Warehouse goods automatic tracking method and system based on RFID technology

Abstract

The invention relates to the technical field of warehouse logistics informatization, in particular to a warehouse goods automatic tracking method and system based on the RFID technology, and the method comprises the steps: obtaining radio frequency identification observation data and reader-writer load data, carrying out the identification analysis of a tag identifier, obtaining a goods unit identifier, obtaining the goods attribute, and forming a region identifier and a timestamp sequence; based on the sequence, extracting the circulation speed and the staying duration as circulation characteristics, acquiring an abnormal threshold and a lagging threshold corresponding to the cargo attribute from a parameter library, and performing abnormal judgment; and when the abnormality is established, determining a reading interval according to the load of the reader-writer, issuing, obtaining a position confirmation event, calculating a lag time difference, and updating the reading interval and issuing when the lag exceeds a threshold value. According to the invention, objective judgment of tracking abnormity and lagging self-adaptive correction are realized.

Description

Technical Field

[0001] This invention relates to the field of information technology in warehousing and logistics, specifically to a method and system for automatic tracking of warehouse goods based on RFID technology. Background Technology

[0002] Modern warehousing, with its multi-category, high-turnover, and multi-operation-point collaboration, places higher demands on the visualization, traceability, and timely correction of goods flow. Tracking capabilities directly impact warehouse location efficiency, inventory accuracy, anomaly handling timeliness, and operational organization optimization, thereby affecting inventory turnover, error costs, and personnel and equipment utilization. Current practices often configure read / write frequencies according to fixed strategies, making it difficult to adapt to differences in the value of different goods and operational stages. Furthermore, the lack of a unified time deviation characterization mechanism between the tracking and business confirmation links, and the lack of standardized constraints on the identification and resolution mechanism from tag identifiers to business goods units, makes it difficult to quickly determine whether anomalies are due to insufficient read / write coverage, reporting congestion, delayed business confirmation, or object confusion caused by inconsistent identifier resolution. This makes it difficult to simultaneously achieve both tracking reliability and resource utilization efficiency. Summary of the Invention

[0003] This invention provides a method and system for automatic tracking of warehouse goods based on RFID technology, which at least solves the problem of how to achieve verifiable anomaly determination and adaptive correction of tracking lag under the load constraints of the reader in warehouse tracking.

[0004] In a first aspect, the present invention provides a method for automatic tracking of warehouse goods based on RFID technology, the method comprising:

[0005] Acquire RFID observation data and reader load data. The RFID observation data includes tag identifiers, area identifiers, and timestamps. Perform identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, form a sequence of area identifiers and timestamps from the RFID observation data.

[0006] Based on the regional identifier and timestamp sequence, the flow speed and dwell time are extracted as flow features. The abnormal threshold and lag threshold corresponding to the cargo attributes are obtained from the parameter library, and the abnormality is judged based on the abnormal threshold.

[0007] When an anomaly is detected, the reading interval is determined based on the reader load data and sent to the reader. The location confirmation event is obtained. The lag time difference is calculated based on the location confirmation event, the area identifier, and the timestamp sequence. When the lag time difference exceeds the lag threshold, the reading interval is updated and sent to the reader.

[0008] In one possible implementation, the identifier resolution includes: retrieving the cargo unit identifier corresponding to the tag identifier in a mapping table between tag identifiers and cargo unit identifiers, and using the retrieved cargo unit identifier as the result of the identifier resolution; wherein, the mapping table is pre-established and stored mapping data, the mapping table includes at least a tag identifier field and a cargo unit identifier field, the tag identifier field and the cargo unit identifier field satisfy a one-to-one correspondence, and an index is established on the tag identifier field to support retrieval by tag identifier field.

[0009] In one possible implementation, forming a region identifier and timestamp sequence from RFID observation data based on cargo unit identifiers includes: grouping RFID observation data according to cargo unit identifiers; sorting RFID observation data corresponding to the same cargo unit identifier by timestamp; and extracting region identifiers and timestamps from the sorted RFID observation data to obtain a region identifier and timestamp sequence.

[0010] In one possible implementation, extracting circulation speed and dwell time as circulation features based on region identifiers and timestamp sequences includes: counting the number of region switching times in the region identifier and timestamp sequence within a preset time window, dividing the number of region switching times by the duration of the preset time window to obtain the circulation speed; and obtaining the dwell time based on the timestamp difference between adjacent region identifiers in the region identifier and timestamp sequence.

[0011] In one possible implementation, the abnormal thresholds include a speed threshold and a dwell time threshold, which are obtained from a parameter library; the abnormality determination includes: determining an abnormality when the flow speed exceeds the speed threshold or the dwell time exceeds the dwell time threshold.

[0012] In one possible implementation, the cargo attributes include a value level, and the reading interval corresponds to the value level, with a smaller reading interval for higher value levels.

[0013] In one possible implementation, the reader load data includes the length of the read request queue or the number of reads per unit time. Determining the read interval based on the reader load data includes: obtaining the read interval parameter corresponding to the length of the read request queue or the number of reads per unit time from the parameter library, and determining the read interval parameter as the read interval.

