair of adjacent vectors, the Euclidean distance between the fused position coordinates of the latter vector and the fused position coordinates of the former vector is calculated, and this distance value is used as the displacement within that time interval. The displacements of all adjacent vector pairs are arranged in chronological order to generate a position difference sequence. Each element in the position difference sequence corresponds to a sampling interval, reflecting the spatial displacement amplitude of the tray within that interval.

[0056] After the position difference sequence is generated, this embodiment calculates the direction change amplitude of the spatiotemporal feature vector. Adjacent vector pairs are traversed in the same timestamp order, and the fused motion direction of each vector is extracted. For each pair of adjacent vectors, the fused motion direction of the preceding and following vectors are considered as unit direction vectors on a two-dimensional plane, and the angle between the two direction vectors is calculated. The angle values ​​of all adjacent vector pairs are arranged in chronological order to generate a direction change amplitude sequence. Each element in the direction change amplitude sequence corresponds to a sampling interval, reflecting the degree of change in the pallet's motion direction within that interval.

[0057] Accordingly, this embodiment performs low displacement determination on the position difference sequence according to a preset static determination threshold. It iterates through each element in the position difference sequence, counting cases where the displacement is lower than the preset static determination threshold for several consecutive sampling intervals. When the number of consecutive intervals reaches a preset static duration period requirement, that period is marked as a low displacement state. Simultaneously, it performs directional change determination on the direction change amplitude sequence according to a preset abrupt change threshold. It iterates through each element in the sequence; when the included angle value of an element exceeds the preset abrupt change threshold, that sampling interval is marked as a directional change state. The low displacement determination results and directional change determination results are aligned and merged according to timestamps to generate motion feature determination results.

[0058] Based on the aforementioned motion feature determination results, this embodiment classifies motion states according to preset state classification rules. The preset state classification rules define the determination conditions and priority relationships for three motion states. When the motion feature determination result for a certain time period shows a low displacement state and no sudden change in direction, the motion state label for that time period is set to a stationary state. When the motion feature determination result for a certain time period shows that the displacement exceeds the preset stationary determination threshold and no sudden change in direction occurs, the motion state label for that time period is set to a steady movement state. When the motion feature determination result for a certain time period shows a sudden change in direction, regardless of the displacement magnitude, the motion state label for that time period is set to a rapidly changing state. The classification results for all time periods are aggregated to generate motion state labels. These motion state labels will be read in subsequent step S501 and used to determine the sampling window length and generate the sampling window configuration.

[0059] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S501: Match the motion state marker with the state window mapping table in the preset window adjustment rules to obtain the window length parameter of the corresponding state, and calculate the window start and end time with the window length parameter and the current timestamp to generate the sampling window configuration; Step S502: The position coordinates of the spatiotemporal feature vectors within the sampling window configuration range are calculated by weighting the normalized weights of each vector confidence evaluation value to generate the event position. The event position is then encapsulated with the start and end timestamps configured in the sampling window, the motion state marker, and the average confidence within the window to generate a perception event.

[0060] Based on the motion state markers generated in step S402 above, this embodiment matches and queries them against the state window mapping table in the preset window adjustment rules. The state window mapping table defines the mapping relationship between each motion state category and its corresponding window length parameter. When the motion state marker is a stationary state, the upper limit of the stationary window is retrieved from the state window mapping table as the window length parameter; this parameter value is relatively large to reduce redundant event generation during the stationary period. When the motion state marker is a smooth movement state, the baseline value of the movement window is retrieved as the window length parameter. When the motion state marker is a rapidly changing state, the lower limit of the rapidly changing window is retrieved as the window length parameter; this parameter value is relatively small to increase the capture density of key turning points.

[0061] After the window length parameter is determined, this embodiment calculates the window start and end times using the current timestamp. The current system timestamp of the edge node is read as the window calculation reference point. Half of the window length parameter is backtracked to obtain the window start timestamp, and half of the window length parameter is extended backward to obtain the window end timestamp. The window start timestamp, window end timestamp, and window length parameter are then structurally encapsulated to generate a sampling window configuration. This sampling window configuration clearly defines the time range covered by this event aggregation.

