Epc screening rule determination method, system, electronic device, and program product

Abstract

The application relates to the technical field of Internet of Things, and provides an EPC screening rule determination method, system, electronic device and program product. The method comprises the following steps: generating coverage count information according to a target EPC set and a non-target EPC set; performing iterative merging on an initial rule set based on the coverage count information to generate a screening rule set; generating a trigger signal according to performance monitoring data when the RFID reader runs the screening rule set and the coverage situation of the screening rule set; generating a repair strategy based on the trigger signal, and updating the screening rule set of the RFID reader according to the repair strategy. The technical problems that the EPC screening rule lacks dynamic self-adaption and self-repairing capability, the prefix coverage range cannot be quantified, and the rule granularity lacks self-diagnosis capability in the prior art can be solved.

Description

Technical Field

[0001] This application relates to the field of Internet of Things (IoT) technology, specifically to an EPC screening rule determination method, system, electronic device, and program product. Background Technology

[0002] Currently, in the field of Radio Frequency Identification (RFID) technology, Electronic Product Codes (EPCs) serve as unique identifiers for items and are widely used for inventory and tracking in scenarios such as warehousing and logistics, smart manufacturing, and new retail. In practical applications, it is typically necessary to identify the target tag from a large number of RFID tags according to specific screening rules, while filtering out interfering tags.

[0003] Existing EPC screening rules primarily rely on manual configuration and static deployment in RFID readers. Maintenance personnel manually write or adjust the mask and session parameters of the screening rules based on the tag distribution in the field environment, and then configure the rules in the RFID reader for inventory. This manual configuration and static deployment approach results in a lack of dynamic adaptation and self-correction capabilities once the rules are deployed. The aforementioned existing technologies all generate screening rules once based on a data snapshot at a specific moment, and the rules remain static after deployment. In real-world applications, the distribution and types of interfering tags continuously change dynamically, and the initial optimization effect of the static rule set gradually deteriorates or even becomes ineffective over time. Existing technologies neither provide mechanisms for automatically detecting performance degradation nor for automatically adjusting rules according to environmental changes, leaving maintenance entirely reliant on manual intervention. Manual correction processes are slow to respond, prone to errors, and difficult to meet real-time business requirements.

[0004] Meanwhile, this type of static rule generation method based on manual configuration cannot quantify prefix coverage and lacks self-diagnostic capabilities at the rule granularity level. Existing technologies using tree-structured nodes (such as rule trees and Merkle prefix trees) do not include quantitative attributes such as coverage scale, making it impossible to quantitatively assess the coverage scale of each prefix for the target and interfering tags during rule generation and merging. This results in a crude and indiscriminate rule optimization process. When the performance of the RFID reader and its filtering rule operating environment deteriorates, due to the lack of quantitative performance profiles at the individual rule granularity level, the system cannot distinguish whether the root cause of the deterioration is a drastic environmental change or a specific rule failure, and therefore cannot output diagnostic conclusions that can guide precise repair. Summary of the Invention

[0005] This application provides an EPC screening rule determination method, system, electronic device, and program product to solve the technical problems in the prior art where EPC screening rules lack dynamic adaptive and self-repair capabilities, cannot quantify prefix coverage, and lack self-diagnostic capabilities at the rule granularity.

[0006] In a first aspect, embodiments of this application provide a method for determining EPC screening rules, including: Coverage count information is generated based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tags; Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set; A trigger signal is generated based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; A repair strategy is generated based on the trigger signal, and the filtering rule set of the RFID reader is updated according to the repair strategy.

[0007] In one embodiment, the coverage count information is represented by an enhanced prefix tree, wherein each node of the enhanced prefix tree contains a target EPC coverage count value and a non-target EPC coverage count value; The step of generating coverage count information based on the target EPC set and the non-target EPC set includes: Initialize the root node of the enhanced prefix tree; Traverse each target EPC in the target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the target EPC, and increment the target EPC coverage count value of all nodes on the path. Traverse each non-target EPC in the non-target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the non-target EPC, and increment the non-target EPC coverage count value of all nodes on the path.

[0008] In one embodiment, each node of the enhanced prefix tree further includes a node depth, a binary prefix sequence, a parent node reference, and a child node dictionary; Wherein, the target EPC coverage count value corresponding to each binary prefix sequence represents the number of all target EPCs prefixed by the binary prefix sequence, and the non-target EPC coverage count value represents the number of all non-target EPCs prefixed by the binary prefix sequence.

[0009] In one embodiment, the step of iteratively merging the initial rule set based on the coverage count information to generate a filtering rule set includes: Starting with the initial rule set, repeat the following steps until the termination condition is met to obtain the filtering rule set: Select one or more merge candidates from the current rule set, each of which includes an ancestor node and a group of rules to be merged; Calculate the comprehensive reward value for each merger candidate separately; Based on the comprehensive reward value, target merging candidates are determined, and a merging operation is performed on the target merging candidates. The merging operation includes deleting the rule group to be merged from the current rule set and adding a prefix rule corresponding to the ancestor node.

[0010] In one embodiment, the comprehensive reward value is determined by a multi-dimensional reward function, which includes a rule reduction benefit item, a non-target coverage cost penalty item, and a cross-level risk penalty item. The rule reduction benefit item is determined based on the number of rules reduced by the merger operation, the non-target coverage cost penalty item is determined based on the number of non-target EPC coverages increased by the merger operation, and the cross-level risk penalty item is determined based on the depth span of the merger operation.

[0011] In one embodiment, selecting one or more merge candidates from the current rule set includes: For any leaf node in the current rule set, traverse upwards along the parent node reference until a target ancestor node that satisfies the first condition is found. The first condition includes that the target EPC coverage count of the target ancestor node is greater than the target EPC coverage count of the leaf node and the depth of the target ancestor node is less than the depth of the leaf node. If the target ancestor node has at least two child nodes, and both of the at least two child nodes have corresponding rules in the current rule set, and the target EPC coverage count of the target ancestor node is greater than or equal to the sum of the target EPC coverage counts covered by the rules corresponding to the at least two child nodes, then the rule group to be merged, which is composed of the rules corresponding to the target ancestor node and the at least two child nodes, is taken as a merging candidate.

[0012] In one embodiment, the iterative merging satisfies a first hard constraint and a second hard constraint; The first hard constraint includes that at any stage of the merging process, the union of the target EPCs covered by the current rule set is equal to the target EPC set; The second hard constraint includes that the number of rules in the final output filtering rule set does not exceed the upper limit of the number of rules in the device; The optimization objective of the iterative merging includes minimizing the total number of non-target EPCs covered by the current rule set, while satisfying the first hard constraint and the second hard constraint.

[0013] In one embodiment, after performing the merge operation, the method further includes: In the mapping table, the prefix rules newly added for the target merging candidate are mapped to the corresponding ancestor nodes, and the mapping relationships corresponding to the deleted rule groups to be merged are removed; The mapping table is used to maintain the mapping relationship between rules and nodes in the enhanced prefix tree.