[0014] In one possible implementation, the location confirmation event includes a cargo unit identifier, a region identifier, and an event timestamp; calculating the lag time difference based on the location confirmation event and the region identifier and timestamp sequence includes: determining the lag time difference as the time difference between the event timestamp of the location confirmation event and the region identifier and the earliest timestamp in the timestamp sequence where the region identifier is equal to the region identifier of the location confirmation event.

[0015] In one possible implementation, updating the read interval when the lag time difference exceeds the lag threshold includes: shortening the read interval to no greater than a preset minimum read interval, where the preset minimum read interval is the minimum polling interval supported by the reader / writer, and sending the shortened read interval to the reader / writer.

[0016] Secondly, the present invention provides an automatic warehouse goods tracking system based on RFID technology, for implementing an automatic warehouse goods tracking method based on RFID technology, the system comprising:

[0017] The data acquisition module is used to acquire RFID observation data and reader load data. The RFID observation data includes tag identifiers, area identifiers, and timestamps. The module performs identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, the RFID observation data is formed into a sequence of area identifiers and timestamps.

[0018] The anomaly detection module is used to extract the flow speed and dwell time as flow features based on the region identifier and timestamp sequence, obtain the anomaly threshold and lag threshold corresponding to the cargo attributes from the parameter library, and make anomaly detection based on the anomaly threshold.

[0019] The interval update module is used to determine the reading interval based on the reader load data and send it to the reader when the anomaly determination is made, obtain the location confirmation event, calculate the lag time difference based on the location confirmation event, the area identifier and the timestamp sequence, and update the reading interval and send it to the reader when the lag time difference exceeds the lag threshold.

[0020] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows:

[0021] By employing time-series serialization and flow feature extraction techniques based on cargo units, verifiable judgment of abnormal flows was achieved; by using threshold configuration based on parameter library and read interval distribution based on load lookup table, matching read / write resources with reader processing capabilities was achieved; and by using location confirmation event alignment and lag time difference calculation techniques, adaptive reading interval shortening and timeliness closure were achieved for tracking lag scenarios. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the execution flow of the method of the present invention;

[0023] Figure 2 This is a structural block diagram of the system of the present invention. Detailed Implementation

[0024] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0025] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0026] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0027] Radio Frequency Identification (RFID) is a non-contact automatic identification technology, typically consisting of electronic tags, readers, and antennas. The electronic tag carries unique identification information, and the reader interacts with the tag in a radio frequency field, performing interrogation and response interactions, thus completing target identification and data collection without visual alignment or manual scanning. A typical deployment of RFID in warehousing scenarios involves: binding tags to goods or cargo carriers; deploying readers at warehouse entrances, passageways, storage locations, or work points; and reporting the tag identifiers collected by the readers, the corresponding area identifiers, and the timestamp of the collection time to the upper-level system via a network. The upper-level system then performs identifier parsing to establish a correspondence between RFID observations and specific cargo units. Based on this data structure and identifier parsing results, the passage and stay of goods in different areas can be mapped into a continuous temporal trajectory, providing a data foundation for subsequent flow status analysis, anomaly identification, and adjustments to read / write strategies.

[0028] like Figure 1 As shown, an automatic tracking method for warehouse goods based on RFID technology includes:

[0029] Acquire RFID observation data and reader load data. The RFID observation data includes tag identifiers, area identifiers, and timestamps. Perform identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, form a sequence of area identifiers and timestamps from the RFID observation data.

[0030] In one embodiment, the system accesses the observation and reporting streams and operational status reporting streams of each RFID reader / writer within the warehouse, forming RFID observation data and reader / writer load data. The RFID observation data includes at least tag identifiers, area identifiers, and timestamps. The area identifier is determined by the configuration relationship between the reader / writer and its installation location, indicating the warehouse unit or work point covered by the reader / writer. The timestamp is generated using a unified clock source, allowing for minor jitter which is corrected through a time synchronization service.

[0031] The reader load data includes at least one or more of the following: read request queue length, number of reads per unit time, and communication round-trip latency, used to characterize the reader's processing pressure in the current time period. Tag identifiers are parsed to obtain cargo unit identifiers, and cargo attributes are retrieved from the warehouse business system. Cargo attributes characterize one or more of the following: cargo category, value level, and operational priority. Subsequently, using the cargo unit identifier as the aggregation key, the RFID observation data is converted into a region identifier and timestamp sequence. The region identifier and timestamp sequence provide a continuous and computable input basis for subsequent extraction of flow rate and dwell time, thereby ensuring the stable operation of the closed-loop link of "observation acquisition—identifier parsing—serialization" in real-time tracking scenarios.

[0032] The identifier resolution includes: retrieving the cargo unit identifier corresponding to the tag identifier in the mapping table between tag identifiers and cargo unit identifiers, and using the retrieved cargo unit identifier as the result of identifier resolution; wherein, the mapping table is pre-established and stored mapping data, the mapping table includes at least a tag identifier field and a cargo unit identifier field, the tag identifier field and the cargo unit identifier field satisfy a one-to-one correspondence relationship, and an index is established for the tag identifier field to support retrieval by tag identifier field.

