Intelligent tray positioning and abnormality early warning method and device

Abstract

The embodiment of the application provides a kind of intelligent tray positioning and abnormal early warning method and device, through clock synchronization and confidence calculation, the accurate acquisition of positioning is realized.Construct monitoring mechanism, combine state determination and trajectory generation, establish reliable position tracking strategy.Introduce early warning optimization, through breakpoint detection and abnormal identification, ensure the continuous improvement of management.The method effectively solves the shortcomings of traditional technology in data acquisition, state determination and abnormal detection, etc., and provides technical support for intelligent warehouse management.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method and device for intelligent pallet positioning and anomaly early warning. Background Technology

[0002] Existing pallet positioning and early warning methods have significant shortcomings. Traditional systems perform poorly in data acquisition and clock synchronization, failing to effectively achieve accurate pallet positioning and impacting monitoring results.

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

[0004] Existing systems have technical shortcomings in anomaly detection. The lack of in-depth analysis of flow trajectories makes it difficult to achieve efficient anomaly identification through probabilistic prediction, thus affecting the accuracy of early warnings. Solving these problems is crucial for improving warehouse management capabilities. Summary of the Invention

[0005] To address the problems in existing technologies, this application provides an intelligent pallet positioning and anomaly early warning method and device, which can effectively solve the shortcomings of traditional technologies in data collection, status determination and anomaly detection, and provide technical support for intelligent warehouse management.

[0006] To solve at least one of the above problems, this application provides the following technical solution: Firstly, this application provides a method for intelligent pallet positioning and anomaly early warning, including: The edge gateway deployed in the storage area reads the pallet code and cargo status data of the pallet electronic tag through the radio frequency identification reader and obtains the pallet storage coordinate data through the Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to calculate the confidence level according to the signal strength threshold condition and the consistency deviation of the historical storage location to obtain a set of pallet data with confidence. The confidence-based pallet data set is grouped according to the pallet code and weighted and fused according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence. The pallet spatiotemporal feature vector sequence is then used to determine the flow status according to the difference between adjacent vectors and the change in the direction of movement to obtain a pallet motion status mark. The pallet motion status mark is then matched with a preset window mapping table to obtain a sampling window configuration. The pallet spatiotemporal feature vectors within the range of the sampling window configuration are then aggregated to obtain a pallet positioning event record. The pallet positioning event records are arranged according to the pallet code and timestamp, and the area mapping is performed according to the boundary definition of the warehouse functional area to obtain the pallet semantic trajectory sequence. The path probability prediction model is called on the pallet semantic trajectory sequence to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. The complete pallet flow trajectory is subjected to anomaly detection according to the preset stagnation threshold condition and standard flow path rules to obtain graded early warning events and push them to the warehouse management terminal.

[0007] Furthermore, it also includes: an edge gateway deployed in the storage area sends a radio frequency excitation signal to the pallet electronic tag through a radio frequency identification reader and receives a response signal returned by the pallet electronic tag to obtain a raw radio frequency signal; and performs decoding processing on the raw radio frequency signal according to a preset tag protocol to extract the pallet code and cargo status data to obtain a tag parsing data packet. The edge gateway receives the broadcast signal emitted by the Bluetooth beacon on the pallet carrier through the Bluetooth positioning base station, and performs triangulation calculation according to the signal arrival angle and signal strength to obtain the pallet storage coordinate data. The tag parsing data packet and the pallet storage coordinate data are associated and matched according to the pallet code, and a collection timestamp is added to generate a pallet sensing data stream.

[0008] Furthermore, it also includes: performing deviation calculation on the local acquisition timestamp of each sensing data in the tray sensing data stream according to the standard time base obtained by the network time protocol to obtain a clock deviation value, and performing timestamp correction processing on the clock deviation value according to a preset compensation rule to obtain a time-aligned data stream; The signal strength of each data point in the time-aligned data stream is determined by interval judgment according to a preset signal strength threshold condition to obtain a signal quality score. The historical position coordinate sliding window is extracted for each data point according to the pallet code, and the Euclidean distance between the current position coordinate and the mean coordinate of the historical position coordinate sliding window is calculated to obtain the position consistency deviation value. The signal quality score and the position consistency deviation value are weighted according to a preset weight configuration to obtain a confidence evaluation value, and then encapsulated with the corresponding data to obtain a pallet data set with confidence.

[0009] Furthermore, it also includes: grouping the confidence-based pallet data set according to the pallet code to obtain a pallet data group set; merging multiple data with similar timestamps in each group of the pallet data group set according to a preset timestamp alignment window; and calculating the weighted average of the warehouse coordinates according to the fusion weight after normalization of the confidence evaluation value of each data to obtain a pallet spatiotemporal feature vector sequence. The pallet spatiotemporal feature vector sequence is processed according to timestamp order to calculate the warehouse position coordinate difference between adjacent vectors to obtain a warehouse position difference sequence. The pallet spatiotemporal feature vector sequence is processed according to timestamp order to calculate the angle between the movement directions of adjacent vectors to obtain a direction change amplitude sequence. The warehouse position difference sequence is processed according to a preset static judgment threshold condition to perform low displacement judgment, and the direction change amplitude sequence is processed according to a preset direction change threshold condition to perform direction change judgment to obtain a flow feature judgment result. The flow feature judgment result is processed according to a preset state classification rule to classify the pallet motion state into static state, stable movement state and rapid change state to obtain a pallet motion state label.

[0010] Furthermore, it also includes: performing a query and matching operation between the pallet movement status marker and the status window correspondence in the preset window mapping table to obtain the window length parameter, and performing a window start and end time calculation between the window length parameter and the current timestamp to obtain the sampling window configuration; The pallet location coordinates are calculated by weighting the pallet location coordinates within the sampling window configuration range according to the weights after normalization of the confidence evaluation values ​​of each vector. The pallet location coordinates are then encapsulated with the start and end timestamps configured in the sampling window, the pallet motion status markers, and the pallet code execution fields to obtain a pallet positioning event record.

[0011] Furthermore, it also includes: grouping the pallet positioning event records according to the pallet code to obtain a pallet event group set, and arranging the pallet positioning event records in each group of the pallet event group set in ascending order of timestamps to obtain the original pallet trajectory sequence; The event location coordinates of each pallet positioning event record in the original pallet trajectory sequence are determined according to the preset storage functional area boundary definition to obtain the functional area assignment result. The functional area assignment result is converted into semantic region encoding and a functional area level label is added. Then, the corresponding pallet positioning event record is encapsulated with field extension to obtain the pallet semantic trajectory sequence.

[0012] Furthermore, it also includes: performing interval calculation on the timestamps of adjacent pallet positioning event records in the pallet semantic trajectory sequence to obtain a time interval sequence; performing over-limit detection on the time interval sequence according to a preset breakpoint judgment threshold condition to obtain a trajectory breakpoint set; calling a path probability prediction model to perform candidate path posterior probability calculation on each breakpoint in the trajectory breakpoint set according to the semantic region encoding and spatiotemporal features of the pallet positioning event records before and after the breakpoint and selecting the path with the highest probability to obtain a probability filling interval; and inserting the probability filling interval into the corresponding breakpoint position of the pallet semantic trajectory sequence to obtain the complete pallet flow trajectory. The dwell time of each semantic region in the complete flow trajectory of the pallet is judged according to the preset dwell threshold condition to obtain the dwell anomaly mark. The semantic region encoding transfer order of the complete flow trajectory of the pallet is calculated according to the standard flow path rules to obtain the deviation anomaly mark. The dwell anomaly mark and the deviation anomaly mark are classified into alarm levels according to the preset classification threshold condition to obtain the graded early warning event and pushed to the warehouse management terminal.