[0014] In one embodiment, the termination condition includes either the number of rules in the current rule set being less than or equal to the upper limit of the number of rules in the device, or no merge candidate being selected.

[0015] In one embodiment, the performance monitoring data includes key performance indicators; the coverage of the filtering rule set includes the target EPCs covered by the filtering rule set; The step of generating a trigger signal based on performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set includes: If the degradation of the key performance indicators exceeds the corresponding degradation trigger threshold for multiple consecutive inventory cycles, a health degradation event is generated. The coverage rate is determined based on the target EPCs covered by the filtering rule set and the current target EPC set in the EPC database. If the coverage rate is lower than a first preset threshold, a target coverage insufficiency event is generated.

[0016] In one embodiment, the key performance indicators include any one of inventory rate, target EPC read rate, non-target EPC misread rate, and rule average hit rate.

[0017] In one embodiment, the repair strategy includes a full rule reconstruction strategy and a local fine-tuning strategy; the generation of the repair strategy based on the trigger signal includes: If the trigger signal is a target coverage insufficiency event, a full rule reconstruction strategy is generated. If the trigger signal is a health degradation event, collect EPC samples in the current environment to form a current environment EPC snapshot set, determine the similarity based on the current environment EPC snapshot set and the historical interference EPC set, and generate a full rule reconstruction strategy if the similarity is lower than the similarity threshold. If the similarity is not lower than the similarity threshold, then the rules in the set of filtering rules that meet the preset degradation conditions are marked as degradation rules, and a full rule reconstruction strategy or a local fine-tuning strategy is generated based on the number of degradation rules and the cause of degradation.

[0018] In one embodiment, the local fine-tuning strategy includes: updating the activity weight of a specific interference EPC in the interference management library and adjusting any one of the list categories to which the specific interference EPC belongs; Wherein, the interference EPC is the EPC that causes the increase in the interference introduction ratio of the degradation rule.

[0019] In one embodiment, the full rule reconstruction strategy includes: obtaining the current target EPC set and the current non-target EPC set, and regenerating a new set of filtering rules based on the current target EPC set and the current non-target EPC set.

[0020] In one embodiment, the preset degradation condition includes any one of the following: The interference introduction ratio exceeds the preset multiple of the average interference introduction ratio of other rules; The target contribution rate is lower than the preset ratio; The rule hit rate in recent inventory counts has remained at zero. The interference introduction ratio and the target contribution are determined as follows: the target EPC coverage count and non-target EPC coverage count corresponding to each rule in the filtering rule set are extracted from the enhanced prefix tree; the interference introduction ratio is determined based on the non-target EPC coverage count corresponding to the rule and the current total non-target EPC misreading; and the target contribution is determined based on the target EPC coverage count corresponding to the rule and the total number of target EPCs.

[0021] In one embodiment, updating the filtering rule set of the RFID reader according to the repair strategy includes: The RFID reader is controlled to perform a screening task using the currently effective set of screening rules and to receive tag response signals; The new set of filtering rules obtained after executing the repair strategy will be sent to the cache area of ​​the RFID reader. The tag response signal is simulated and matched with the new set of filtering rules to obtain a simulated matching result; If the simulated matching result indicates that the new filtering rule set meets the verification conditions, then an atomic switching command is sent to the RFID reader during the interval between the two inventory tasks to switch the currently effective filtering rule set to the new filtering rule set.

[0022] In one embodiment, the verification conditions include: the intersection size of the target EPC set matched by the new filtering rule set and the current target EPC set in the EPC database is equal to the total number of target EPCs, and the number of rules in the new filtering rule set does not exceed the upper limit of the number of device rules.

[0023] Secondly, embodiments of this application provide an EPC screening rule determination system, including: a rule generation layer, an autonomous management layer, and an infrastructure layer; The rule generation layer includes an enhanced prefix tree construction module and a merging optimization module; The autonomous management layer includes a real-time inventory performance monitoring module, a problem diagnosis module, a repair execution module, and a verification and update module; The infrastructure layer includes an RFID reader and an antenna. The RFID reader is used to read EPC data of multiple RFID tags within the coverage area through the antenna to obtain a target EPC set and a non-target EPC set, and to run a set of filtering rules. The enhanced prefix tree construction module is used to generate coverage count information based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tag; The merging and optimization module is used to iteratively merge the initial rule set based on the coverage count information to generate a filtering rule set; The real-time inventory performance monitoring module is used to monitor the performance monitoring data of the RFID reader when running the filtering rule set and the coverage of the filtering rule set; The problem diagnosis module is used to generate a trigger signal based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; and to generate a repair strategy based on the trigger signal. The repair execution module is used to execute the repair strategy; The verification update module is used to update the filtering rule set of the RFID reader according to the repair strategy after the repair execution module executes the repair strategy.

[0024] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the EPC screening rule determination method described in the first aspect.

[0025] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the EPC screening rule determination method described in the first aspect.

[0026] The EPC screening rule determination method, system, electronic device, and program product provided in this application obtain a target EPC set and a non-target EPC set and generate coverage count information. They then precisely quantify the number of target tags and interfering tags covered by the prefix of each candidate rule, achieving granular coverage quantification. Based on this, the initial rule set is iteratively merged using the coverage count information to generate a screening rule set with better coverage. This proactively sorts out and eliminates redundancy in the coverage relationships between rules from the rule generation stage, achieving granular coverage quantification. After the screening rule set is deployed on the RFID reader, a trigger signal is generated based on the reader's performance monitoring data and the coverage status of the screening rule set, enabling real-time tracking and perception of changes in the reader's operating status. Based on the trigger signal, rules with abnormal coverage are diagnosed and located, and a repair strategy is generated. The RFID reader's screening rule set is then updated according to the repair strategy, allowing the screening rules to dynamically adjust according to changes in the reader's actual operating performance and coverage, continuously adapting to the dynamically changing inventory environment. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in 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.

[0028] Figure 1 This is the overall architecture diagram of the EPC screening rule determination system provided in the embodiments of this application; Figure 2 This is a flowchart illustrating the EPC screening rule determination method provided in the embodiments of this application; Figure 3 This is an example diagram of the enhanced prefix tree structure and its construction provided in the embodiments of this application; Figure 4 This is a schematic diagram of greedy rule merging provided in an embodiment of this application; Figure 5 This is a flowchart of the greedy merging iteration provided in the embodiments of this application; Figure 6 This is a core flowchart of autonomous management provided in the embodiments of this application; Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. 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.