[0062] Accordingly, this embodiment performs filtering and aggregation processing on the spatiotemporal feature vectors within the configured sampling window range. From the spatiotemporal feature vector sequence generated in step S302, all vectors whose fused timestamps fall within the start and end time intervals of the configured sampling window are selected to form a vector set within the window. The confidence evaluation value of each vector in the vector set within the window is extracted, and each confidence evaluation value is normalized by dividing it by the sum of the built-in confidence evaluation values ​​of the set to obtain the aggregation weight corresponding to each vector.

[0063] Based on the aforementioned aggregation weights, this embodiment calculates a weighted average of the position coordinates of each vector within the window. The fused position coordinates of each vector in the vector set within the window are extracted, and each coordinate value is multiplied by its corresponding aggregation weight and then summed to obtain the weighted position coordinates. These weighted position coordinates are used as the representative position for this aggregation to generate the event position. The event position reflects the overall spatial position of the tray within the sampling window period.

[0064] After the event location is generated, this embodiment encapsulates it and its related attribute fields. The window start timestamp and window end timestamp are read from the sampling window configuration, and the motion state marker for the current time period is read from step S402. The arithmetic mean of the confidence evaluation values ​​of each vector in the vector set within the window is calculated to obtain the average confidence score within the window. The event location, window start timestamp, window end timestamp, motion state marker, number of vectors within the window, and average confidence score within the window are structurally encapsulated to generate a perceived event. This perceived event will be uploaded to the cloud in subsequent step S601 for the construction of the original trajectory sequence.

[0065] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S601: The cloud obtains the sensing events uploaded by each edge node through the message receiving interface and groups them according to the tray identifier to generate a tray event group set. The sensing events in each group in the tray event group set are arranged in ascending order of timestamp to generate the original trajectory sequence. Step S602: Determine the coordinate attribution of the event positions of each sensing event in the original trajectory sequence according to the preset semantic region boundary definition to generate a region attribution result. Convert the region attribution result into a semantic region code and add a region level label. Then, perform field expansion and encapsulation with the corresponding sensing event to generate a semantic trajectory sequence.

[0066] The cloud acquires the sensing events uploaded by each edge node through a message receiving interface. These sensing events are the aggregated results generated in step S502. After generating the sensing events, each edge node packages them according to preset upload batch rules and transmits them to the cloud's message receiving interface via a network channel. The cloud unpacks the received sensing events and extracts the tray identifier field from each sensing event.

[0067] After the sensing events are received, this embodiment groups them according to their tray identifiers. All received sensing events are traversed, and sensing events with the same tray identifier are grouped into the same group. Each group corresponds to all sensing event records for one tray, and each event within a group originates from data collected by that tray at different time periods and within the coverage area of ​​different edge nodes. All groups are aggregated to generate a tray event group set, where each group uses its tray identifier as an index key.

[0068] Accordingly, this embodiment performs timestamp sorting on each group in the tray event grouping set. It iterates through each group in the tray event grouping set, reading the window start timestamp field of all sensing events within the group. The sensing events within each group are then sorted in ascending order according to their window start timestamps, ensuring that sensing events within the same tray are arranged sequentially according to their acquisition time. The sorted grouping results are used as the original trajectory sequence for that tray, reflecting the temporal sensing records of that tray from its first acquisition to its most recent acquisition.

[0069] Based on the original trajectory sequence generated in step S601 above, this embodiment performs semantic region mapping processing on it. A preset semantic region boundary definition is maintained in the cloud. This preset semantic region boundary definition includes the spatial boundary polygon coordinates of each semantic region, a unique region code, and region hierarchy information. The semantic regions are divided according to business scenarios into spatial ranges with clear logistics meanings, such as warehouse inbound area, storage area, picking area, outbound area, and transit area. There are hierarchical relationships of inclusion or adjacency between these areas.

[0070] After the preset semantic region boundary definition is loaded, this embodiment performs coordinate attribution determination for each perceived event in the original trajectory sequence. It iterates through each perceived event in the original trajectory sequence and extracts the event location coordinates for each event. The event location coordinates are then compared with the boundary polygons of each region in the preset semantic region boundary definition to determine the inclusion relationship between the point and the polygon, thus determining the semantic region to which the coordinates belong and generating a region attribution result.