[0013] Secondly, this application provides an intelligent pallet positioning and anomaly early warning device, comprising: The pallet sensing module is used by the edge gateway deployed in the storage area to read the pallet code and cargo status data of the pallet electronic tag through the radio frequency identification reader and obtain the pallet storage location coordinate data through the Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to calculate the confidence level according to the signal strength threshold condition and the consistency deviation of the historical storage location to obtain a set of pallet data with confidence. The event location module is used to group the confidence-based pallet data set according to the pallet code and perform weighted fusion according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence. The module then performs flow status determination on the pallet spatiotemporal feature vector sequence according to the difference between adjacent vectors and the magnitude of change in the movement direction to obtain a pallet movement status mark. The module matches the pallet movement status mark with a preset window mapping table to obtain a sampling window configuration. Finally, the module aggregates the pallet spatiotemporal feature vectors within the sampling window configuration range to obtain a pallet location event record. The anomaly warning module is used to arrange the pallet positioning event records according to the pallet code and timestamp, and perform area mapping according to the boundary definition of the storage functional area to obtain the pallet semantic trajectory sequence. The module calls the path probability prediction model to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. The module performs anomaly detection on the complete pallet flow trajectory according to the preset stagnation threshold conditions and standard flow path rules to obtain graded warning events and push them to the storage management terminal.

[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 intelligent tray positioning and anomaly warning 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 intelligent tray positioning and anomaly warning method.

[0016] Fifthly, this application provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the intelligent tray positioning and anomaly warning method.

[0017] As can be seen from the above technical solution, this application provides an intelligent pallet positioning and anomaly early warning method and device, which achieves accurate positioning through clock synchronization and confidence calculation. A monitoring mechanism is constructed, combining state determination and trajectory generation to establish a reliable location tracking strategy. Early warning optimization is introduced, ensuring continuous improvement in management through breakpoint detection and anomaly identification. This method effectively solves the shortcomings of traditional technologies in data acquisition, state determination, and anomaly detection, providing technical support for intelligent warehouse management. 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 intelligent tray positioning and anomaly warning method in the embodiments of this application; Figure 2 This is a structural diagram of the intelligent pallet positioning and anomaly warning device in the embodiments of this application. Detailed Implementation

[0020] 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.

[0021] The acquisition, storage, use, and processing of data in this application comply with relevant laws and regulations.

[0022] To address the problems existing in current technologies, this application provides an intelligent pallet positioning and anomaly early warning method and apparatus. Through clock synchronization and confidence calculation, accurate positioning is achieved. A monitoring mechanism is constructed, combining state determination and trajectory generation to establish a reliable location tracking strategy. Early warning optimization is introduced, ensuring continuous management improvement through breakpoint detection and anomaly identification. This method effectively solves the shortcomings of traditional technologies in data acquisition, state determination, and anomaly detection, providing technical support for intelligent warehouse management.

[0023] To effectively address the shortcomings of traditional technologies in data acquisition, status determination, and anomaly detection, and to provide technical support for intelligent warehouse management, this application provides an embodiment of an intelligent pallet positioning and anomaly early warning method. See [link to embodiment]. Figure 1 The intelligent pallet positioning and anomaly early warning method specifically includes the following: Step S101: The edge gateway deployed in the storage area reads the pallet code and cargo status data of the pallet electronic tag through the radio frequency identification reader and obtains the pallet storage location coordinate data through the Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to calculate the confidence level according to the signal strength threshold condition and the consistency deviation of the historical storage location to obtain a set of pallet data with confidence. The edge gateway deployed in the warehouse area serves as the data access node in this embodiment, sending radio frequency excitation signals to the pallet electronic tags via RFID readers. Upon receiving the excitation signal, the pallet electronic tag returns a response signal carrying the pallet code and cargo status data. The edge gateway decodes this response signal according to a preset tag protocol, extracting the pallet code field and cargo status field and encapsulating them into a tag parsing data packet.

[0024] Simultaneously with the generation of the tag parsing data packet, the edge gateway receives a broadcast signal emitted by a Bluetooth beacon on the pallet carrier via a Bluetooth positioning base station. In this embodiment, triangulation calculations are performed on the broadcast signal based on its arrival angle and signal strength. The angle and strength values ​​observed by multiple Bluetooth positioning base stations are substituted into the geometry solving module to obtain the pallet's location coordinates within the storage space. The edge gateway then performs an association matching between the tag parsing data packet and the pallet location coordinates according to the pallet code. For successfully matched data pairs, a locally acquired timestamp is appended to generate a pallet sensing data stream.

[0025] Accordingly, this embodiment performs clock synchronization calibration processing on the tray sensing data stream. The edge gateway obtains the current standard time reference from the standard time server according to the network time protocol, calculates the deviation between the local acquisition timestamp of each sensing data in the tray sensing data stream and the standard time reference, and obtains the clock deviation value corresponding to each data. The edge gateway performs timestamp correction processing on the clock deviation value according to a preset compensation rule, and fills the corrected timestamp back into the corresponding sensing data to generate a time-aligned data stream.

[0026] After the time-aligned data stream is generated, this embodiment performs confidence calculation on it. The edge gateway extracts the signal strength value of each data point in the time-aligned data stream, compares it with a preset signal strength threshold condition to determine the interval, and assigns a corresponding signal quality score based on the interval level in which the signal strength falls. Simultaneously, the edge gateway extracts the historical warehouse coordinate sliding window of the corresponding pallet according to the pallet code, calculates the Euclidean distance between the current warehouse coordinate and the mean coordinate within the historical warehouse coordinate sliding window, and obtains the warehouse consistency deviation value.

[0027] Based on the aforementioned signal quality score and position consistency deviation value, this embodiment performs a weighted calculation to generate a confidence assessment value according to a preset weight configuration. The formula for the confidence assessment value is as follows: M = u·Q - v·D, Where M represents the confidence assessment value, Q represents the signal quality score, D represents the warehouse consistency deviation value, u represents the signal quality weighting coefficient, and v represents the deviation penalty weighting coefficient. The values ​​of both are pre-configured based on the signal stability characteristics of the warehousing environment. The edge gateway encapsulates the confidence assessment value with corresponding data to generate a confidence-weighted pallet data set. This confidence-weighted pallet data set will be read in subsequent step S201 for grouping and weighted fusion processing according to the pallet code.

[0028] Step S102: Group the confidence-based pallet data set according to the pallet code and perform weighted fusion according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence. Perform flow status determination on the pallet spatiotemporal feature vector sequence according to the difference between adjacent vectors and the change in movement direction to obtain a pallet motion status mark. Match the pallet motion status mark with a preset window mapping table to obtain a sampling window configuration. Aggregate the pallet spatiotemporal feature vectors within the sampling window configuration range to obtain a pallet positioning event record. Based on the confidence-based tray data set generated in step S101 above, this embodiment performs grouping processing on it according to the tray code. The edge gateway traverses each data item in the confidence-based tray data set, extracts the tray code field as the grouping key, and groups data with the same tray code into the same group to generate a tray data group set.