[0030] This application provides an EPC screening rule determination system, referring to... Figure 1 As shown, it includes: a rule generation layer, an autonomous management layer, and an infrastructure layer; The rule generation layer includes an enhanced prefix tree construction module and a merging optimization module; The autonomous management layer includes a real-time inventory performance monitoring module, a problem diagnosis module, a repair execution module, and a verification and update module; The infrastructure layer includes an RFID reader and an antenna. The RFID reader is used to read EPC data of multiple RFID tags within the coverage area through the antenna to obtain a target EPC set and a non-target EPC set, and to run a set of filtering rules. The enhanced prefix tree construction module is used to generate coverage count information based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tag; The merging and optimization module is used to iteratively merge the initial rule set based on the coverage count information to generate a filtering rule set; The real-time inventory performance monitoring module is used to monitor the performance monitoring data of the RFID reader when running the filtering rule set and the coverage of the filtering rule set; The problem diagnosis module is used to generate a trigger signal based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; and to generate a repair strategy based on the trigger signal. The repair execution module is used to execute the repair strategy; The verification update module is used to update the filtering rule set of the RFID reader according to the repair strategy after the repair execution module executes the repair strategy.

[0031] The infrastructure layer includes RFID readers and antennas, as well as an EPC database, a health baseline library, an interference management library, an autonomous event log library, and the RFID readers and antennas themselves. The EPC database stores the latest set of target EPCs, i.e., the EPC data of all target tags corresponding to items to be identified, providing input data for the rule generation layer. The health baseline library stores the baseline averages and degradation trigger thresholds of various key performance indicators collected by the RFID readers under normal operating conditions. The performance monitoring module generates health degradation events by comparing the KPIs of the current inventory cycle with the degradation trigger thresholds in the baseline library. The interference management library stores the set of non-target EPCs, i.e., the EPC data corresponding to interfering tags, and records the activity weight and category of each interfering EPC, providing non-target EPC input data for the rule generation layer and providing the basis for updating interfering EPCs for local fine-tuning strategies. The autonomous event log library stores self-healing event reports throughout the entire lifecycle of autonomous behavior. This report includes at least the trigger event type and data, intermediate data and conclusions during the diagnostic process, decision basis, copies of the old and new screening rule sets, and shadow operation verification results, serving as technical evidence that the system possesses self-healing capabilities.

[0032] After the RFID reader reads the EPC data returned by all RFID tags within the coverage area through its antenna, it queries the EPC database for each read EPC. If the EPC exists in the target EPC set in the database, it is assigned to that set. If the EPC exists in the non-target EPC set in the interference management database, it is assigned to that set. If the EPC exists in neither the target nor the non-target set, it is processed according to preset rules (e.g., marked as an unknown tag upon first appearance, or ignored). This query and classification process is executed by the RFID reader or its connected edge computing nodes, resulting in the target EPC set and the non-target EPC set after classification.

[0033] The specific execution flow of each module of the system is described in the method embodiments below. The EPC screening rule determination method provided in the embodiments of this application is described below, and the EPC screening rule determination method described below can be referred to in correspondence with the EPC screening rule determination system described above.

[0034] The EPC screening rule determination method provided in this application is actually an autonomous EPC screening rule method. The term "autonomy" in this application refers to the EPC screening rule determination system's ability to overcome reliance on manual operation and maintenance, possessing the capability to autonomously perceive environmental changes, autonomously diagnose performance degradation, autonomously classify and repair problems, and autonomously verify the security of repair results. This forms a fully automated closed loop of "monitoring-diagnosis-decision-repair-verification," continuously maintaining full coverage and low-error coverage performance of the target EPCs without human intervention during long-term operation. Specifically, by acquiring the target EPC set and non-target EPC sets and generating coverage count information, the number of target tags and interfering tags covered by the prefix of each candidate rule is precisely quantified, achieving quantification of the coverage range at the rule granularity. Based on the coverage count information, the initial rule set is iteratively merged to generate a screening rule set with better coverage, actively sorting out and eliminating redundancy in the coverage relationships between rules during the rule generation stage. When the screening rule set is deployed on the RFID reader, a trigger signal is generated based on the reader's performance monitoring data and the actual coverage of the screening rule set, achieving real-time tracking and perception of the operating status, and using performance baselines and... The system autonomously monitors changes in the distribution of interfering tags in real-time KPI monitoring; based on trigger signals, it diagnoses and locates rules with abnormal coverage, distinguishing between two root causes: drastic environmental changes and rule degradation, through a problem diagnosis module; based on the diagnostic conclusions, it autonomously decides and executes repair strategies, including full reconstruction or local fine-tuning; it updates the RFID reader's filtering rule set according to the repair strategy, and achieves safe and seamless online deployment of repaired rules through shadow operation verification and atomic switching, enabling the filtering rules to be dynamically adjusted according to changes in the reader's actual operating performance and coverage, forming a fully automated closed loop of "monitoring-diagnosis-decision-repair-verification," continuously adapting to dynamically changing inventory environments. All of the above steps are linked in a closed loop by the system without human intervention, which is the core meaning of the "autonomy" in this application.

[0035] Figure 2 This is a flowchart illustrating the EPC screening rule determination method provided in an embodiment of this application. (Refer to...) Figure 2 This application provides a method for determining EPC screening rules, which may include: S110. Generate coverage count information based on the target EPC set and the non-target EPC set.

[0036] The target EPC set mentioned above refers to the set of EPCs corresponding to the items to be identified, that is, the set of EPCs corresponding to the target tags that the user expects to identify from the inventory site. The non-target EPC set refers to the set of EPCs corresponding to interfering tags, that is, the set of electronic product codes corresponding to tags that exist in the field environment, which the user does not expect to identify but may interfere with the inventory results. The EPC is a unique identifier for the item stored in the RFID tag, and it is a binary code.

[0037] The prefix is ​​a bit sequence in the EPC from the most significant bit to a certain length. Coverage count information is statistical data describing the coverage scale of each prefix on the target EPC set and non-target EPC sets. Specifically, for each prefix, the coverage count information includes the target EPC coverage count value and the non-target EPC coverage count value corresponding to that prefix. The target EPC coverage count value indicates the number of EPCs in the target EPC set that match the prefix, and the non-target EPC coverage count value indicates the number of EPCs in the non-target EPC set that match the prefix.

[0038] In one specific embodiment, for each prefix, all EPCs in the target EPC set are traversed, and the number of EPCs matching the prefix is ​​counted to obtain the target EPC coverage count value corresponding to the prefix; all EPCs in the non-target EPC set are traversed, and the number of EPCs matching the prefix is ​​counted to obtain the non-target EPC coverage count value corresponding to the prefix.

[0039] S120. Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set.

[0040] The initial rule set is the set of rules initially configured before iterative merging, with each rule corresponding to a prefix. The filtering rule set is the set of rules finally deployed on the RFID reader after iterative merging.

[0041] Iterative merging involves repeatedly merging rules in the initial rule set. Each merge combines two or more rules with similar coverage relationships into a new rule with broader coverage, until a preset termination condition is met. The changes in the target EPC coverage count and non-target EPC coverage count before and after merging are evaluated based on coverage count information to determine whether to perform the merge and its priority. The merge operation is performed if the merged target EPC coverage count is not lower than a preset coverage threshold and the non-target EPC coverage count is not higher than a preset false coverage threshold; otherwise, the merge is not performed.