[0033] In one embodiment, identifier resolution is limited to retrieval-based resolution based on a mapping table. The limitation is that the output of identifier resolution does not rely on temporary rule concatenation or manual intervention, but is directly determined by the established mapping relationship between label identifiers and cargo unit identifiers, thereby constraining the resolution boundary and reducing the risk of the same label being repeatedly interpreted in different business processes.

[0034] In practical implementation, a mapping table between tag identifiers and cargo unit identifiers is maintained. This mapping table must contain at least two fields: a tag identifier field and a cargo unit identifier field. The mapping table can be stored in a relational database or a key-value store, and a unique index is created for the tag identifier to support high-concurrency retrieval. During identifier parsing, the tag identifier is first normalized, including one or more of the following: case-consistent encoding, illegal character removal, and check bit verification, to prevent retrieval failures due to encoding differences. Then, using the tag identifier as the search key, the cargo unit identifier is queried in the mapping table, and the retrieved cargo unit identifier is used as the identifier parsing result.

[0035] To ensure the availability of parsing results in the real-time link, the mapping table retrieval can adopt a two-level structure: A cached index of recently active tag identifiers is maintained on the reader access side or edge gateway side. When the cache is hit, the cargo unit identifier is returned directly; when the cache is missed, the central storage is accessed and the result is backfilled into the cache. Cache entries are set with validity periods and maximum capacity, and are evicted based on the most recent access time. To avoid short-term inconsistencies in the mapping table during business changes, updates are controlled by version numbers or update timestamps. The version number or update timestamp is recorded simultaneously during cache backfilling, and a refresh is triggered when the version number is found to be outdated or the update timestamp is too old.

[0036] A constraint relationship is established between identifier resolution and cargo attribute retrieval: After the cargo unit identifier is determined, the cargo master data table of the warehouse business system is accessed using the cargo unit identifier as the primary key to read fields such as cargo category, value level, and operation priority. When cargo master data is missing, it enters a "pending completion" state while retaining the cargo unit identifier, allowing subsequent serialization and anomaly detection to still be executed. Simultaneously, the missing cargo unit identifier is written to the completion queue, to be completed by the business system during the next synchronization. Through these constraints, the input, processing path, and output of identifier resolution all have auditable data basis, and stable operation can be easily achieved in actual deployment through indexing, caching, and version control.

[0037] The process of forming a region identifier and timestamp sequence from RFID observation data based on cargo unit identifiers includes: grouping RFID observation data according to cargo unit identifiers; sorting RFID observation data corresponding to the same cargo unit identifier by timestamp; and extracting region identifiers and timestamps from the sorted RFID observation data to obtain the region identifier and timestamp sequence.

[0038] In one embodiment, the formation of the region identifier and timestamp sequence is limited to a three-step process of grouping, sorting, and field extraction. The limitation is that discrete observation records of the same cargo unit are transformed into an ordered sequence that can be used to calculate the flow rate and dwell time, thereby constraining the input form of subsequent calculations and reducing temporal ambiguity.

[0039] In practice, the RFID observation data is first grouped according to the cargo unit identifier, forming a grouping result with the cargo unit identifier as the key and the observation record set as the value. Grouping can be done in memory based on a hash table or in a streaming processing framework based on keyed partitioning. Secondly, each group of observation records is sorted by timestamp, with the sorting rule being from smallest to largest timestamp. To adapt to the actual situation of out-of-order observation reporting, an out-of-order tolerance window is introduced before sorting. Late records are allowed to participate in the sorting within the out-of-order tolerance window. Late records exceeding the out-of-order tolerance window are marked as late observations and enter the compensation channel. The compensation channel can be used only to update the end of the sequence or write to the audit log, avoiding real-time link jitter caused by frequent reordering.

[0040] Next, region identifiers and timestamps are extracted from the sorted observation records to generate a region identifier and timestamp sequence. To make the sequence more consistent with the physical process of warehouse tracking, duplicate observations are compressed during the extraction process: when adjacent records have the same region identifier and the time interval is less than the minimum dwell time, only the earliest timestamp is retained as the entry time of the region, and the number of observations in the region is accumulated as a quality marker; when the region identifier changes, the timestamp of the change point is recorded as the cross-region transition time. To reduce "jumping zones" caused by occasional misreading, a consistency confirmation rule is introduced when the region identifier changes: the new region identifier must appear repeatedly in several consecutive observation records, or the number of times it appears within the shortest confirmation time window must reach a threshold, before the new region is written into the region identifier and timestamp sequence; change points that do not meet the conditions are temporarily stored as candidate change points, and if they are subsequently covered by continuous observations of the original region identifier, the candidate change points are discarded.

[0041] The final obtained region identifier and timestamp sequence must at least contain region entry time information arranged in ascending order of time, and may optionally include observation frequency quality markers to facilitate the subsequent removal or downweighting of low-quality segments when extracting flow rate and dwell time. With these constraints, the serialized output can stably connect to subsequent flow feature extraction stages, and remains feasible and reproducible even in typical warehouse environments such as dense reader / writer environments, occlusion, and multipath reflection.