[0029] After the pallet data group set is generated, this embodiment performs weighted fusion on the data within each group according to the timestamp alignment rule. The edge gateway sorts the data within each group according to the corrected timestamp, and identifies multiple data points whose timestamp differences fall within the preset timestamp alignment window as observation records at the same sampling time. For multiple observation records at the same sampling time, the edge gateway extracts the confidence evaluation value of each record and performs normalization processing to obtain the fusion weight. The warehouse coordinates of each record are then weighted and averaged according to the fusion weight to obtain the fused warehouse coordinates at that sampling time. The edge gateway encapsulates the fused warehouse coordinates with the corresponding fused timestamp, pallet code, and fused cargo status to generate a pallet spatiotemporal feature vector. The pallet spatiotemporal feature vectors of all sampling times within each group are arranged in ascending order of timestamp to generate a pallet spatiotemporal feature vector sequence.

[0030] Accordingly, this embodiment performs flow status determination on the pallet spatiotemporal feature vector sequence. The edge gateway traverses adjacent vectors in the pallet spatiotemporal feature vector sequence according to timestamp order, calculates the Euclidean distance between the fused warehouse coordinates of the latter vector and the fused warehouse coordinates of the former vector, and obtains the warehouse difference value. The edge gateway arranges the warehouse difference values ​​of all adjacent vector pairs in chronological order to generate a warehouse difference value sequence. At the same time, the edge gateway extracts the movement direction of adjacent vector pairs and calculates the angle value between the two directions, arranges the angle values ​​of all adjacent vector pairs in chronological order to generate a direction change amplitude sequence.

[0031] After the storage position difference sequence and the direction change amplitude sequence are generated, this embodiment performs flow characteristic determination on them. The edge gateway compares each element in the storage position difference sequence with a preset static judgment threshold condition, and counts cases where the storage position difference is lower than the preset static judgment threshold condition within a continuous number of sampling intervals and marks them as low displacement states. The edge gateway compares each element in the direction change amplitude sequence with a preset direction change threshold condition, and marks a direction change state when the included angle value of an element exceeds the preset direction change threshold condition. The edge gateway classifies the low displacement state and the direction change state according to a preset state classification rule. When a period of time presents a low displacement state and no direction change state occurs, it is classified as a static state. When a period of time presents a storage position difference state and no direction change state occurs, it is classified as a stable movement state. When a period of time presents a direction change state, it is classified as a rapid change state, and a pallet movement state label is generated.

[0032] Based on the aforementioned pallet motion status markers, this embodiment performs a matching query against a preset window mapping table. The preset window mapping table defines the mapping relationship between each motion status category and its corresponding window length parameter. The edge gateway retrieves the window length parameter from the preset window mapping table based on the pallet motion status markers for the current time period, calculates the window start and end times using the window length parameter and the current timestamp, and generates a sampling window configuration.

[0033] After the sampling window is configured and generated, this embodiment performs aggregation processing on the pallet spatiotemporal feature vectors within its range. The edge gateway selects all vectors from the pallet spatiotemporal feature vector sequence whose fused timestamps fall within the start and end time intervals configured in the sampling window. It extracts the confidence evaluation value of each vector and performs normalization processing to obtain the aggregation weight. The fused warehouse coordinates of each vector are calculated by weighting according to the aggregation weight to obtain the event warehouse coordinates. The edge gateway encapsulates the event warehouse coordinates with the start and end timestamps, pallet motion status markers, and pallet code execution fields configured in the sampling window to generate a pallet positioning event record. The pallet positioning event record will be read in subsequent step S103 and used for sorting according to the pallet code and timestamp and performing region mapping processing.

[0034] Step S103: Arrange the pallet positioning event records according to the pallet code and timestamp, and perform area mapping according to the boundary definition of the warehouse functional area to obtain the pallet semantic trajectory sequence. Call the path probability prediction model to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. Perform anomaly detection on the complete pallet flow trajectory according to the preset stagnation threshold condition and standard flow path rules to obtain graded early warning events and push them to the warehouse management terminal.

[0035] Based on the tray location event records generated in step S102 above, this embodiment performs grouping processing according to the tray code. The cloud service obtains the tray location event records uploaded by each edge gateway through the message receiving interface, extracts the tray code field in each record as the grouping key, and groups records with the same tray code into the same group to generate a tray event group set.

[0036] After the tray event group set is generated, this embodiment sorts the tray positioning event records within each group in ascending order of timestamp. The cloud service reads the window start timestamp field of all records in each group and sorts the records in ascending order of the window start timestamp, so that the positioning event records of the same tray are arranged sequentially according to the collection time, generating the original tray trajectory sequence.

[0037] Accordingly, this embodiment performs region mapping processing on the original pallet trajectory sequence. The cloud service maintains a preset warehouse functional area boundary definition, which includes the spatial boundary polygon coordinates of each functional area, a unique functional area code, and functional area hierarchy information. The cloud service traverses each pallet positioning event record in the original pallet trajectory sequence, extracts the event warehouse coordinates of each record, determines the inclusion relationship between the event warehouse coordinates and the boundary polygon execution points and polygons of each functional area in the preset warehouse functional area boundary definition, determines the functional area to which the coordinates belong, and generates a functional area attribution result. The cloud service converts the functional area attribution result into a semantic region code and adds a functional area hierarchy annotation, then expands and encapsulates it with the corresponding pallet positioning event record execution field to generate a semantic pallet trajectory sequence.

[0038] After the semantic trajectory sequence of the tray is generated, this embodiment performs breakpoint detection processing on it. The cloud service traverses each tray positioning event record in the semantic trajectory sequence, arranged in ascending order of timestamp, calculates the difference between the window start timestamps of two adjacent records, and generates a time interval sequence. The cloud service compares each element in the time interval sequence with a preset breakpoint determination threshold condition. When a certain time interval value exceeds the preset breakpoint determination threshold condition, the position is marked as a breakpoint position, and the semantic region encoding and spatiotemporal features of the two records before and after the breakpoint are extracted to generate a trajectory breakpoint set.

[0039] Based on the aforementioned set of trajectory breakpoints, this embodiment invokes a path probability prediction model to perform probability filling processing. For each breakpoint in the trajectory breakpoint set, the cloud service reads the semantic region encoding of the records before and after the breakpoint as start and end constraints for path inference. It retrieves historical records of the same start and end region pairs from historical flow path statistics and extracts the sequence of traversed regions as a candidate path set. The path probability prediction model calculates the posterior probability based on the frequency of each candidate path in the historical records and the distribution of dwell time in each functional area. It selects the candidate path with the highest posterior probability and whose total path duration matches the breakpoint time span as the inference result. Virtual location event records are generated for each functional area traversed in the inferred path according to time allocation rules and encapsulated as probability filling intervals. The cloud service inserts the probability filling intervals into the corresponding breakpoint positions of the tray semantic trajectory sequence to generate a complete tray flow trajectory.

[0040] After the complete pallet flow trajectory is generated, this embodiment performs anomaly detection processing. The cloud service segments the complete pallet flow trajectory according to semantic region encoding, calculates the continuous dwell time of the pallet in each functional area, compares the dwell time with the preset dwell threshold condition of the functional area, and generates a dwell anomaly mark when the dwell time exceeds the dwell threshold condition. At the same time, the cloud service extracts the semantic region encoding transfer order of the complete pallet flow trajectory, matches it with the standard flow path rule execution mode, and calculates the path deviation degree and generates a deviation anomaly mark when there is an unregistered functional area or the transfer order deviates from the standard path.