[0042] For the new rules obtained through merging, recalculate their coverage count information, that is, calculate the target EPC coverage count value and the non-target EPC coverage count value corresponding to the new rule. Repeat the above merging operation until the termination condition is met, then stop merging and output the current rule set as the filtering rule set.

[0043] S130. Generate a trigger signal based on the performance monitoring data of the RFID reader when running the filtering rule set and the coverage of the filtering rule set.

[0044] Performance monitoring data refers to various operational metrics collected by the RFID reader during inventory operations using a set of filtering rules. In one specific embodiment, performance monitoring data includes, but is not limited to: slot conflict rate, empty slot percentage, tag read success rate, and inventory cycle duration. The slot conflict rate is the ratio of the number of slots with tag conflicts per unit inventory cycle to the total number of slots; the empty slot percentage is the ratio of the number of slots without tag responses to the total number of slots; the tag read success rate is the ratio of the number of successfully read tags to the expected total number of tags; and the inventory cycle duration is the time required to complete one full inventory operation.

[0045] Coverage refers to the actual coverage statistics of the filtering rule set over the target EPC set and non-target EPC sets during the current operation. In a specific embodiment, coverage includes the actual coverage rate and false alarm rate of the current filtering rule set. The actual coverage rate is the ratio of the number of target tags successfully identified by the current filtering rule set to the total number of tags in the target EPC set; the false alarm rate is the ratio of the number of interfering tags incorrectly identified as non-target tags by the current filtering rule set to the total number of interfering tags.

[0046] Performance monitoring data and coverage status together reflect the actual operational effectiveness of the current filtering rule set on the RFID reader. The trigger signal is an anomaly indication signal generated based on performance monitoring data and coverage status. A trigger signal is generated when any indicator in the performance monitoring data exceeds a preset normal operating threshold, or when the coverage rate is lower than a preset coverage threshold or the false alarm rate is higher than a preset false alarm rate threshold. The trigger signal indicates that the current filtering rule set is not functioning correctly and a diagnostic and repair process needs to be initiated.

[0047] S140. Generate a repair strategy based on the trigger signal, and update the filtering rule set of the RFID reader / writer according to the repair strategy.

[0048] The repair strategy is a scheme to adjust the current filtering rule set to improve the operating performance of the RFID reader. Upon receiving a trigger signal, a diagnostic process is initiated based on the performance monitoring data and coverage corresponding to the trigger signal to locate specific problems in the current filtering rule set. Based on the diagnostic results, a corresponding repair strategy is generated. The repair strategy includes two methods: full reconstruction and local fine-tuning. Full reconstruction involves re-executing steps S110 and S120 to regenerate the filtering rule set based on the current EPC set; local fine-tuning involves targeted adjustments to some rules in the current filtering rule set to improve the reader's operating performance.

[0049] The generated repair strategy is applied to the RFID reader, updating the currently running set of filtering rules in the RFID reader. After the update is complete, the RFID reader continues to perform inventory operations using the updated set of filtering rules.

[0050] The EPC filtering rule determination method provided in this application obtains a target EPC set and a non-target EPC set and generates coverage count information. It then precisely quantifies the number of target tags and interfering tags covered by the prefix of each candidate rule, achieving granular coverage quantification. Based on this, the initial rule set is iteratively merged using the coverage count information to generate a filtering rule set with better coverage. This proactively sorts out and eliminates redundancy in the coverage relationships between rules from the rule generation stage, achieving granular coverage quantification. After the filtering rule set is deployed on the RFID reader, a trigger signal is generated based on the reader's performance monitoring data and the coverage status of the filtering rule set, enabling real-time tracking and perception of changes in the reader's operating status. Based on the trigger signal, rules with abnormal coverage are diagnosed and located, and a repair strategy is generated. The filtering rule set of the RFID reader is then updated according to the repair strategy, allowing the filtering rules to be dynamically adjusted according to changes in the reader's actual operating performance and coverage, continuously adapting to the dynamically changing inventory environment.

[0051] In one embodiment, to achieve accurate quantification of the number of target and interfering tags covered by different EPC prefixes, the coverage count information is represented by an enhanced prefix tree. (Refer to...) Figure 3 The diagram shows the data structure of an enhanced prefix tree. Each node in the enhanced prefix tree contains the following attribute fields: target EPC coverage count and non-target EPC coverage count. Each node in the enhanced prefix tree also contains the following attribute fields: node depth, binary prefix sequence, parent node reference, and child node dictionary. Wherein, the target EPC coverage count value corresponding to each binary prefix sequence represents the number of all target EPCs prefixed by the binary prefix sequence, and the non-target EPC coverage count value represents the number of all non-target EPCs prefixed by the binary prefix sequence.

[0052] This application enhances the prefix tree by setting node depth, binary prefix sequence, parent node reference, and child node dictionary at each node, and using the binary prefix sequence as the node positioning identifier. This allows each node to store a target EPC coverage count value that accurately counts the number of all target EPCs prefixed by the binary prefix sequence, and store a non-target EPC coverage count value that accurately counts the number of all non-target EPCs prefixed by the binary prefix sequence. Thus, each node stores the target EPC coverage count value and non-target EPC coverage count value in association with the node depth, binary prefix sequence, parent node reference, and child node dictionary. This enables the target EPC coverage count value and non-target EPC coverage count value corresponding to any binary prefix sequence to be directly obtained through the node corresponding to the prefix sequence, achieving a precise mapping from the binary prefix sequence to its coverage count value.

[0053] For any node in the enhanced prefix tree The depth of this node is The binary prefix sequence of this node is This indicates the distance from the root node to the node. The bit sequence corresponding to the path. . Represents binary prefix sequence The first in 1 bit Represents binary prefix sequence The first in 1 bit.

[0054] Let the target EPC set be The non-target EPC set is The target EPC coverage count value of node v is defined as follows: The non-target EPC coverage count value is defined as follows: ,in EPC The former Bit binary prefix. for or Any EPC in the, The cardinality of the set is the number of elements within the curly braces that satisfy the condition; that is, the count.

[0055] The step of generating coverage count information based on the target EPC set and the non-target EPC set includes: S1101. Initialize the root node of the enhanced prefix tree.

[0056] The root node has a depth of 0, the binary prefix sequence is empty, and both the target EPC coverage count and the non-target EPC coverage count are 0.

[0057] S1102. Traverse each target EPC in the target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the target EPC, and increment the target EPC coverage count value of all nodes on the path.