[0042] Based on the regional identifier and timestamp sequence, the flow speed and dwell time are extracted as flow features. The abnormal threshold and lag threshold corresponding to the cargo attributes are obtained from the parameter library, and the abnormality is judged based on the abnormal threshold.

[0043] After the region identifier and timestamp sequence are formed, the flow rate and dwell time are extracted as inputs from this sequence as flow features, and anomaly detection is performed accordingly. Flow rate is used to characterize the frequency of region transitions per unit time, and dwell time is used to characterize the duration for which goods remain unchanged within the same region. The parameter library stores anomaly thresholds and lag thresholds according to goods attributes. The anomaly threshold is used to define the normal boundaries of flow rate and dwell time, and the lag threshold is used to define the time deviation boundaries in the subsequent location confirmation process. Anomaly detection is based on the anomaly threshold to perform threshold comparison on the flow features, output an anomaly flag, and write the anomaly flag and corresponding time window information into the tracking status. This provides clear triggering conditions for subsequent read interval issuance and lag time difference calculation, forming a closed-loop link of feature extraction, threshold acquisition, and anomaly detection.

[0044] Extracting circulation speed and dwell time as circulation features based on region identifiers and timestamp sequences includes: counting the number of region switching times in the region identifier and timestamp sequence within a preset time window, dividing the number of region switching times by the duration of the preset time window to obtain the circulation speed; and obtaining the dwell time based on the timestamp difference between adjacent region identifiers in the region identifier and timestamp sequence.

[0045] In one embodiment, the flow feature extraction is limited to statistical calculation based on time windows. The additional limitation is to solidify the generation path of flow speed and dwell time into "switching count and time difference accumulation", thereby constraining different implementers to still output features with consistent caliber when the observation density is inconsistent.

[0046] In practice, a preset time window length is set. The time window operates on the region identifier and timestamp sequence in a sliding manner, and the window boundary is used to truncate sequence segments based on the timestamp. The number of region switches within the window is counted, with the rule being that two adjacent records with unequal region identifiers are counted as one switch; when the same region appears consecutively due to repeated readings, the switch count is not increased. The throughput is calculated by the number of region switches and the preset time window length, which can be calculated using the following expression:

[0047]

[0048] in, For the flow rate, This is the number of times the region can be switched within a preset time window. The preset time window duration is used. The dwell time is obtained by accumulating the difference between adjacent timestamps: when two adjacent record areas have the same identifier, the difference between the timestamps is accumulated to obtain the dwell time in that area; when two adjacent record areas have different identifiers, the timestamp of the record before the change point is regarded as the departure time, and the timestamp of the record after the change point is regarded as the entry time. This difference is recorded as the cross-area conversion interval and is not included in the dwell time of any area.

[0049] To mitigate the impact of misreading on region switching, a confirmation rule is introduced for region switching counting. A new region identifier is only counted in the switching count if it appears a threshold number of times within the shortest confirmation time window or appears repeatedly in several consecutive records. Changes that do not meet the confirmation rule are temporarily stored as candidate changes, which are discarded if subsequently covered by continuous observations of the original region identifier. To accommodate out-of-order reporting, an out-of-order tolerance window is set. Late records within the out-of-order tolerance window are allowed to be included in the window statistics. Late records exceeding the out-of-order tolerance window are entered into a compensation queue, used only for generating compensation logs or updating audit fields, avoiding frequent recalculations in the real-time link. To prevent distorted dwell time due to sparse observations, a minimum observation threshold is set. When the number of observation records within the window is insufficient, an insufficient data flag is output and feature output is delayed until the observation count meets the threshold before calculation resumes. Through these constraints, the input for calculating flow rate and dwell time, window boundaries, counting rules, and abnormal data processing paths are all clear and executable, allowing those skilled in the art to directly implement and reproduce the same flow characteristics.

[0050] The abnormal thresholds include speed thresholds and dwell time thresholds, which are obtained from the parameter library. Abnormal judgment includes: when the flow speed exceeds the speed threshold or the dwell time exceeds the dwell time threshold, an abnormality is judged.

[0051] In one embodiment, the anomaly threshold is limited to a speed threshold and a dwell threshold. The new limitation is that the anomaly determination is decomposed into two independent threshold comparison conditions, and the threshold is explicitly obtained from the parameter library according to the cargo attribute, thereby constraining the trigger boundary of the anomaly marker and facilitating unified configuration during deployment.

[0052] In practice, the parameter library maintains threshold entries, each containing at least a speed threshold and a dwell threshold. These threshold entries are indexed with cargo attributes, each containing at least one of either a cargo category or a value level. Threshold retrieval uses the cargo attribute as the search key to return both the speed threshold and the dwell threshold. When both cargo category and value level exist, the cargo category is used first to match the threshold entry; if no match is found, the value level is used. If neither matches, a default threshold entry is used, and this default threshold entry is recorded in the tracking status for later configuration updates.