[0041] Based on the aforementioned delay anomaly markers and deviation anomaly markers, this embodiment performs alarm level classification according to preset classification threshold conditions. The cloud service determines the timeout degree corresponding to the delay anomaly marker and the deviation degree corresponding to the deviation anomaly marker according to the preset classification threshold conditions, classifying them into three alarm levels: Attention, Warning, and Severe, and generating graded early warning events. The graded early warning event includes the pallet code, current event warehouse location coordinates, trigger time, alarm level, anomaly type, and suggested handling action. The cloud service sends the graded early warning event to the warehouse management terminal through a message push interface. The warehouse management terminal displays the location of the abnormal pallet with visual annotations on the electronic map interface, allowing operators to view detailed early warning information and enter handling results.

[0042] As described above, the intelligent pallet positioning and anomaly early warning method provided in this application can achieve accurate positioning through clock synchronization and confidence calculation. A monitoring mechanism is constructed, combining state determination and trajectory generation to establish a reliable location tracking strategy. Early warning optimization is introduced, ensuring continuous improvement in management through breakpoint detection and anomaly identification. This method effectively solves the shortcomings of traditional technologies in data acquisition, state determination, and anomaly detection, providing technical support for intelligent warehouse management.

[0043] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S201: The edge gateway deployed in the storage area sends an RF excitation signal to the pallet electronic tag through an RF reader and receives the response signal returned by the pallet electronic tag to obtain the raw RF signal. The raw RF signal is then decoded according to a preset tag protocol to extract the pallet code and cargo status data to obtain the tag parsing data packet. Step S202: The edge gateway receives the broadcast signal emitted by the Bluetooth beacon on the pallet carrier through the Bluetooth positioning base station and performs triangulation calculation according to the signal arrival angle and signal strength to obtain the pallet storage coordinate data. The tag parsing data packet and the pallet storage coordinate data are associated and matched according to the pallet code and a collection timestamp is added to generate a pallet sensing data stream.

[0044] An edge gateway deployed in the warehouse area serves as the data access node in this embodiment. It continuously sends radio frequency (RF) excitation signals to the pallet electronic tags within its coverage area via an RF reader. The RF excitation signal employs a preset carrier frequency and modulation scheme to form a stable RF coverage field within the warehouse space. Upon entering this RF coverage field, the pallet electronic tags are awakened by the excitation signal, transitioning from a dormant state to a response state and returning a response signal carrying the tag's stored data to the RF reader. The edge gateway receives the response signal through the RF reader's antenna array, obtaining the raw RF signal.

[0045] After acquiring the raw radio frequency signal, this embodiment performs decoding processing on it according to a preset tag protocol. The edge gateway performs baseband demodulation and frame synchronization detection on the raw radio frequency signal, identifying the preamble and frame boundaries in the response signal. The edge gateway performs field segmentation on the intra-frame bitstream according to the data frame structure defined by the preset tag protocol, extracting the pallet code field and the cargo status field. The pallet code field contains the pallet's unique identifier, and the cargo status field contains whether the pallet is currently carrying goods and the goods category code. The edge gateway performs checksum verification on the extracted field data, and after successful verification, encapsulates the pallet code and cargo status data into a tag parsing data packet.

[0046] Accordingly, this embodiment obtains the spatial location information of the pallet through Bluetooth positioning base stations. Multiple Bluetooth positioning base stations deployed in the storage area continuously monitor the broadcast signals emitted by Bluetooth beacons on the pallet carrier. Each Bluetooth positioning base station independently measures the angle of arrival and signal strength of the received broadcast signal and adds its own base station identifier. The edge gateway collects the measurement data reported by multiple Bluetooth positioning base stations and aggregates the measurement results collected from multiple base stations for the same Bluetooth beacon within a configurable time window.

[0047] After aggregating the measurement results from multiple base stations, this embodiment performs triangulation calculations. The edge gateway reads the pre-calibrated spatial coordinates of each Bluetooth positioning base station and converts the signal arrival angles measured by each base station into directional rays emanating from the base station's location. The edge gateway performs spatial intersection calculations on multiple directional rays and uses the geometric center of the intersection area as a preliminary estimate of the pallet location coordinates. The edge gateway further incorporates the signal strength measured by each base station as a distance constraint to correct the preliminary estimate, obtaining the pallet location coordinate data.

[0048] Based on the aforementioned tag parsing data packet and pallet location coordinate data, this embodiment performs an association matching process. The edge gateway extracts the pallet code from the tag parsing data packet and the beacon-bound pallet code from the Bluetooth beacon's broadcast signal, and performs a consistency comparison between the two pallet codes. When the two pallet codes match successfully, the edge gateway merges the fields of the tag parsing data packet and the pallet location coordinate data, appends a collection timestamp generated by the edge gateway's local clock, and generates a pallet sensing data stream. The pallet sensing data stream will be read in subsequent step S301 for performing clock synchronization calibration processing according to the network time protocol.

[0049] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S301: Calculate the clock deviation value by performing deviation calculation on the local acquisition timestamp of each sensing data in the tray sensing data stream according to the standard time base obtained by the network time protocol, and perform timestamp correction processing on the clock deviation value according to the preset compensation rule to obtain a time-aligned data stream; Step S302: Perform interval determination on the signal strength of each data in the time-aligned data stream according to the preset signal strength threshold condition to obtain a signal quality score. Extract the historical position coordinate sliding window for each data according to the pallet code and calculate the Euclidean distance between the current position coordinate and the mean of the coordinates in the historical position coordinate sliding window to obtain the position consistency deviation value. Perform weighted calculation on the signal quality score and the position consistency deviation value according to the preset weight configuration to obtain the confidence evaluation value and encapsulate it with the corresponding data to obtain a pallet data set with confidence.

[0050] Based on the tray sensing data stream generated in step S202 above, this embodiment performs clock synchronization calibration processing on it. The edge gateway periodically interacts with the standard time server according to the network time protocol to obtain the current standard time reference and maintain a record of the deviation between the local clock and the standard time. The edge gateway traverses each sensing data in the tray sensing data stream, extracts the local acquisition timestamp field of each data, calculates the difference between it and the standard time reference, and obtains the clock deviation value corresponding to each data.

[0051] After the clock deviation value is calculated, this embodiment performs timestamp correction processing on it according to a preset compensation rule. The preset compensation rule defines the compensation direction and the calculation method for the compensation amount of the clock deviation value. The edge gateway determines the compensation direction based on the positive or negative sign of the clock deviation value, and adds or subtracts the corresponding deviation amount from the locally acquired timestamp to obtain the corrected timestamp. The edge gateway fills the corrected timestamp back into the timestamp field of the corresponding sensed data, generating a time-aligned data stream. The timestamps of each data item in the time-aligned data stream have been calibrated to a unified standard time reference.

[0052] Accordingly, this embodiment performs signal quality assessment on the time-aligned data stream. The edge gateway traverses each data point in the time-aligned data stream and extracts the signal strength value recorded during Bluetooth positioning for each data point. The edge gateway performs interval determination based on the signal strength value and a preset signal strength threshold condition. The preset signal strength threshold condition is divided into multiple strength intervals, and a corresponding quality level is configured for each interval. The edge gateway assigns a corresponding signal quality score based on the interval level in which the signal strength value falls; the higher the signal strength, the higher the signal quality score.

[0053] After the signal quality score is generated, this embodiment performs a warehouse consistency deviation calculation on each data entry. The edge gateway groups each data entry in the time-aligned data stream according to the warehouse code, and maintains a historical warehouse coordinate sliding window for each warehouse code. The historical warehouse coordinate sliding window records the warehouse coordinate sequence of that warehouse in several recent sampling periods. The edge gateway extracts the warehouse coordinates of the current data, calculates the Euclidean distance between the current data and the arithmetic mean of all coordinates in the historical warehouse coordinate sliding window, and obtains the warehouse consistency deviation value. The warehouse consistency deviation value reflects the degree of deviation between the current observation position and the recent position distribution center of that warehouse.