[0058] Traverse each target EPC in the target EPC set. For each target EPC, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the target EPC until the leaf node corresponding to the EPC is reached. Specifically, for each bit in the binary bit sequence of the target EPC, use the bit value as the key to check if a corresponding child node exists in the child node dictionary of the current node; if it exists, move down to the next level along the child node; if it does not exist, create a new child node with the bit value as the key, and then move down to the next level along the new child node. For example, suppose the binary prefix sequence of the current node is 101, and its child node dictionary stores the child node N1 (binary prefix sequence is 1010) corresponding to key 0 and the child node N2 (binary prefix sequence is 1011) corresponding to key 1. When processing the next bit value of 0, search for the key 0 in the child node dictionary of the current node. If a child node N1 is found, the match is successful, and processing continues along child node N1 to process subsequent bits. If processing the next bit value of 0 but the key 0 is not found in the child node dictionary of the current node, the match fails, and a new child node with the key 0 is created. The binary prefix sequence of this new child node is 1010, and processing continues along this new child node to process subsequent bits. Repeat the above process until all bits of the EPC have been processed.

[0059] After path expansion is completed, the target EPC coverage count of all nodes on the path is incremented by 1. A path is a sequence of all nodes traversed from the root node, following the binary bit sequence of the EPC layer by layer down to the corresponding leaf node. For example, assuming the binary bit sequence of the target EPC is 1→0→1→1, in the enhanced prefix tree, starting from the root node, it sequentially passes through nodes N1 (prefix 1), N2 (prefix 10), N3 (prefix 101), and N4 (prefix 1011). These four nodes together constitute the complete path from the root node to the leaf node N4. The target EPC coverage count of all nodes on the path is incremented by 1, that is, the target EPC coverage counts of nodes N1, N2, N3, and N4 are all incremented by 1.

[0060] S1103. Traverse each non-target EPC in the non-target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the non-target EPC, and increment the non-target EPC coverage count value of all nodes on the path.

[0061] Iterate through each non-target EPC in the set of non-target EPCs and insert it into the enhanced prefix tree in the same way. For each non-target EPC, starting from the root node, match and expand the child nodes layer by layer according to the binary bit sequence of the non-target EPC until the leaf node corresponding to the EPC is reached. After completing the path expansion, increment the non-target EPC coverage count of all nodes on the path by 1.

[0062] In the constructed enhanced prefix tree, any node can quantify the number of target EPCs and non-target EPCs among all EPCs prefixed with that node's binary prefix sequence. This quantization result provides direct data support for the subsequent iterative merging of the initial rule set based on coverage count information.

[0063] This application constructs an enhanced prefix tree, setting a target EPC coverage count and a non-target EPC coverage count at each node. It then expands child nodes layer by layer according to the binary bit sequence of the target EPC, incrementing the target EPC coverage count for all nodes along the path, and similarly, expands child nodes layer by layer according to the binary bit sequence of the non-target EPC, incrementing the non-target EPC coverage count for all nodes along the path. This ensures that for any node in the enhanced prefix tree, the target EPC coverage count accurately represents the number of EPCs in the target EPC set whose binary prefix sequence is a prefix of that node, and the non-target EPC coverage count accurately represents the number of EPCs in the non-target EPC set whose binary prefix sequence is a prefix of that node. Thus, each node in the enhanced prefix tree accurately records the respective number of target EPCs and non-target EPCs among all EPCs whose binary prefix sequence is a prefix of that node. This achieves precise quantification of the target EPC coverage count and the non-target EPC coverage count corresponding to any prefix, providing a data foundation for evaluating the changes in the target EPC coverage count and the non-target EPC coverage count before and after merging in subsequent iterative merging, enabling iterative merging to be performed based on accurate coverage counts.

[0064] In one embodiment, this application employs a greedy rule merging algorithm to iteratively merge the initial rule set. The basic idea is to start with the initially generated full-match rule set (each rule corresponding to a complete EPC), and repeatedly search for mergeable rule groups by leveraging the parent-child relationships between nodes in the enhanced prefix tree. This involves replacing a set of longer child node rules covered by a shorter ancestor node prefix rule, thereby gradually reducing the total number of rules. Each merge changes the composition of the rule set. The merged rule set, while satisfying the target full coverage constraint, has a decreasing number of rules, eventually approaching the upper limit of the device's rule count, while simultaneously controlling the erroneous coverage range of non-target tags. Figure 4 As shown in the diagram, this illustration illustrates the corresponding changes in the rule set in the enhanced prefix tree before and after a typical merging operation: multiple leaf node rules are replaced by a single ancestor node rule with a smaller depth. This application implements the above idea into an improved greedy algorithm, configured to iteratively merge the initial set of fully matching rules (i.e., the initial rule set mentioned above) into a simplified set of prefix rules, thus obtaining the filtering rule set.

[0065] In one embodiment, reference is made to Figure 5 As shown, let the initial rule set be... Each rule A node in the corresponding augmented prefix tree The maximum number of rules for a device is The iterative merging satisfies the first hard constraint and the second hard constraint; The first hard constraint includes the condition that at any stage of the merging process, the union of the target EPCs covered by the current rule set is equal to the target EPC set; the first hard constraint is expressed as follows: , Representation rules The target EPC set covered Target EPC set, full target EPC set (i.e., the complete set of EPCs corresponding to all items to be identified), and union of all target EPCs covered by the current rule set at any stage of the merging process. The set of target EPCs must be equal to the initial input, and no target EPCs can be omitted. In other words, the merge operation cannot cause any target EPC to be lost from the coverage of the rule set, and every EPC in the target EPC set is always covered by at least one rule in the current rule set.

[0066] The second hard constraint includes the number of rules contained in the final output filtering rule set. The number of device rules shall not exceed the upper limit; the second hard constraint is expressed as That is, the number of rules in the final output filtering rule set must not exceed the maximum number of rules that the RFID reader / writer device can support.

[0067] The optimization objective of the iterative merging includes: minimizing the total number of non-target EPCs covered by the current rule set, while satisfying the first hard constraint and the second hard constraint. Let... Representation rules The set of non-target EPCs covered, the optimization objective is expressed as That is, under the premise of satisfying the two constraints of full target coverage (first hard constraint) and the number of rules not exceeding the device limit (second hard constraint), the total number of non-target EPCs covered by the current rule set is maximized. Minimize, that is, reduce the coverage of the rule set on interfering labels as much as possible, thereby reducing false coverage. It is the union of the non-target EPCs covered by the current rule set, representing the total set of interference labels covered by the current rule set.

[0068] This application sets a first hard constraint to ensure that the union of the target EPCs covered by the current rule set at any stage of the merging process equals the target EPC set. This ensures that no target EPC is lost from the rule set's coverage during each merging operation, maintaining complete coverage of the target EPC set during iterative merging and rule simplification. A second hard constraint ensures that the number of rules in the final output filtering rule set does not exceed the device's rule limit, guaranteeing that the merged filtering rule set can be deployed and function normally on RFID reader devices, avoiding rule loading failures or reader malfunctions due to exceeding device hardware limitations. The optimization objective is set to minimize the total number of non-target EPCs covered by the current rule set while satisfying the first and second hard constraints. This allows iterative merging to continuously reduce the rule set's coverage of non-target EPCs while maintaining full target coverage and the device's rule limit, thus reducing false coverage of interfering tags. This achieves joint optimization of the rule set under three constraint dimensions: full target coverage, the device's rule limit, and minimum non-target coverage.