[0053] During anomaly detection, the flow rate is compared with a speed threshold; if the flow rate exceeds the speed threshold, an anomaly is detected. Similarly, the dwell time is compared with a dwell time threshold; if the dwell time exceeds the dwell time threshold, an anomaly is detected. An anomaly flag is output immediately upon fulfilling either condition. To avoid frequent triggering due to a single fluctuation, a continuous window constraint can be introduced: the anomaly flag is only set to valid if the same cargo unit meets any of the anomaly conditions within multiple consecutive preset time windows. The number of continuous windows is configured as a parameter in the parameter library and can change according to cargo attributes.

[0054] To avoid false positives triggered by insufficient data windows, preconditions are added to anomaly detection. These preconditions include the number of observation records within the window reaching the minimum observation threshold and the timestamp coverage ratio within the window reaching the minimum coverage ratio. If the preconditions are not met, an insufficient data flag is output, while the anomaly flag of the previous window remains unchanged. To facilitate troubleshooting and backtracking, the status record window start and end times, flow speed, dwell time, speed threshold, dwell threshold, and anomaly flags are tracked. These record fields support the consistency of the trigger chain for subsequent read interval distribution and location lag verification. By explicitly defining the anomaly thresholds as speed and dwell thresholds, and providing an executable process for data retrieval by cargo attribute and threshold comparison, anomaly detection has clear boundaries, configurable parameters, and auditable output.

[0055] Goods attributes include value level, and the reading interval corresponds to the value level, with a smaller reading interval for higher value levels.

[0056] In one embodiment, the cargo attribute is limited to including a value level, and a monotonic correspondence between the reading interval and the value level is established. The additional limitation is that the value level is used as a constraint input for determining the subsequent reading interval, so that the resource allocation rules after an anomaly is triggered have a clear parameterized mapping rather than relying on empirical criteria.

[0057] In practice, the value level is provided by the cargo master data and is represented by discrete level coding, including at least two levels: low value and high value. The parameter library maintains mapping entries from value levels to target read intervals. The target read interval is represented by a time length and follows a correspondence where higher value levels have smaller target read intervals. The mapping entries also include minimum and maximum read intervals to constrain the read intervals from exceeding device and business boundaries. The minimum read interval is not less than the minimum stable polling interval supported by the RFID reader, and the maximum read interval is not greater than the maximum update interval allowed for warehouse tracking.

[0058] When determining the read interval, the target read interval is first retrieved based on the value level, and then pruned according to the reader load data. The pruning rule maintains monotonicity: the pruned read interval corresponding to a high value level is still no greater than the pruned read interval corresponding to a low value level. The reader load data can be represented by the read request queue length or the number of reads per unit time. The parameter library configures load segmentation thresholds for the reader load data. When the load is in a high load range, the read interval is adjusted towards the maximum read interval but does not exceed the maximum read interval; when the load is in a low load range, the read interval is adjusted towards the target read interval but does not fall below the minimum read interval.

[0059] To ensure traceability in subsequent processes, the tracking status records include the value level, target read interval, trimmed read interval, and reader load data summary field. The summary field must at least include the read request queue length or the number of reads per unit time. Through these constraints, the correspondence between value level and read interval can be directly configured by the parameter library and stably executed during operation. Furthermore, the read interval is always constrained by device capabilities and business limitations, providing a reliable input basis for issuing read intervals after an anomaly occurs.

[0060] When an anomaly is detected, the reading interval is determined based on the reader load data and sent to the reader. The location confirmation event is obtained. The lag time difference is calculated based on the location confirmation event, the area identifier, and the timestamp sequence. When the lag time difference exceeds the lag threshold, the reading interval is updated and sent to the reader.

[0061] When an anomaly is detected, a closed-loop process for determining and correcting the reading interval is executed. First, the reading interval is determined using the reader's load data as a constraint input and then sent to the reader, enabling it to poll and read tags according to the updated interval in subsequent time periods. Next, location confirmation events are acquired. These events, generated by warehouse operations, confirm the arrival or passage of goods in a specific area. The location confirmation events are aligned with the area identifier and timestamp sequence, and a lag time difference is calculated to characterize the time deviation between location confirmation and RFID observation. When the lag time difference exceeds a lag threshold, a reading interval update is triggered, adjusting the reading interval to a shorter value and resending it to the reader. This increases the reader's reading frequency during critical time periods, forming a closed-loop chain of "anomaly triggering—interval sending—event alignment—lag calculation—interval resending," ensuring the tracking link maintains usable timeliness boundaries even in anomaly scenarios.

[0062] The reader load data includes the length of the read request queue or the number of reads per unit time. Determining the read interval based on the reader load data includes: obtaining the read interval parameter corresponding to the length of the read request queue or the number of reads per unit time from the parameter library, and determining the read interval parameter as the read interval.

[0063] In one embodiment, the reader load data is limited to the length of the read request queue or the number of reads per unit time. The added constraint is to solidify the determination of the read interval as a parameterized path of "lookup table by load," preventing the read interval from relying on empirical standards and becoming unreproducible. This constraint is used to constrain the input criteria and value sources for read interval adjustment, ensuring that different readers achieve consistent interval configurations under the same load level.