[0054] Based on the aforementioned signal quality score and position consistency deviation value, this embodiment performs a weighted calculation to generate a confidence assessment value according to a preset weight configuration. The formula for the confidence assessment value is as follows: R = p·S - q·E, Where R represents the confidence assessment value, S represents the signal quality score, E represents the warehouse consistency deviation value, p represents the signal quality weighting coefficient, and q represents the deviation penalty weighting coefficient. The values ​​of both are pre-configured based on the signal propagation characteristics and pallet turnover speed characteristics of the warehousing environment. The edge gateway encapsulates the confidence assessment value and corresponding data into fields to generate a confidence-weighted pallet data set. This confidence-weighted pallet data set will be read in subsequent step S401 for grouping and weighted fusion processing according to the pallet code.

[0055] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S401: The confidence-based pallet data set is grouped according to the pallet code to obtain a pallet data group set. Multiple data with similar timestamps in each group of the pallet data group set are merged according to a preset timestamp alignment window. The weighted average of the warehouse coordinates is calculated according to the fusion weight after normalization of the confidence evaluation value of each data to obtain the pallet spatiotemporal feature vector sequence. Step S402: Calculate the warehouse position difference sequence by performing the warehouse position coordinate difference calculation on the warehouse position feature vector sequence according to the timestamp order of adjacent vectors; calculate the direction change amplitude sequence by performing the angle between the movement directions of adjacent vectors on the warehouse position feature vector sequence according to the timestamp order; perform low displacement judgment on the warehouse position difference sequence according to the preset static judgment threshold condition; and perform direction change abrupt judgment on the direction change amplitude sequence according to the preset direction change threshold condition to obtain the flow feature judgment result; and classify the flow feature judgment result into static state, stable movement state, and rapid change state according to the preset state classification rules to obtain the warehouse movement state label.

[0056] Based on the confidence-based tray data set generated in step S302 above, this embodiment performs grouping processing on it according to the tray code. The edge gateway traverses each data record in the confidence-based tray data set, extracts the tray code field as the grouping key, and groups data with the same tray code into the same group, generating a tray data group set. Each group in the tray data group set corresponds to all observation records of a tray in the current processing cycle.

[0057] After the tray data group set is generated, this embodiment performs merging processing on the data within each group according to a preset timestamp alignment window. The edge gateway sorts the data within each group in ascending order according to the corrected timestamps, and uses the length of the preset timestamp alignment window as the criterion to identify multiple data points whose timestamp differences fall within the preset timestamp alignment window as concurrent observation records at the same sampling time. For the multiple concurrent observation records identified at the same sampling time, the edge gateway merges them into a data group to be fused.

[0058] Accordingly, this embodiment performs weighted mean calculation on each data group to be fused. The edge gateway extracts the confidence assessment value of each data point within the data group to be fused, calculates the sum of all confidence assessment values ​​within the group as the normalization base, and divides the confidence assessment value of each data point by the normalization base to obtain the fusion weight. The edge gateway extracts the warehouse location coordinates of each data point, multiplies each coordinate value by the corresponding fusion weight, and then sums them to obtain the fused warehouse location coordinates at that sampling time. The edge gateway encapsulates the fused warehouse location coordinates with the corresponding fusion timestamp, pallet code, fused cargo status, and average confidence score within the group to generate a pallet spatiotemporal feature vector. The edge gateway arranges the pallet spatiotemporal feature vectors of all sampling times within each group in ascending order of timestamp to generate a pallet spatiotemporal feature vector sequence.

[0059] After the pallet spatiotemporal feature vector sequence is generated, this embodiment performs a warehouse location difference calculation on it. The edge gateway traverses adjacent vectors in the pallet spatiotemporal feature vector sequence according to timestamp order, extracts the fused warehouse location coordinates of the latter vector and the former vector, and calculates the Euclidean distance between the two as the warehouse location difference within that time interval. The edge gateway arranges the warehouse location differences of all adjacent vector pairs in chronological order to generate a warehouse location difference sequence.

[0060] Based on the aforementioned pallet spatiotemporal feature vector sequence, this embodiment synchronously performs the calculation of the moving direction angle. The edge gateway traverses two adjacent vectors in timestamp order, calculates the moving direction vector at two sampling times based on the fused warehouse coordinates of the preceding and following vectors, treats the two direction vectors as unit vectors on a two-dimensional plane, and calculates their angle value. The edge gateway arranges the angle values ​​of all adjacent vector pairs in chronological order to generate a sequence of direction change amplitudes.

[0061] Accordingly, this embodiment performs flow feature determination on the position difference sequence and the direction change amplitude sequence. The edge gateway traverses each element in the position difference sequence, compares it with a preset static determination threshold, and marks the time periods in which the position difference is lower than the preset static determination threshold within a certain number of consecutive sampling intervals as low displacement states. The edge gateway traverses each element in the direction change amplitude sequence, compares it with a preset direction change threshold, and marks the direction change state when the included angle value of an element exceeds the preset direction change threshold, generating a flow feature determination result.

[0062] After the flow feature determination result is generated, this embodiment performs category division according to preset state classification rules. The edge gateway classifies the flow feature determination result based on the combination of low displacement state and direction change state in each time period. When a certain time period shows a low displacement state and no direction change state occurs, it is classified as a stationary state. When the warehouse position difference in a certain time period exceeds the preset stationary determination threshold and no direction change state occurs, it is classified as a stable movement state. When a direction change state occurs in a certain time period, regardless of the size of the warehouse position difference, it is classified as a rapid change state, generating a pallet movement state marker. The pallet movement state marker will be read in the subsequent step S501 and used to perform a matching query with a preset window mapping table.

[0063] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S501: Perform a query and match between the pallet movement status marker and the status window correspondence in the preset window mapping table to obtain the window length parameter, and perform a window start and end time calculation between the window length parameter and the current timestamp to obtain the sampling window configuration; Step S502: The pallet location coordinates of the pallet spatiotemporal feature vectors within the sampling window configuration range are calculated by weighted average according to the weights after normalization of the confidence evaluation values ​​of each vector. The event location coordinates are then encapsulated with the start and end timestamps configured in the sampling window, the pallet motion status marker, and the pallet code execution field to obtain a pallet positioning event record.

[0064] Based on the pallet motion state markers generated in step S402, this embodiment performs a query and matching operation with a preset window mapping table. The preset window mapping table predefines the mapping relationship between each motion state category and its corresponding window length parameter. A longer window length is configured for stationary states to reduce the event generation frequency during stationary periods; a medium window length is configured for smooth movement states to balance positioning accuracy and event density; and a shorter window length is configured for rapidly changing states to improve the granularity of capturing key turning points. The edge gateway reads the pallet motion state markers for the current time period and retrieves the window length parameter corresponding to that state category from the preset window mapping table.

[0065] After obtaining the window length parameter, this embodiment calculates the window start and end times using the current timestamp. The edge gateway reads the current timestamp of the local system as the window calculation reference point, backtracks half of the window length parameter to obtain the window start timestamp, and extends half of the window length parameter backward to obtain the window end timestamp. The edge gateway structurally encapsulates the window start timestamp, window end timestamp, and window length parameter to generate a sampling window configuration. The sampling window configuration clearly defines the time range covered by this event aggregation.