[0069] The step of iteratively merging the initial rule set based on the coverage count information to generate a filtering rule set includes: Starting with the initial rule set, repeat the following steps until the termination condition is met to obtain the filtering rule set: S210. Select one or more merge candidates from the current rule set, each of the merge candidates including an ancestor node and a rule group to be merged.

[0070] The rule group to be merged is a set of rules in the current rule set that takes the ancestor node as its ancestor. That is, the prefixes corresponding to all rules in the rule group to be merged are descendant prefixes of the prefix corresponding to the ancestor node. The prefix rule corresponding to the ancestor node is to replace the set of child node rules it covers with the prefix rule of the ancestor node. In other words, the shorter prefix rule of the ancestor node replaces the set of longer child node rules it covers, so that the number of rules decreases.

[0071] S220. Calculate the comprehensive reward value for each merged candidate.

[0072] The overall reward value is used to evaluate the degree of performance improvement of the rule set after performing the merge operation. The higher the overall reward value, the better the simplification and coverage control of the rule set after performing the merge operation.

[0073] S230. Determine target merging candidates based on the comprehensive reward value, and perform a merging operation on the target merging candidates. The merging operation includes deleting the rule group to be merged from the current rule set and adding a prefix rule corresponding to the ancestor node. That is, replace all rules in the rule group to be merged with the prefix rule of the ancestor node, so that the number of rules in the rule set decreases by 1. When the rule group to be merged contains n rules, delete n rules and add 1 rule, resulting in a net reduction of n-1 rules.

[0074] Repeat the above process of selecting merging candidates, calculating the comprehensive reward value, determining the target merging candidate and performing the merging operation (i.e., S220-S230) until the termination condition is met, stop merging, and output the current rule set as the filtering rule set.

[0075] This application selects one or more merging candidates from the current rule set. Each merging candidate includes an ancestor node and a group of rules to be merged, such that the prefixes corresponding to all rules in the group to be merged are descendant prefixes of the prefixes corresponding to the ancestor node. This replaces all rules in the group to be merged with the prefix rules of the ancestor node, achieving the replacement of longer child node rules with shorter ancestor node prefix rules. By calculating the comprehensive reward value of each merging candidate, the improvement in the overall performance of the rule set after performing the merging operation is quantitatively evaluated, providing a quantifiable basis for the selection of the merging operation. The target is determined based on the comprehensive reward value. Merge candidates and perform a merge operation. Remove the rule group to be merged from the current rule set and add the prefix rule corresponding to the ancestor node. This reduces the number of rules in the rule set by 1, thus gradually decreasing the number of rules. At the same time, the target EPC coverage count and non-target EPC coverage count of the prefix rule corresponding to the newly added ancestor node are used as quantitative indicators of the performance of the replaced rule. This allows each merge to quantitatively evaluate the coverage performance of the merged rule by using the target EPC coverage count and non-target EPC coverage count as the rule number decreases, achieving a balance between rule set simplification and coverage performance control.

[0076] In one embodiment, to guide the direction of the merge iteration, this module calculates a comprehensive reward value for each merge candidate. The comprehensive reward value is determined by a multi-dimensional reward function, which includes a rule reduction benefit term, a non-target coverage cost penalty term, and a cross-level risk penalty term. The rule reduction benefit item is determined based on the number of rules reduced by the merger operation, the non-target coverage cost penalty item is determined based on the number of non-target EPC coverages increased by the merger operation, and the cross-level risk penalty item is determined based on the depth span of the merger operation.

[0077] For the merge candidate (ancestor node A, the rule group to be merged) Let S be... The set of nodes corresponding to the rules, and the total reward. The calculation formula is: ; in, This is a preset positive coefficient used to reduce the number of reward rules; This is a preset negative coefficient used to penalize the increase in the number of non-target EPC coverages caused by the merging operation; This is a preset second negative coefficient used to penalize the risk of loss of control over coverage caused by cross-level merging operations. The number of rules in the rule group to be merged. This represents the size of the node set.

[0078] Specifically, the first item The reduction of benefit items for rules is determined based on the number of rules reduced during the merge operation. (Rule group to be merged) Includes After merging, one new prefix rule corresponding to the ancestor node is added, resulting in a net reduction in the number of rules. Item, multiplied by a positive coefficient Obtaining the rule reduces the benefit items.

[0079] Second item This is a penalty item for non-target coverage costs, determined based on the increase in non-target EPC coverage due to the consolidation operation. The non-target EPC coverage count value of ancestor node A. For nodes Non-target EPC coverage count value, This is the sum of the non-target EPC coverage counts of all nodes in the rule group to be merged. The coverage of ancestor node A covers the coverage of all its descendant nodes v∈S, therefore... Greater than or equal to The difference between the two represents the number of non-target EPC coverages added due to the merging operation, multiplied by the first negative coefficient. Obtain a penalty for non-target coverage costs.

[0080] Third item This is a cross-level risk penalty item, determined based on the depth and span of the consolidation operation. Among them, Let A be the depth of ancestor node A. For any node in the rule group to be merged depth, The average depth of all nodes in the rule group to be merged is given by the difference between the two values, which represents the depth span of the merge operation. Multiplying this by the second negative coefficient γ yields the cross-level risk penalty term. The absolute value is set much smaller than and This design ensures that the weight of the cross-level risk penalty is lower than that of the rule reduction benefit penalty and the non-target coverage cost penalty. The intention behind the cross-level penalty is to make the system prioritize merging with smaller depth spans when the overall reward values ​​of multiple merge candidates are similar, so as to preserve the fine-grained flexibility of the rule set and facilitate subsequent local diagnosis and fine-tuning.

[0081] This application calculates a comprehensive reward value for each merge candidate. The comprehensive reward value includes a rule reduction benefit item, a non-target coverage cost penalty item, and a cross-level risk penalty item. The rule reduction benefit item is determined based on the number of rules reduced by the merge operation, ensuring that merge candidates with greater rule reductions receive higher reward values, thus guiding iterative merges to prioritize merge operations with significant rule simplification effects. The non-target coverage cost penalty item is determined based on the increase in non-target EPC coverage after the merge operation, ensuring that merge candidates with less increase in non-target EPC coverage after the merge receive lower penalties, thus guiding iterative merges to prioritize merge operations with the least impact on interference label coverage. Controlling the erroneous coverage of non-target labels; the cross-level risk penalty is determined based on the depth span of the merge operation. The larger the depth span, the wider the EPC coverage of the prefix rule after the merge, and the higher the risk of the coverage getting out of control. This makes iterative merging prioritize merging with smaller depth spans when multiple candidate comprehensive reward values ​​are similar, avoiding difficulties in subsequent rule adjustments due to a single merge with an excessively large span. The above three dimensions together constitute a quantitative evaluation of the comprehensive reward value. The merge candidate with the highest comprehensive reward value is determined as the target merge candidate, so that each iterative merge is based on the quantitative comparison of comprehensive reward values ​​to make the optimal selection, achieving a balance between rule simplification and coverage performance control.