[0064] In practice, reader load data is periodically reported by the reader or access gateway. The read request queue length represents the backlog of read requests to be processed, and the number of reads per unit time represents the throughput of the reader within the statistical period. The statistical period length is a fixed configuration item, kept consistent within the same database. The parameter library maintains mapping entries from load to read interval. Each mapping entry includes at least a load range and a read interval parameter. The load range can be segmented by the read request queue length or by the number of reads per unit time, and a corresponding read interval parameter is configured for each range. When determining the read interval, the matching load range is first located based on the current read request queue length or the number of reads per unit time. Then, the read interval parameter corresponding to that load range is read from the parameter library, and the read interval parameter is determined as the read interval.

[0065] To ensure the maintainability of the mapping relationship, the load range and read interval parameters are stored in a discrete table structure and support online updates. During online updates, the update version number is written to the parameter library, and the read interval parameter is cached on the access gateway side, carrying the version number. A refresh is triggered when a version number change is detected. To avoid frequent switching caused by load measurement jitter, a hysteresis rule is introduced: when the length of the read request queue fluctuates near the boundary between two adjacent load ranges, the switch to the new load range is only performed if the cross-range condition is met for multiple consecutive statistical periods; the number of consecutive statistical periods is configured by the parameter library. When the read interval is issued, the polling period configuration item is written to the reader through the management channel, and the reader is required to send back configuration confirmation information. The configuration confirmation information includes at least the effective timestamp and the effective read interval. The access side uses the configuration confirmation information as the basis for determining whether the read interval has taken effect. Through the above constraints, the determination and issuance of the read interval form a closed-loop configuration process based on load data, and the value path is clear, configurable, and reproducible.

[0066] Location confirmation events include cargo unit identifiers, area identifiers, and event timestamps; calculating the lag time difference based on the location confirmation event and the area identifier and timestamp sequence includes: determining the lag time difference as the time difference between the event timestamp of the location confirmation event and the area identifier and the earliest timestamp in the timestamp sequence where the area identifier is equal to the area identifier of the location confirmation event.

[0067] In one embodiment, the location confirmation event is defined as including a cargo unit identifier, a region identifier, and an event timestamp. The added constraint is the explicit definition of the three-part field structure for event alignment, and the solidification of the lag time difference calculation as "the difference between the event timestamp and the earliest matching timestamp in the sequence." This constraint is used to constrain the correspondence between location confirmation events and RFID observations, avoiding cross-cargo confusion caused by relying solely on region identifiers.

[0068] In practice, location confirmation events are generated by the warehouse operation system. The generation source can be one or more of the following: manual barcode scanning, gate access records, conveyor line node sensor triggers, and loading / unloading process confirmations. Regardless of the source, each location confirmation event is bound to a specific cargo using the cargo unit identifier as the primary key, and carries a region identifier and an event timestamp. The region identifier maintains the same encoding rule as the region identifier in the reader configuration. When calculating the lag time difference, the corresponding region identifier and timestamp sequence segment is first filtered using the cargo unit identifier. Then, within this segment, the earliest timestamp where the region identifier equals the region identifier of the location confirmation event is found. The difference between the event timestamp of the location confirmation event and this earliest timestamp is determined as the lag time difference. To handle cases where no matching region identifier exists in the region sequence, a missing match flag is set: if no matching region identifier is found within the preset search window, the lag time difference is set to missing and the missing match flag is output. Simultaneously, the read interval update strategy is triggered to enter conservative mode, shortening the read interval to a preset safety value and continuing for several statistical cycles until a matching region identifier appears, at which point normal calculation resumes.

[0069] To handle cases where the same region identifier appears multiple times in a sequence, the earliest timestamp is used as the alignment reference to ensure that the lag time difference reflects the time deviation of the first observation of the region by RFID. When alignment to the most recent entry time into the region is required, the search rule can be switched to "most recent entry time," and an alignment mode identifier can be configured in the parameter library to maintain consistency across the entire warehouse. To ensure timestamp comparability, the event timestamp of the location confirmation event and the timestamp of the region identifier and timestamp sequence use the same clock source or are mapped to the same time reference through a time synchronization service. The time synchronization service records the offset and periodically calibrates it to ensure that the offset is within an acceptable range. Through the above constraints, the field structure of the location confirmation event is clear, the event alignment path is executable, and the lag time difference calculation has clear inputs, search rules, and boundaries for handling abnormal paths.

[0070] Updating the read interval when the lag time difference exceeds the lag threshold includes: shortening the read interval to no greater than the preset minimum read interval, where the preset minimum read interval is the minimum polling interval supported by the reader / writer, and sending the shortened read interval to the reader / writer.

[0071] In one embodiment, the read interval update is defined as shortening the read interval to no greater than a preset minimum read interval when the lag time difference exceeds a lag threshold. The added constraint is to clarify that the "update direction" is shortening and to provide a lower limit boundary, avoiding uncertain update actions or exceeding device capabilities. This constraint is used to constrain the response strategy when the lag exceeds the limit, giving lag correction a feasible execution path.