[0066] Accordingly, this embodiment performs filtering processing on the tray spatiotemporal feature vectors within the sampling window configuration range. The edge gateway traverses each vector from the tray spatiotemporal feature vector sequence generated in the aforementioned step S401, extracts the fusion timestamp field of each vector, and determines whether it falls within the interval defined by the start and end timestamps configured in the sampling window. The edge gateway extracts all vectors whose fusion timestamps fall within the interval to form a vector set within the window.

[0067] After the vector set within the window is constructed, this embodiment performs a weighted average calculation to generate event warehouse coordinates. The edge gateway extracts the confidence evaluation value of each vector in the vector set within the window, calculates the sum of all confidence evaluation values ​​in the set as the normalization base, and divides the confidence evaluation value of each vector by the normalization base to obtain the aggregation weight. The edge gateway extracts the fused warehouse coordinates of each vector, multiplies each coordinate value by the corresponding aggregation weight, and then sums them to obtain the event warehouse coordinates. The event warehouse coordinates reflect the comprehensive spatial position of the pallet within the sampling window period.

[0068] Based on the aforementioned event location coordinates, this embodiment encapsulates them and related attribute fields. The edge gateway reads the window start timestamp and window end timestamp from the sampling window configuration, reads the pallet movement status marker for the current time period from step S402, and extracts the pallet code field from the vector set within the window. The edge gateway encapsulates the event location coordinates, window start timestamp, window end timestamp, pallet movement status marker, pallet code, and number of vectors within the window into a field to generate a pallet positioning event record. The pallet positioning event record will be uploaded to the cloud in subsequent step S601 for sorting and region mapping according to the pallet code and timestamp.

[0069] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S601: Group the pallet positioning event records according to the pallet code to obtain a pallet event group set, and sort the pallet positioning event records in each group of the pallet event group set in ascending order of timestamp to obtain the original pallet trajectory sequence; Step S602: Perform coordinate attribution determination on the event warehouse coordinates of each pallet positioning event record in the original pallet trajectory sequence according to the preset warehouse functional area boundary definition to obtain the functional area attribution result. Convert the functional area attribution result into semantic region encoding and add functional area level label. Then, perform field extension encapsulation with the corresponding pallet positioning event record to obtain the pallet semantic trajectory sequence.

[0070] Based on the tray location event records generated in step S502 above, this embodiment performs grouping processing according to the tray code. The cloud service obtains the tray location event records uploaded by each edge gateway through the message receiving interface, traverses all received records, and extracts the tray code field from each record. The cloud service groups tray location event records with the same tray code into the same group. Each group corresponds to all location events generated by a tray in different time periods and within the coverage area of ​​different edge gateways, generating a tray event group set.

[0071] After the tray event group set is generated, this embodiment performs timestamp sorting on the tray positioning event records within each group. The cloud service traverses each group in the tray event group set and reads the window start timestamp field of all tray positioning event records within the group. The cloud service sorts the tray positioning event records within each group in ascending order according to the window start timestamp, so that the positioning events of the same tray are arranged sequentially according to the collection time, generating the original tray trajectory sequence. The original tray trajectory sequence reflects the temporal positioning records of the tray from its first collection to its most recent collection.

[0072] Accordingly, this embodiment performs coordinate attribution determination processing on the original trajectory sequence of the pallet. The cloud service maintains preset warehouse functional area boundary definitions, which include the coordinate sequence of the vertices of the spatial boundary polygons of each functional area, the unique code of the functional area, and the functional area hierarchy information. The functional areas are divided according to warehousing business scenarios into inbound temporary storage area, shelf storage area, picking operation area, verification and packaging area, and outbound loading area. There are hierarchical relationships of inclusion or adjacency between these functional areas.

[0073] After the preset warehouse functional area boundary definition is loaded, this embodiment performs a line-by-line judgment on each pallet positioning event record in the original pallet trajectory sequence. The cloud service traverses each pallet positioning event record in the original pallet trajectory sequence and extracts the event warehouse location coordinates of each record. The cloud service performs a point-to-polygon inclusion relationship judgment between the event warehouse location coordinates and the boundary polygons of each functional area in the preset warehouse functional area boundary definition to determine the functional area to which the coordinates belong and generates a functional area assignment result. When an event warehouse location coordinate falls into multiple functional areas with inclusion relationships, the cloud service selects the deepest functional area as the final assignment result.

[0074] Based on the aforementioned functional area attribution results, this embodiment performs encoding conversion and field expansion processing. The cloud service converts the functional area names in the attribution results into semantic region codes according to preset encoding rules, reads the hierarchical information of the functional area from the preset warehouse functional area boundary definition, and appends it as a functional area hierarchical label. The cloud service encapsulates the semantic region code and functional area hierarchical label as new fields with the corresponding pallet positioning event records to generate semantic positioning event records. After replacing all pallet positioning event records in the original pallet trajectory sequence with the corresponding semantic positioning event records, the cloud service generates a pallet semantic trajectory sequence. The pallet semantic trajectory sequence will be read in subsequent step S701 for breakpoint detection and path probability filling processing.

[0075] In one embodiment of the intelligent tray positioning and anomaly warning method of this application, it may further include the following: Step S701: Perform interval calculation on the timestamps of adjacent pallet positioning event records in the pallet semantic trajectory sequence to obtain a time interval sequence. Perform over-limit detection on the time interval sequence according to the preset breakpoint judgment threshold to obtain a trajectory breakpoint set. For each breakpoint in the trajectory breakpoint set, call the path probability prediction model to perform candidate path posterior probability calculation according to the semantic region encoding and spatiotemporal features of the pallet positioning event records before and after the breakpoint, and select the path with the highest probability to obtain a probability filling interval. Insert the probability filling interval into the corresponding breakpoint position of the pallet semantic trajectory sequence to obtain the complete pallet flow trajectory. Step S702: The dwell time of each semantic region in the complete flow trajectory of the pallet is judged according to the preset dwell threshold condition to obtain the dwell anomaly mark. The semantic region encoding transfer order of the complete flow trajectory of the pallet is calculated according to the standard flow path rules to obtain the deviation anomaly mark. The dwell anomaly mark and the deviation anomaly mark are classified into alarm levels according to the preset classification threshold condition to obtain the graded early warning event and pushed to the warehouse management terminal.

[0076] Based on the tray semantic trajectory sequence generated in step S602 above, this embodiment calculates the time interval for adjacent tray positioning event records. The cloud service traverses each tray positioning event record in the tray semantic trajectory sequence, arranged in ascending order of timestamps, and sequentially reads two adjacent records and extracts their respective window start timestamps. The cloud service calculates the difference between the window start timestamp of the later record and the window end timestamp of the previous record, and uses this difference as the time interval for that position. The cloud service arranges the time intervals of all adjacent record pairs in sequence order to generate a time interval sequence.

[0077] After the time interval sequence is generated, this embodiment performs over-limit detection on it according to a preset breakpoint determination threshold. The cloud service traverses each element in the time interval sequence and compares each time interval value with the preset breakpoint determination threshold. When a time interval value exceeds the preset breakpoint determination threshold, the cloud service marks that position as a breakpoint. For each breakpoint, the cloud service extracts the semantic region code, event warehouse coordinates, and timestamp of the previous record from the pallet semantic trajectory sequence, and simultaneously extracts the semantic region code, event warehouse coordinates, and timestamp of the next record, generating a trajectory breakpoint set.