[0082] In one embodiment, selecting one or more merge candidates from the current rule set includes: For any leaf node in the current rule set, traverse upwards along the parent node reference until a target ancestor node that satisfies the first condition is found. The first condition includes that the target EPC coverage count of the target ancestor node is greater than the target EPC coverage count of the leaf node and the depth of the target ancestor node is less than the depth of the leaf node. If the target ancestor node has at least two child nodes, and both of the at least two child nodes have corresponding rules in the current rule set, and the target EPC coverage count of the target ancestor node is greater than or equal to the sum of the target EPC coverage counts covered by the rules corresponding to the at least two child nodes, then the rule group to be merged, which is composed of the rules corresponding to the target ancestor node and the at least two child nodes, is taken as a merging candidate.

[0083] This application selects one or more merging candidates from the current rule set and generates merging candidates in a targeted manner through the "nearest effective ancestor node" mechanism. Specifically, for any leaf node N in the current rule set, it traverses upwards along its parent node reference, that is, it visits its parent node layer by layer. For each visited ancestor node, it determines whether its target EPC coverage count value A.t_count is greater than the target EPC coverage count value N.t_count of N, and whether its depth A.depth is less than the depth N.depth of N. The ancestor node that first satisfies the above conditions is taken as the target ancestor node A.

[0084] If the target ancestor node A has at least two child nodes, and both of these child nodes have corresponding rules in the current rule set, and the target EPC coverage count A.t_count of the target ancestor node A is greater than or equal to the sum of the target EPC coverage counts t_count of the rules corresponding to each of the at least two child nodes, then the rule group to be merged, consisting of the target ancestor node A and the rules corresponding to the at least two child nodes, is considered as a merge candidate. The rule group to be merged consists of the rules corresponding to the at least two child nodes of the target ancestor node A. All rules in the rule group to be merged have A as their ancestor node; that is, the prefixes corresponding to these rules are all descendant prefixes of the binary prefix sequence of A. Therefore, after merging, they can be uniformly replaced by the prefix rules corresponding to A.

[0085] Repeat the above process to select all merging candidates that meet the conditions from the current rule set.

[0086] This application searches upwards from any leaf node in the current rule set along its parent node references to find a target ancestor node whose target EPC coverage count is greater than that of the leaf node and whose depth is less than that of the leaf node. This ensures that the target EPC coverage count of the ancestor node covers the number of target EPCs in the leaf node, and that the ancestor node is located above the leaf node. This allows the prefix rule corresponding to the ancestor node to cover a wider range of target EPCs with a shorter binary prefix sequence, in addition to covering the target EPCs covered by the rules corresponding to the leaf node. By determining that the ancestor node has at least two child nodes, and that both child nodes have corresponding rules in the current rule set, the number of rules in the rule group to be merged is at least two, ensuring that the merging operation can... Reduce the number of rules; by determining that the target EPC coverage count of the target ancestor node is greater than or equal to the sum of the target EPC coverage counts of the rules corresponding to the at least two child nodes, the number of target EPCs covered by the prefix rules corresponding to the merged target ancestor node can cover the sum of the target EPC coverage counts of all rules in the rule group to be merged, ensuring that the merging operation will not cause the target EPC to be missed; by using the rule group to be merged, which is composed of the rules corresponding to the target ancestor node and the at least two child nodes, as a merge candidate, the generation of the merge candidate simultaneously satisfies the triple constraints that the ancestor node depth is less than the leaf node, all child nodes have corresponding rules in the rule set, and the target coverage number meets the coverage condition, ensuring that each generated merge candidate has executability and target coverage integrity.

[0087] Under the constraints of the aforementioned hard constraints and optimization objectives, the process of selecting merging candidates (S210), calculating the comprehensive reward value (S220), determining target merging candidates, and performing the merging operation (S230) is repeated. After each merging, it is verified whether the current rule set satisfies the first hard constraint, that is, whether the new rule set obtained after merging still covers all target EPCs; at the same time, if the merged rule set does not satisfy the second hard constraint, the merging is stopped. When the termination condition is met, the current rule set is output as the filtering rule set.

[0088] In one embodiment, after performing the merge operation, the method further includes: In the mapping table, the prefix rules newly added for the target merging candidate are mapped to the corresponding ancestor nodes, and the mapping relationships corresponding to the deleted rule groups to be merged are removed; The mapping table is used to maintain the mapping relationship between rules and nodes in the enhanced prefix tree.

[0089] This application maps newly added prefix rules for target merging candidates to their corresponding ancestor nodes in a mapping table, and removes the mapping relationships corresponding to deleted rule groups to be merged. This ensures that each rule in the mapping table has a unique correspondence with a node in the enhanced prefix tree. When subsequent rules change, the enhanced prefix tree node corresponding to the rule can be directly located through the mapping table without having to retrace the enhanced prefix tree to find the node position, reducing the node location overhead when rules change. By maintaining the mapping relationship between rules and nodes in the enhanced prefix tree, the correspondence between rules and nodes remains synchronized after the merge operation is executed, avoiding the inability to locate the correct node position in subsequent diagnostic or fine-tuning operations due to inconsistent mapping relationships.

[0090] In one embodiment, the termination condition includes either the number of rules in the current rule set being less than or equal to the upper limit of the number of rules in the device, or no merge candidate being selected.

[0091] This application stops merging when the number of rules in the current rule set is less than or equal to the upper limit of the number of rules on the device, ensuring that the output filtering rule set meets the maximum number of rules that the RFID reader device can handle, thus preventing the final rule set from being undeployable due to exceeding device limitations. By setting the termination condition to stop merging when no merging candidates can be selected, the merging process automatically stops when it is impossible to further simplify the rules through merging operations, avoiding infinite loops in the iteration process, and ensuring that the final output filtering rule set reaches the most concise state under the current constraints.

[0092] The iterative merging in this application adopts a greedy rule merging, referring to... Figure 5 As shown, the overall iterative optimization process is as follows: S1. Input the initial rule set The initial rule set is a full-match rule set, with each rule corresponding to a complete EPC. At the same time, a mapping table between rules and enhanced prefix tree nodes is constructed, and the maximum number of rules for the device is set. .

[0093] S2. Generate a candidate set for merging based on the nearest effective ancestor mechanism. Specifically, for the current rule set Traverse upwards along the parent node references of any leaf node in the rule set, and take the first ancestor node whose target EPC coverage count is greater than that of the leaf node and whose depth is less than that of the leaf node as the target ancestor node. If the ancestor node has at least two child nodes and both of the child nodes have corresponding rules in the current rule set, and the target EPC coverage count of the ancestor node is greater than or equal to the sum of the target EPC coverage counts covered by the rules corresponding to the at least two child nodes, then the rule group to be merged, consisting of the ancestor node and the rules corresponding to the at least two child nodes, is taken as a merge candidate. All selected merger candidates constitute the aforementioned merger candidate set. .