[0072] In practice, the parameter library maintains a lag threshold and a preset minimum read interval. The preset minimum read interval is the minimum polling interval supported by the reader / writer. This minimum polling interval is provided by the reader / writer device parameters or obtained by the access side through capability negotiation and is written into the device capability table during reader registration. When the lag time difference exceeds the lag threshold, the read interval update rule is to shorten the current read interval, ensuring that the shortened read interval is not greater than the preset minimum read interval. When the current read interval is less than or equal to the preset minimum read interval, the read interval remains unchanged, and the lag exceedance event is recorded. To avoid frequent updates causing congestion in the management channel, an update suppression window is introduced. The update suppression window is a statistical period of fixed duration or fixed number of updates, allowing only one read interval update within the update suppression window. If a lag exceedance occurs again within the update suppression window, only the lag exceedance count is accumulated, and a decision on whether to continue maintaining the minimum read interval is made at the end of the window.

[0073] To ensure the traceability of configuration distribution, distribution records are written to the tracking status. Record fields must include at least the trigger timestamp, lag time difference, lag threshold, read interval before update, read interval after update, and preset minimum read interval. After the read interval is distributed, the reader is required to send back configuration confirmation information. If the access side does not receive configuration confirmation information, it will retry. The number of retries and the retry interval are configured by the parameter library. If confirmation is still not received after exceeding the maximum number of retries, a distribution failure flag is output, and the system enters a degradation mode. In degradation mode, the original read interval is maintained, and priority is given to ensuring continuous reporting of observation data. By limiting the read interval update to a minimum polling interval supported by the reader, and supplementing with distribution confirmation, suppression windows, and failure handling boundaries, read interval correction in lag-exceeding scenarios has a clear, implementable, and device-capable execution closed loop.

[0074] like Figure 2 As shown, an automatic warehouse goods tracking system based on RFID technology is used to implement an automatic warehouse goods tracking method based on RFID technology. The system includes:

[0075] The data acquisition module acquires RFID observation data and reader load data. RFID observation data includes tag identifiers, area identifiers, and timestamps. It performs identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, it forms a sequence of area identifiers and timestamps from the RFID observation data. The hardware includes at least an RFID reader, antenna assembly, reader control unit, and communication interface. The reader polls and responds to the electronic tags through the antenna to collect tag identifiers. The area identifier is fixed by a configuration table of reader installation points and coverage areas. The timestamp is assigned by the reader's local clock or the gateway's unified clock service and reported with the observations. This module is configured with an edge computing unit or gateway processor to perform identifier parsing locally. Typically, this involves accessing a mapping table memory or cache index to map tag identifiers to cargo unit identifiers and acquiring cargo attributes through a service interface or local cache. This module further groups and sorts the observation and reporting streams by time within the edge computing unit using the cargo unit identifier as the aggregation key, forming a region identifier and timestamp sequence and writing it into the local buffer or reporting it to the backend processing node. At the same time, it collects and reports reader load data, which can be provided by the reader firmware counter, including one or more of the following: read request queue length, number of reads per unit time, and communication round-trip latency.

[0076] The anomaly detection module extracts flow speed and dwell time as flow features based on region identifiers and timestamp sequences. It retrieves anomaly thresholds and lag thresholds corresponding to cargo attributes from a parameter library and performs anomaly detection based on these thresholds. Hardware-wise, it can be deployed on a warehouse local server, edge computing server, or data center computing node, and includes at least a general-purpose processor, memory, and a high-speed network interface. This module receives region identifiers, timestamp sequences, and cargo attributes from upstream, uses the processor to perform feature calculations for flow speed and dwell time, and maintains the rolling window state of each cargo unit in memory. The parameter library can be implemented as a local database, embedded key-value store, or centralized configuration service, with corresponding hardware including persistent storage media and a caching layer for quickly retrieving anomaly thresholds and lag thresholds by cargo attributes. The anomaly detection process involves the processor performing threshold comparisons and outputting anomaly flags. The anomaly flags and time window information are written to the state storage area, and the interval update module is notified via a message bus or network interface to trigger subsequent transmission links. To ensure real-time performance, a hardware-accelerated memory queue or shared memory circular buffer can be configured to reduce data copy overhead.

[0077] The interval update module is used to determine the reading interval based on the reader load data and send it to the reader when an anomaly is detected. It also acquires a location confirmation event, calculates the lag time difference based on the location confirmation event, area identifier, and timestamp sequence, and updates the reading interval and sends it to the reader when the lag time difference exceeds a lag threshold. The hardware includes at least a reader management channel interface, an edge controller or configuration distribution gateway, and an interface unit for aligning location confirmation events. This module receives anomaly flags and reader load data. The controller processor looks up the reading interval in a parameter library based on the load data and generates configuration instructions. It then sends the polling interval parameters to the target reader via Ethernet, serial port, or a dedicated management protocol channel. Upon receiving this, the reader firmware updates the polling schedule and sends back configuration confirmation information. Location confirmation events are generated by warehouse operational terminals or automated equipment. Hardware sources may include barcode scanners, gate controllers, conveyor line node sensor controllers, or industrial tablets at workstations. This module collects event records containing cargo unit identifiers, area identifiers, and event timestamps via a network interface, and performs alignment calculations with the area identifier and timestamp sequence on the processor to obtain the lag time difference. When the lag time difference exceeds a lag threshold, the module generates an updated read interval and resends it to the reader / writer. Simultaneously, it records audit fields such as the previous and subsequent read intervals, lag time difference, and trigger timestamp in local storage to support subsequent fault location and maintenance.