[0071] After the region attribution result is generated, this embodiment performs encoding conversion and field expansion processing. The region name in the region attribution result is converted into a semantic region code according to a preset encoding table. The hierarchical information of the region is read from the preset semantic region boundary definition and appended as a region hierarchical label. The semantic region code and region hierarchical label are encapsulated as new fields along with corresponding sensing events to generate semantic sensing events. After replacing all sensing events in the original trajectory sequence with the corresponding semantic sensing events, a semantic trajectory sequence is generated. This semantic trajectory sequence will be read in subsequent step S701 for trajectory breakpoint detection and breakpoint set generation.

[0072] In one embodiment of the IoT-based tray dynamic tracking method of this application, it may further include the following: Step S701: Calculate the time interval of the timestamps of adjacent perceived events in the semantic trajectory sequence to generate a time interval sequence, perform over-limit detection on the time interval sequence according to the preset breakpoint judgment threshold condition, and extract the semantic regions and timestamps of the events before and after the over-limit position to generate a trajectory breakpoint set. Step S702: For each breakpoint in the trajectory breakpoint set, according to the spatiotemporal feature vectors of events before and after the breakpoint and the historical flow path statistics, call the trajectory inference model to calculate the probability of candidate paths and select the path with the highest posterior probability to generate a probability filling interval. Insert the probability filling interval into the corresponding breakpoint position of the semantic trajectory sequence to generate a complete trajectory record. Write the complete trajectory record into persistent storage and perform anomaly detection on the complete trajectory record according to the preset dwell threshold conditions and standard flow path rules to generate alarm events.

[0073] Based on the semantic trajectory sequence generated in step S602, this embodiment calculates the time interval for adjacent sensing events. It iterates through the sensing events in the semantic trajectory sequence, arranged in ascending order of timestamps, sequentially reading two adjacent events and extracting their respective window start timestamps. For each pair of adjacent events, the difference between the window start timestamp of the subsequent event and the window end timestamp of the preceding event is calculated, and this difference is used as the time interval at that position. The time intervals of all adjacent event pairs are arranged in sequence to generate a time interval sequence.

[0074] After the time interval sequence is generated, this embodiment performs over-limit detection on it according to a preset breakpoint determination threshold. Each element in the time interval sequence is traversed, and each time interval value is compared with the preset breakpoint determination threshold. When a time interval value exceeds the preset breakpoint determination threshold, that position is marked as a breakpoint. For each breakpoint, the semantic region code, timestamp, and event position of the event preceding that position are extracted from the semantic trajectory sequence, and the semantic region code, timestamp, and event position of the event following that position are also extracted. All breakpoint positions and their preceding and following event information are collected to generate a trajectory breakpoint set.

[0075] Accordingly, this embodiment calls the trajectory inference model to perform probability filling for each breakpoint in the trajectory breakpoint set. For each breakpoint, the spatiotemporal feature vectors of the events before and after the breakpoint are read as boundary conditions for inference input, and the difference between the timestamps of the events before and after the breakpoint is calculated to obtain the breakpoint time span. The historical flow path statistics of the tray identifier are retrieved from the cloud historical data storage. The historical flow path statistics record the typical path sequence between semantic region pairs traversed by the tray in the past flow process and the distribution of the dwell time in each region.

[0076] After the inference input is prepared, this embodiment calls the trajectory inference model to generate candidate paths and calculate probabilities. The trajectory inference model retrieves historical records of the same start and end region pairs from the historical flow path statistics based on the semantic region encoding of events before and after the breakpoint, and extracts the sequence of regions traversed in the historical records as a set of candidate paths. For each candidate path, the posterior probability of the path is calculated based on its frequency of occurrence in the historical records and the distribution of dwell time in each region. The candidate path with the highest posterior probability and whose total path duration matches the breakpoint time span is selected as the inference result. Virtual sensing event sequences are generated for each region traversed in the inferred path according to time allocation rules. Each virtual event is then appended with an inference label and inference confidence level and encapsulated into a probability-filled interval.