[0078] Accordingly, this embodiment calls the path probability prediction model to perform probability filling for each breakpoint in the trajectory breakpoint set. For each breakpoint, the cloud service reads the semantic region codes of the records before and after the breakpoint as start and end constraints for path inference, and calculates the difference in timestamps between the records before and after the breakpoint to obtain the breakpoint time span. The cloud service retrieves the historical flow records corresponding to the tray code from the historical flow path statistics database, filters historical path segments with the same start and end semantic region codes, and extracts the intermediate semantic region sequences traversed by each segment as a candidate path set.

[0079] After the candidate path set is constructed, this embodiment calls the path probability prediction model to perform posterior probability calculation. The path probability prediction model calculates the prior probability based on the frequency of each candidate path's occurrence in historical records, and calculates the likelihood probability based on the matching degree between the typical dwell time distribution of each intermediate semantic region and the breakpoint time span. The prior probability and the likelihood probability are multiplied to obtain the posterior probability of each candidate path. The cloud service selects the candidate path with the highest posterior probability and whose total path duration matches the breakpoint time span as the inference result. Virtual tray positioning event records are generated for each intermediate semantic region traversed in the inferred path according to time allocation rules. An inference tag and inference confidence are added to each virtual record, and then it is encapsulated as a probability-filled interval. The cloud service inserts the probability-filled interval into the corresponding breakpoint position of the tray semantic trajectory sequence to generate the complete tray flow trajectory.

[0080] Based on the aforementioned complete pallet flow trajectory, this embodiment performs a dwell anomaly detection. The cloud service segments the complete pallet flow trajectory according to semantic region encoding, identifies the time periods in which the pallet continuously exists within each semantic region, and calculates the dwell time. The cloud service reads the upper limit of the normal dwell time corresponding to each semantic region from the preset dwell threshold condition configuration, and compares the dwell time with the corresponding upper limit of the normal dwell time. When the dwell time in a certain semantic region exceeds the upper limit of the normal dwell time, the cloud service generates a dwell anomaly marker and records the timeout degree.

[0081] Simultaneously with the generation of the delay anomaly marker, this embodiment performs path deviation detection on the complete pallet flow trajectory. The cloud service extracts the semantic region encoding transfer order of the complete pallet flow trajectory, forming a region transfer sequence. The cloud service performs pattern matching between the region transfer sequence and standard flow path rules, which define the semantic region access order that various pallets should follow in normal business processes. The cloud service detects whether there are unregistered semantic regions or transfer jumps that do not conform to the standard order in the region transfer sequence, calculates the deviation degree, and generates a deviation anomaly marker.

[0082] Based on the aforementioned delay anomaly markers and deviation anomaly markers, this embodiment performs alarm level classification according to preset grading threshold conditions. The cloud service comprehensively evaluates the timeout degree corresponding to the delay anomaly marker and the deviation degree corresponding to the deviation anomaly marker, and classifies them into three alarm levels—attention, warning, and severe—according to preset grading threshold conditions, generating graded early warning events. The graded early warning event includes the pallet code, current event warehouse location coordinates, trigger time, alarm level, anomaly type, and suggested handling action. The cloud service sends the graded early warning event to the warehouse management terminal through a message push interface. The warehouse management terminal displays the location of the abnormal pallet with visual annotations on the electronic map interface, allowing operators to view detailed early warning information and enter handling results.

[0083] To effectively address the shortcomings of traditional technologies in data acquisition, status determination, and anomaly detection, and to provide technical support for intelligent warehouse management, this application provides an embodiment of an intelligent pallet positioning and anomaly early warning device for implementing all or part of the aforementioned intelligent pallet positioning and anomaly early warning method. See [link to embodiment]. Figure 2 The intelligent pallet positioning and anomaly early warning device specifically includes the following components: Pallet sensing module 10 is used to deploy an edge gateway in the storage area to read the pallet code and cargo status data of the pallet electronic tag through an RFID reader and obtain the pallet location coordinate data through a Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to perform confidence calculation according to the signal strength threshold condition and the historical location consistency deviation to obtain a set of pallet data with confidence. Event location module 20 is used to group the confidence-based pallet data set according to the pallet code and perform weighted fusion according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence; to perform flow status determination on the pallet spatiotemporal feature vector sequence according to the difference between adjacent vectors and the change in movement direction to obtain a pallet movement status mark; to perform matching on the pallet movement status mark with a preset window mapping table to obtain a sampling window configuration; and to perform aggregation on the pallet spatiotemporal feature vectors within the sampling window configuration range to obtain a pallet location event record. The anomaly warning module 30 is used to arrange the pallet positioning event records according to the pallet code and timestamp, and perform area mapping according to the boundary definition of the warehouse functional area to obtain the pallet semantic trajectory sequence. The module calls the path probability prediction model to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. The module performs anomaly detection on the complete pallet flow trajectory according to the preset stagnation threshold conditions and standard flow path rules to obtain graded warning events and push them to the warehouse management terminal.

[0084] As described above, the intelligent pallet positioning and anomaly early warning device provided in this application embodiment can achieve accurate positioning through clock synchronization and confidence calculation. A monitoring mechanism is constructed, combining state determination and trajectory generation to establish a reliable location tracking strategy. Early warning optimization is introduced, ensuring continuous improvement in management through breakpoint detection and anomaly identification. This method effectively solves the shortcomings of traditional technologies in data acquisition, state determination, and anomaly detection, providing technical support for intelligent warehouse management.

[0085] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the intelligent tray positioning and anomaly warning method.

[0086] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described intelligent tray positioning and anomaly warning method.

[0087] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described intelligent tray positioning and anomaly warning method.

[0088] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, 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.

[0089] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), 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.

[0090] 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.

[0091] 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.

[0092] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent pallet positioning and anomaly early warning, characterized in that, The method includes: The edge gateway deployed in the storage area reads the pallet code and cargo status data of the pallet electronic tag through the radio frequency identification reader and obtains the pallet storage coordinate data through the Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to calculate the confidence level according to the signal strength threshold condition and the consistency deviation of the historical storage location to obtain a set of pallet data with confidence. The confidence-based pallet data set is grouped according to the pallet code and weighted and fused according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence. The pallet spatiotemporal feature vector sequence is then used to determine the flow status according to the difference between adjacent vectors and the change in the direction of movement to obtain a pallet motion status mark. The pallet motion status mark is then matched with a preset window mapping table to obtain a sampling window configuration. The pallet spatiotemporal feature vectors within the range of the sampling window configuration are then aggregated to obtain a pallet positioning event record. The pallet positioning event records are arranged according to the pallet code and timestamp, and the area mapping is performed according to the boundary definition of the warehouse functional area to obtain the pallet semantic trajectory sequence. The path probability prediction model is called on the pallet semantic trajectory sequence to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. The complete pallet flow trajectory is subjected to anomaly detection according to the preset stagnation threshold condition and standard flow path rules to obtain graded early warning events and push them to the warehouse management terminal.

2. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The edge gateway deployed in the storage area reads the pallet code and cargo status data from the pallet electronic tag using an RFID reader and obtains the pallet location coordinate data through a Bluetooth positioning base station to generate a pallet sensing data stream, including: The edge gateway deployed in the storage area sends radio frequency excitation signals to the pallet electronic tags through radio frequency identification readers and receives the response signals returned by the pallet electronic tags to obtain the raw radio frequency signals. The raw radio frequency signals are then decoded according to a preset tag protocol to extract the pallet code and cargo status data to obtain the tag parsing data packet. The edge gateway receives the broadcast signal emitted by the Bluetooth beacon on the pallet carrier through the Bluetooth positioning base station, and performs triangulation calculation according to the signal arrival angle and signal strength to obtain the pallet storage coordinate data. The tag parsing data packet and the pallet storage coordinate data are associated and matched according to the pallet code, and a collection timestamp is added to generate a pallet sensing data stream.

3. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The process involves performing clock synchronization calibration on the pallet sensing data stream according to the network time protocol to obtain a time-aligned data stream. Then, confidence calculation is performed on the time-aligned data stream based on signal strength threshold conditions and historical warehouse position consistency deviations to obtain a confidence-weighted pallet data set, including: The local acquisition timestamps of each sensing data in the tray sensing data stream are used to calculate the clock deviation value according to the standard time base obtained by the network time protocol. The clock deviation value is then used to perform timestamp correction processing according to the preset compensation rule to obtain a time-aligned data stream. The signal strength of each data point in the time-aligned data stream is determined by interval judgment according to a preset signal strength threshold condition to obtain a signal quality score. The historical position coordinate sliding window is extracted for each data point according to the pallet code, and the Euclidean distance between the current position coordinate and the mean coordinate of the historical position coordinate sliding window is calculated to obtain the position consistency deviation value. The signal quality score and the position consistency deviation value are weighted according to a preset weight configuration to obtain a confidence evaluation value, and then encapsulated with the corresponding data to obtain a pallet data set with confidence.

4. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The process involves grouping the confidence-based pallet data set according to pallet codes and performing weighted fusion according to timestamp alignment rules to obtain a pallet spatiotemporal feature vector sequence. Then, the process involves determining the pallet movement status by analyzing the difference between adjacent pallet positions and the magnitude of changes in the movement direction, including: The confidence-based pallet data set is grouped according to the pallet code to obtain a pallet data group set. Multiple data with similar timestamps in each group of the pallet data group set are merged according to a preset timestamp alignment window. The weighted average of the warehouse coordinates is calculated according to the fusion weight after normalization of the confidence evaluation value of each data to obtain the pallet spatiotemporal feature vector sequence. The pallet spatiotemporal feature vector sequence is processed according to timestamp order to calculate the warehouse position coordinate difference between adjacent vectors to obtain a warehouse position difference sequence. The pallet spatiotemporal feature vector sequence is processed according to timestamp order to calculate the angle between the movement directions of adjacent vectors to obtain a direction change amplitude sequence. The warehouse position difference sequence is processed according to a preset static judgment threshold condition to perform low displacement judgment, and the direction change amplitude sequence is processed according to a preset direction change threshold condition to perform direction change judgment to obtain a flow feature judgment result. The flow feature judgment result is processed according to a preset state classification rule to classify the pallet motion state into static state, stable movement state and rapid change state to obtain a pallet motion state label.

5. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The step of matching the pallet motion state marker with a preset window mapping table to obtain a sampling window configuration, and then aggregating the pallet spatiotemporal feature vectors within the sampling window configuration range to obtain pallet positioning event records, includes: The window length parameter is obtained by querying and matching the correspondence between the pallet movement status marker and the status window in the preset window mapping table. The sampling window configuration is obtained by calculating the window start and end time with the window length parameter and the current timestamp. The pallet location coordinates are calculated by weighting the pallet location coordinates within the sampling window configuration range according to the weights after normalization of the confidence evaluation values ​​of each vector. The pallet location coordinates are then encapsulated with the start and end timestamps configured in the sampling window, the pallet motion status markers, and the pallet code execution fields to obtain a pallet positioning event record.

6. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The process of arranging the pallet positioning event records according to the pallet code and timestamp, and mapping the resulting area according to the warehouse functional area boundary definition, yields a semantic trajectory sequence for the pallet, including: The pallet positioning event records are grouped according to the pallet code to obtain a pallet event group set. The pallet positioning event records in each group of the pallet event group set are sorted in ascending order of timestamp to obtain the original pallet trajectory sequence. The event location coordinates of each pallet positioning event record in the original pallet trajectory sequence are determined according to the preset storage functional area boundary definition to obtain the functional area assignment result. The functional area assignment result is converted into semantic region encoding and a functional area level label is added. Then, the corresponding pallet positioning event record is encapsulated with field extension to obtain the pallet semantic trajectory sequence.

7. The intelligent pallet positioning and anomaly early warning method according to claim 1, characterized in that, The process of calling the path probability prediction model to perform breakpoint detection and probability filling on the semantic trajectory sequence of the pallet to obtain the complete pallet flow trajectory, and performing anomaly detection on the complete pallet flow trajectory according to the preset delay threshold conditions and standard flow path rules to obtain graded early warning events and push them to the warehouse management terminal includes: The time interval sequence is obtained by performing interval calculation on the timestamps of adjacent pallet positioning event records in the pallet semantic trajectory sequence. The time interval sequence is then subjected to over-limit detection according to the preset breakpoint judgment threshold to obtain a trajectory breakpoint set. For each breakpoint in the trajectory breakpoint set, the path probability prediction model is called to perform candidate path posterior probability calculation according to the semantic region encoding and spatiotemporal features of the pallet positioning event records before and after the breakpoint, and the path with the highest probability is selected to obtain a probability filling interval. The probability filling interval is then inserted into the corresponding breakpoint position of the pallet semantic trajectory sequence to obtain the complete pallet flow trajectory. The dwell time of each semantic region in the complete flow trajectory of the pallet is judged according to the preset dwell threshold condition to obtain the dwell anomaly mark. The semantic region encoding transfer order of the complete flow trajectory of the pallet is calculated according to the standard flow path rules to obtain the deviation anomaly mark. The dwell anomaly mark and the deviation anomaly mark are classified into alarm levels according to the preset classification threshold condition to obtain the graded early warning event and pushed to the warehouse management terminal.

8. A smart pallet positioning and anomaly early warning device, characterized in that, The device includes: The pallet sensing module is used by the edge gateway deployed in the storage area to read the pallet code and cargo status data of the pallet electronic tag through the radio frequency identification reader and obtain the pallet storage location coordinate data through the Bluetooth positioning base station to generate a pallet sensing data stream. The pallet sensing data stream is clocked and calibrated according to the network time protocol to obtain a time-aligned data stream. The time-aligned data stream is then used to calculate the confidence level according to the signal strength threshold condition and the consistency deviation of the historical storage location to obtain a set of pallet data with confidence. The event location module is used to group the confidence-based pallet data set according to the pallet code and perform weighted fusion according to the timestamp alignment rule to obtain a pallet spatiotemporal feature vector sequence. The module then performs flow status determination on the pallet spatiotemporal feature vector sequence according to the difference between adjacent vectors and the magnitude of change in the movement direction to obtain a pallet movement status mark. The module matches the pallet movement status mark with a preset window mapping table to obtain a sampling window configuration. Finally, the module aggregates the pallet spatiotemporal feature vectors within the sampling window configuration range to obtain a pallet location event record. The anomaly warning module is used to arrange the pallet positioning event records according to the pallet code and timestamp, and perform area mapping according to the boundary definition of the storage functional area to obtain the pallet semantic trajectory sequence. The module calls the path probability prediction model to perform breakpoint detection and probability filling to obtain the complete pallet flow trajectory. The module performs anomaly detection on the complete pallet flow trajectory according to the preset stagnation threshold conditions and standard flow path rules to obtain graded warning events and push them to the storage management terminal.

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 intelligent pallet positioning and anomaly warning 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 intelligent tray positioning and anomaly warning method according to any one of claims 1 to 7.

Images