[0078] 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.

[0079] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for automatic tracking of warehouse goods based on RFID technology, characterized in that, The method includes: Acquire RFID observation data and reader load data. The RFID observation data includes tag identifiers, area identifiers, and timestamps. Perform identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, form a sequence of area identifiers and timestamps from the RFID observation data. Based on the region identifier and timestamp sequence, the flow speed and dwell time are extracted as flow features. The abnormal threshold and lag threshold corresponding to the cargo attributes are obtained from the parameter library. Anomalies are determined based on the abnormal thresholds. When the anomaly determination is established, the reading interval is determined based on the reader load data and sent to the reader. A location confirmation event is obtained. The lag time difference is calculated based on the location confirmation event and the area identifier and timestamp sequence. When the lag time difference exceeds the lag threshold, the reading interval is updated and sent to the reader.

2. The method according to claim 1, characterized in that, The identifier resolution includes: Retrieve the cargo unit identifier corresponding to the label identifier from the mapping table between label identifiers and cargo unit identifiers, and use the retrieved cargo unit identifier as the result of the identifier parsing; The mapping table is pre-established and stored mapping data. The mapping table includes at least a label identifier field and a cargo unit identifier field. The label identifier field and the cargo unit identifier field have a one-to-one correspondence. An index is established on the label identifier field to support retrieval by the label identifier field.

3. The method according to claim 1, characterized in that, The process of forming the region identifier and timestamp sequence from the RFID observation data based on the cargo unit identifier includes: The RFID observation data are grouped according to the cargo unit identifier; the RFID observation data corresponding to the same cargo unit identifier are sorted by timestamp. The region identifier and timestamp are extracted from the sorted RFID observation data to obtain the region identifier and timestamp sequence.

4. The method according to claim 3, characterized in that, Extracting the flow speed and the dwell time as the flow features based on the region identifier and timestamp sequence includes: Within a preset time window, the number of times the region identifier and timestamp sequence are switched is counted, and the number of times the region is switched is divided by the duration of the preset time window to obtain the flow rate. The duration of stay is obtained based on the difference between the timestamps corresponding to the region identifier and the timestamps of adjacent region identifiers in the timestamp sequence.

5. The method according to claim 4, characterized in that, The abnormal thresholds include a speed threshold and a dwell threshold, which are obtained from the parameter library. The anomaly determination includes determining an anomaly when the flow rate exceeds the speed threshold or the dwell time exceeds the dwell threshold.

6. The method according to claim 1, characterized in that, The cargo attributes include a value level, and the reading interval corresponds to the value level, with a smaller reading interval for higher value levels.

7. The method according to claim 1, characterized in that, The reader load data includes the length of the read request queue or the number of reads per unit time. Determining the read interval based on the reader load data includes: Obtain the read interval parameter corresponding to the length of the read request queue or the number of reads per unit time from the parameter library, and determine the read interval parameter as the read interval.

8. The method according to claim 1, characterized in that, Location confirmation events include cargo unit identifiers, area identifiers, and event timestamps; Calculating the lag time difference based on the location confirmation event and the region identifier and timestamp sequence includes: determining the lag time difference as the time difference between the event timestamp of the location confirmation event and the earliest timestamp in the region identifier and timestamp sequence where the region identifier is equal to the region identifier of the location confirmation event.

9. The method according to claim 8, characterized in that, Updating the read interval when the lag time difference exceeds the lag threshold includes: The reading interval is shortened to no greater than a preset minimum reading interval, which is the minimum polling interval supported by the reader / writer, and the shortened reading interval is sent to the reader / writer.

10. An automatic warehouse goods tracking system based on RFID technology, used to implement the automatic warehouse goods tracking method based on RFID technology as described in any one of claims 1-9, characterized in that, The system includes: The data acquisition module is used to acquire RFID observation data and reader load data. The RFID observation data includes tag identifiers, area identifiers, and timestamps. The module performs identifier parsing on the tag identifiers to obtain cargo unit identifiers and acquire cargo attributes. Based on the cargo unit identifiers, the RFID observation data is formed into a sequence of area identifiers and timestamps. The anomaly detection module is used to extract the flow speed and dwell time as flow features based on the region identifier and timestamp sequence, obtain the anomaly threshold and lag threshold corresponding to the cargo attributes from the parameter library, and make anomaly detection based on the anomaly threshold. The interval update module is used to determine the reading interval based on the reader load data and send it to the reader when the anomaly determination is made, obtain the location confirmation event, calculate the lag time difference based on the location confirmation event, the area identifier and the timestamp sequence, and update the reading interval and send it to the reader when the lag time difference exceeds the lag threshold.

Images