[0077] Based on the aforementioned probability-filling interval, this embodiment inserts it into the corresponding breakpoint position of the semantic trajectory sequence. The events before and after each breakpoint in the semantic trajectory sequence are located, and the virtual sensing event sequence in the probability-filling interval is inserted in timestamp order between the events before and after the breakpoint. After completing the insertion operation for all breakpoints, the insertion results and the original sensing events are reordered by timestamp to generate a complete trajectory record.

[0078] After the complete trajectory record is generated, this embodiment writes it to persistent storage and performs anomaly detection. A unique transaction identifier is assigned to the complete trajectory record, and a write timestamp is appended before writing it to the cloud database. The complete trajectory record is segmented according to semantic region encoding, and the continuous dwell time of the pallet within each semantic region is calculated. This dwell time is compared with a preset normal dwell threshold for that region; when the dwell time exceeds the threshold, an abnormal dwell alarm is generated. Simultaneously, the semantic region transfer sequence of the complete trajectory record is extracted and pattern matched with preset standard flow path rules. When an unregistered region or sequence deviation occurs, a path deviation alarm is generated. The abnormal dwell alarm and path deviation alarm are sent to the warehouse management system via a message push interface as anomaly event output for pallet dynamic tracking.

[0079] To effectively address the shortcomings of traditional technologies in data acquisition, status recognition, and trajectory processing, and to provide technical support for pallet tracking, this application provides an embodiment of an IoT-based pallet dynamic tracking device for implementing all or part of the aforementioned IoT-based pallet dynamic tracking method. See [link to embodiment]. Figure 2 The IoT-based pallet dynamic tracking device specifically includes the following components: The tray sensing module 10 is used by edge nodes to obtain tray sensing data through radio frequency identification readers and Bluetooth positioning base stations, and to aggregate the data according to a preset message queue protocol to obtain an original sensing data set. The original sensing data set is then used to calculate confidence level according to signal strength threshold conditions and historical stability measures to obtain confidence-bearing sensing data. The confidence-bearing sensing data with the same tray identifier is then weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector. The motion labeling module 20 is used to classify the state based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector according to the preset motion judgment threshold to obtain motion state labels, determine the sampling window length according to the motion state labels according to the preset window adjustment rules to obtain sampling window configuration, and aggregate the spatiotemporal feature vectors within the sampling window configuration range according to the weighted mean rule to obtain the perceived event. The trajectory tracking module 30 is used to receive the perceived events in the cloud and arrange them according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is then mapped to a semantic trajectory sequence according to a preset semantic region boundary definition. The semantic trajectory sequence is then detected according to the breakpoint judgment threshold condition to obtain a trajectory breakpoint set. Based on the trajectory breakpoint set, a probability filling is performed according to the trajectory inference model to obtain a complete trajectory record. The complete trajectory record is then written to persistent storage and an alarm event is generated.

[0080] As described above, the IoT-based pallet dynamic tracking device provided in this application can achieve accurate location acquisition through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively addresses the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking.

[0081] From a hardware perspective, in order to effectively address the shortcomings of traditional technologies in data acquisition, status recognition, and trajectory processing, and to provide technical support for pallet tracking, this application provides an embodiment of an electronic device for implementing all or part of the aforementioned IoT-based dynamic pallet tracking method. The electronic device specifically includes the following components: The system comprises a processor, memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to realize information transmission between the IoT-based tray dynamic tracking device and core business systems, user terminals, and related databases and other related devices; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the IoT-based tray dynamic tracking method and the IoT-based tray dynamic tracking device, the content of which is incorporated herein, and repeated details will not be described again.

[0082] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0083] In practical applications, the IoT-based tray dynamic tracking method can be partially executed on the electronic device side as described above, or all operations can be completed in the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations on this. If all operations are completed in the client device, the client device may further include a processor.

[0084] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0085] Figure 3 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 3 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 3 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0086] In one embodiment, the IoT-based tray dynamic tracking method can be integrated into the central processing unit 9100. The central processing unit 9100 can be configured to perform the following control: Step S101: The edge node obtains the pallet sensing data from the RFID reader and Bluetooth positioning base station and aggregates it according to the preset message queue protocol to obtain the original sensing data set. The original sensing data set is used to calculate the confidence level according to the signal strength threshold condition and historical stability metric to obtain the confidence-bearing sensing data. The confidence-bearing sensing data of the same pallet identifier is weighted and fused according to the timestamp alignment rule to obtain the spatiotemporal feature vector. Step S102: Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, state classification is performed according to the preset motion judgment threshold to obtain motion state labels. The sampling window length is determined according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to the weighted mean rule to obtain the perceived event. Step S103: The cloud receives the perceived event and arranges it according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

[0087] As described above, the electronic device provided in this application achieves accurate positioning through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively addresses the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for tray tracking.

[0088] In another embodiment, the IoT-based pallet dynamic tracking device can be configured separately from the central processing unit 9100. For example, the IoT-based pallet dynamic tracking device can be configured as a chip connected to the central processing unit 9100, and the IoT-based pallet dynamic tracking method function can be implemented through the control of the central processing unit.

[0089] like Figure 3 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 3All components shown; in addition, the electronic device 9600 may also include Figure 3 For components not shown, please refer to existing technologies.

[0090] like Figure 3 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0091] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0092] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0093] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0094] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0095] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0096] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored audio via the speaker 9131.

[0097] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the IoT-based tray dynamic tracking method with a server or client as the execution subject in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the IoT-based tray dynamic tracking method with a server or client as the execution subject in the above embodiments. For example, when the processor executes the computer program, it implements the following steps: Step S101: The edge node obtains the pallet sensing data from the RFID reader and Bluetooth positioning base station and aggregates it according to the preset message queue protocol to obtain the original sensing data set. The original sensing data set is used to calculate the confidence level according to the signal strength threshold condition and historical stability metric to obtain the confidence-bearing sensing data. The confidence-bearing sensing data of the same pallet identifier is weighted and fused according to the timestamp alignment rule to obtain the spatiotemporal feature vector. Step S102: Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, state classification is performed according to the preset motion judgment threshold to obtain motion state labels. The sampling window length is determined according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to the weighted mean rule to obtain the perceived event. Step S103: The cloud receives the perceived event and arranges it according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

[0098] As described above, the computer-readable storage medium provided in this application embodiment achieves accurate location acquisition through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively addresses the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking.

[0099] Embodiments of this application also provide a computer program product capable of implementing all steps of the IoT-based tray dynamic tracking method with the execution subject being a server or client in the above embodiments. When executed by a processor, this computer program / instruction implements the steps of the IoT-based tray dynamic tracking method. For example, the computer program / instruction implements the following steps: Step S101: The edge node obtains the pallet sensing data from the RFID reader and Bluetooth positioning base station and aggregates it according to the preset message queue protocol to obtain the original sensing data set. The original sensing data set is used to calculate the confidence level according to the signal strength threshold condition and historical stability metric to obtain the confidence-bearing sensing data. The confidence-bearing sensing data of the same pallet identifier is weighted and fused according to the timestamp alignment rule to obtain the spatiotemporal feature vector. Step S102: Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, state classification is performed according to the preset motion judgment threshold to obtain motion state labels. The sampling window length is determined according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to the weighted mean rule to obtain the perceived event. Step S103: The cloud receives the perceived event and arranges it according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

[0100] As described above, the computer program product provided in this application achieves accurate positioning through signal fusion and confidence calculation. A tracking mechanism is constructed, combining state classification and window adjustment to establish a reliable trajectory generation strategy. Trajectory optimization is introduced, using breakpoint detection and probability filling to ensure continuous improvement in tracking. This method effectively addresses the shortcomings of traditional technologies in data acquisition, state recognition, and trajectory processing, providing technical support for pallet tracking.

[0101] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0102] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0103] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0104] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0105] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for dynamic pallet tracking based on the Internet of Things, characterized in that, The method includes: Edge nodes acquire tray sensing data from RFID readers and Bluetooth positioning base stations and aggregate them according to a preset message queue protocol to obtain a raw sensing data set. The raw sensing data set is then used to calculate confidence level based on signal strength threshold conditions and historical stability metrics to obtain confidence-bearing sensing data. The confidence-bearing sensing data of the same tray identifier is then weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector. Based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector, a motion state label is obtained by classifying the state according to a preset motion judgment threshold. The sampling window length is determined according to the motion state label according to a preset window adjustment rule to obtain the sampling window configuration. The spatiotemporal feature vectors within the sampling window configuration range are aggregated according to a weighted mean rule to obtain the perceived event. The cloud receives the perceived events and arranges them according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is mapped according to the preset semantic region boundary definition to obtain the semantic trajectory sequence. The semantic trajectory sequence is detected according to the breakpoint judgment threshold condition to obtain the trajectory breakpoint set. Based on the trajectory breakpoint set, the trajectory inference model is used to fill the probability to obtain the complete trajectory record. The complete trajectory record is written to persistent storage and an alarm event is generated.

2. The IoT-based tray dynamic tracking method according to claim 1, characterized in that, The edge nodes acquire tray sensing data from the RFID reader and Bluetooth positioning base station, and aggregate the data according to a preset message queue protocol to obtain the raw sensing data set, including: Edge nodes acquire tray identification and signal strength data through radio frequency identification readers and obtain tray location coordinates and timestamp data through Bluetooth positioning base stations to generate multi-source data streams. The multi-source data streams are then processed for field extraction and format standardization according to preset data parsing rules to generate standardized sensing data sequences. The standardized sensing data sequence is accessed through a channel and written to a cache queue according to a preset message queue protocol to generate a data queue to be aggregated. The data queue to be aggregated is grouped and aggregated according to the tray identifier and sorted in ascending order according to the timestamp to generate the original sensing data set.

3. The IoT-based dynamic pallet tracking method according to claim 1, characterized in that, The process involves calculating confidence levels for the original sensing data set based on signal strength threshold conditions and historical stability metrics to obtain confidence-bearing sensing data. Then, the confidence-bearing sensing data with the same tray identifier is weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector, including: The signal strength of each data point in the original sensing data set is compared and judged according to a preset signal strength threshold condition to generate a signal quality score. The historical sensing data sliding window is extracted for each data point according to the tray identifier, and the spatial consistency deviation between the current data and the data in the historical sensing data sliding window is calculated to generate a stability metric value. The signal quality score and the stability metric value are weighted and summed according to a preset weight configuration to generate a confidence evaluation value, and then encapsulated with the corresponding data to generate confidence-bearing sensing data. The confidence-sensing data with the same tray identifier is timestamped according to the standard time obtained by the network time protocol to generate a time-aligned sensing data sequence. Multiple data with similar timestamps in the time-aligned sensing data sequence are weighted and fused according to the fusion weight after normalization of their respective confidence evaluation values ​​to generate a spatiotemporal feature vector.

4. The IoT-based tray dynamic tracking method according to claim 1, characterized in that, The motion state label is obtained by classifying the position difference sequence and direction change amplitude based on the spatiotemporal feature vector according to a preset motion determination threshold condition, including: The position difference sequence is generated by calculating the position coordinate difference of adjacent vectors according to the timestamp order of the spatiotemporal feature vectors; the direction change amplitude sequence is generated by calculating the angle between the motion directions of adjacent vectors according to the timestamp order of the spatiotemporal feature vectors. The position difference sequence is subjected to low displacement determination within a continuous period according to a preset static determination threshold condition, and the direction change amplitude sequence is subjected to a sudden change determination according to a preset abrupt change threshold condition to generate motion feature determination results. The motion feature determination results are classified into static state, stable movement state and rapid change state according to a preset state classification rule to generate motion state labels.

5. The IoT-based dynamic pallet tracking method according to claim 1, characterized in that, The step of determining the sampling window length according to the motion state marker and a preset window adjustment rule to obtain the sampling window configuration, and aggregating the spatiotemporal feature vectors within the sampling window configuration range according to a weighted average rule to obtain the perceived event includes: The motion state marker is matched with the state window mapping table in the preset window adjustment rules to obtain the window length parameter of the corresponding state. The window length parameter is then used to calculate the window start and end time with the current timestamp to generate the sampling window configuration. The position coordinates of the spatiotemporal feature vectors within the sampling window configuration range are weighted and averaged according to the weights after normalization of the confidence evaluation values ​​of each vector to generate the event position. The event position is then encapsulated with the start and end timestamps configured in the sampling window, the motion state marker, and the average confidence within the window to generate a perception event.

6. The IoT-based dynamic pallet tracking method according to claim 1, characterized in that, The cloud receives the perceived events and arranges them according to tray identifiers and timestamps to obtain an original trajectory sequence. The original trajectory sequence is then mapped according to a preset semantic region boundary definition to obtain a semantically encoded trajectory sequence, including: The cloud obtains the sensing events uploaded by each edge node through the message receiving interface and groups them according to the tray identifier to generate a tray event group set. The sensing events in each group in the tray event group set are arranged in ascending order of timestamp to generate the original trajectory sequence. The coordinate attribution of each sensing event in the original trajectory sequence is determined according to the preset semantic region boundary definition to generate a region attribution result. The region attribution result is converted into a semantic region code and a region level label is added. Then, the corresponding sensing event is encapsulated with field expansion to generate a semantic trajectory sequence.

7. The IoT-based tray dynamic tracking method according to claim 1, characterized in that, The process of detecting trajectory breakpoints in the semantic trajectory sequence according to the breakpoint determination threshold condition, obtaining a trajectory breakpoint set, performing probability filling based on the trajectory breakpoint set according to the trajectory inference model to obtain a complete trajectory record, writing the complete trajectory record to persistent storage and generating an alarm event includes: The time interval sequence is generated by calculating the interval between the timestamps of adjacent perceived events in the semantic trajectory sequence. The time interval sequence is then subjected to over-limit detection according to the preset breakpoint judgment threshold condition, and the semantic regions and timestamps of the events before and after the over-limit position are extracted to generate a trajectory breakpoint set. For each breakpoint in the trajectory breakpoint set, the trajectory inference model is called to calculate the probability of candidate paths based on the spatiotemporal feature vectors of events before and after the breakpoint and the historical flow path statistics. The path with the highest posterior probability is selected to generate a probability filling interval. The probability filling interval is inserted into the corresponding breakpoint position of the semantic trajectory sequence to generate a complete trajectory record. The complete trajectory record is written to persistent storage, and anomaly detection is performed on the complete trajectory record according to the preset dwell threshold conditions and standard flow path rules to generate alarm events.

8. A pallet dynamic tracking device based on the Internet of Things, characterized in that, The device includes: The tray sensing module is used by edge nodes to acquire tray sensing data through radio frequency identification readers and Bluetooth positioning base stations, and aggregate the data according to a preset message queue protocol to obtain a raw sensing data set. The raw sensing data set is then used to calculate confidence level according to signal strength threshold conditions and historical stability measures to obtain confidence-bearing sensing data. The confidence-bearing sensing data with the same tray identifier is then weighted and fused according to timestamp alignment rules to obtain a spatiotemporal feature vector. The motion labeling module is used to classify the state based on the position difference sequence and direction change amplitude of the spatiotemporal feature vector according to the preset motion judgment threshold to obtain motion state labels, determine the sampling window length according to the motion state labels according to the preset window adjustment rules to obtain the sampling window configuration, and aggregate the spatiotemporal feature vectors within the sampling window configuration range according to the weighted mean rule to obtain the perceived event. The trajectory tracking module is used to receive the perceived events in the cloud and arrange them according to the tray identifier and timestamp to obtain the original trajectory sequence. The original trajectory sequence is then mapped to a semantic trajectory sequence according to a preset semantic region boundary definition. The semantic trajectory sequence is then detected according to the breakpoint judgment threshold condition to obtain a trajectory breakpoint set. Based on the trajectory breakpoint set, a probability filling is performed according to the trajectory inference model to obtain a complete trajectory record. The complete trajectory record is then written to persistent storage and an alarm event is generated.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the IoT-based tray dynamic tracking method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the IoT-based tray dynamic tracking method as described in any one of claims 1 to 7.

Images