[0094] S3, Judgment If the value is empty, it indicates that there are no candidates to merge, and the current rule set is output. If not, proceed to step S4.

[0095] S4. For each merge candidate Calculate the overall reward value .

[0096] S5. Select the candidate with the highest comprehensive reward value as the target merging candidate. The target merging candidate is denoted as... , This represents the ancestor node corresponding to the target merging candidate. This indicates the group of rules to be merged corresponding to the target merge candidate. This represents the independent variable that corresponds to the maximum value.

[0097] S6. Performing the merge operation includes: from the current rule set Delete the merge rule group corresponding to the target merge candidate. And added Corresponding prefix rules A new set of rules is obtained. This rule set will serve as the current rule set for the next iteration. . This represents a single-element set consisting of prefix rules from newly added ancestor nodes. Understandably, the first iteration uses the initial rule set. As the current rule set . The difference operation represents the difference from the current rule set. Delete the rule group to be merged The remaining set of rules.

[0098] The effects of merging include a reduction in the number of rules and an increase in the potential for incorrect rule overwriting. After performing a merge operation, the new rule set is equal to deleting all rules from the rule group to be merged from the current rule set and then adding the prefix rules corresponding to the ancestor nodes. .in, This represents the number of rules in the rule group to be merged. The number of rules in the new rule set. The rule groups to be merged. It may contain multiple rules, and the newly added ones There is only one rule, therefore the net number of rules is reduced. strip.

[0099] The potential increase in false coverage refers to the following: after the merge operation, the number of non-target EPCs covered by the newly added ancestor node prefix rule may be greater than or equal to the total number of non-target EPCs covered by the rule group to be merged before the merge. This is because, before the merge operation, multiple rules in the rule group to be merged each covered a portion of the non-target EPCs. These rules correspond to nodes with greater depth and longer prefixes, resulting in a relatively narrow coverage for each rule. After the merge, these multiple rules are replaced by the prefix rule corresponding to an ancestor node. This ancestor node has a smaller depth and shorter prefix, and its coverage is wider than that of any rule in the rule group to be merged. It not only covers all the non-target EPCs originally covered by the rule group to be merged but may also cover additional non-target EPCs. This can be expressed by the formula: , This represents the set of non-target EPCs covered by the newly added ancestor node prefix rule. This represents the union of all non-target EPCs covered by rules in the rule group to be merged. This is precisely the role of the second term, the non-target coverage cost penalty term, in the total reward formula of this application. This term takes a negative value; the more the false coverage increases, the smaller the total reward value, thereby guiding the algorithm to avoid merging operations with excessive false coverage.

[0100] S7. Update the mapping table between the rules and the enhanced prefix tree nodes.

[0101] S8. Determine if the termination conditions are met. Termination conditions include the current rule set (i.e., the new rule set obtained by performing the merge operation). The number of rules in the system is less than or equal to the maximum number of rules for the device. ) or there are no valid merge candidates in the current rule set (i.e. (Empty). If not satisfied, repeat the above process of candidate generation, reward calculation, and merging (i.e., S2-S7) until the termination condition is met. At this point, stop merging and output the final set of filtering rules. This rule set satisfies both the global constraints (i.e., the first and second hard constraints mentioned above) and the optimization objective.

[0102] The rule generation layer also includes a rule post-processing and device adaptation module. This module converts the iteratively merged filtering rule set into an instruction format that the target RFID reader can directly parse and execute. Specifically, for each rule in the filtering rule set, the corresponding binary prefix is ​​converted to hexadecimal representation, and parameters such as EPC storage area identifier, matching start address, and matching length are filled in according to the Select command format compatible with the target RFID reader to generate a deployable rule configuration file. The matching length is adapted to an integer multiple of 4 to meet the instruction format requirements of the target RFID reader. The generated rule configuration file is then sent to the target RFID reader, enabling the RFID reader to load and execute the filtering rule set.

[0103] Based on the above system, the method of this application further includes: converting the filtering rule set generated by iterative merging into an instruction format that can be directly parsed and executed by the target RFID reader, for details of which can be found in the description of the rule post-processing and device adaptation module above.

[0104] The autonomous management layer also includes a performance baseline management module. The starting point of the autonomous closed loop is establishing an objective performance reference system. This module is configured to establish an objective reference standard for measuring the health of the system after deployment and stable operation. A set of key performance indicators (KPIs) sensitive to system performance are selected to form a baseline vector. These KPIs include any one of the following: inventory rate, target EPC read rate, non-target EPC misread rate, and average rule hit rate. Inventory refers to the RFID reader performing an inventory operation using the currently deployed filtering rule set, identifying target tags (items to be identified) from a large number of tags while filtering out interfering tags (non-target tags). The inventory rate is the total number of tags successfully read by the RFID reader per unit time when using the filtering rule set for inventory, i.e., the total number of successfully inventoryed tags (obtained by adding the number of target tags and the number of non-target tags read) divided by the total task time. The target EPC read rate is the proportion of target EPCs successfully inventoryed to the total number of target EPCs that should be inventoryed when using the current filtering rule set for inventory. The non-target EPC misread rate is the proportion of non-target EPCs mistakenly recorded during inventory when applying the filtering rule set; that is, the ratio of the number of non-target EPCs covered by the current rule set to the total number of EPCs in the non-target EPC set. The rule average hit rate is the average frequency at which each filtering rule is successfully matched during the inventory process; that is, the ratio of the sum of the hit counts of all rules in the filtering rule set to the total number of rules when applying the filtering rule set for inventory.

[0105] The baseline establishment process is as follows: The system enters baseline learning mode and continuously monitors and records KPI data for M normal inventory cycles. Statistical analysis is then performed on the collected KPI data. Taking the non-target EPC misread rate as an example, let the... The observed values ​​for each inventory cycle are Baseline mean and standard deviation The calculation is as follows: ; ; in, This represents the number of inventory cycles.

[0106] The degradation trigger threshold Threshold is set as: Threshold = μ + 3σ. The average vector of each KPI and the corresponding threshold data are encrypted, signed, and stored in the health baseline library.

[0107] This application also provides a baseline reset mechanism, which means that when a significant change occurs in the external environment (such as the replanning of the reservoir area or the replacement of the target EPC set leading to rule reconstruction), this module can be triggered manually or automatically to re-execute the above baseline establishment process.

[0108] In one embodiment, the real-time inventory performance monitoring module continuously monitors the system's performance status with low computational overhead during system operation and performs unified event generation processing for various trigger sources. After the baseline is established, the system enters a continuous online monitoring state. A monitoring agent is deployed in the embedded firmware of the RFID reader or in the edge computing node connected to it. After each inventory task is completed, the monitoring agent uses the inventory result statistics reported by the RFID reader to calculate the key performance indicators for the current inventory cycle. The inventory result statistics include the number of target tags read, the number of non-target tags read, and the total task time. This calculation process is designed with low computational complexity to avoid affecting the real-time performance of the inventory operation.

[0109] The real-time inventory performance monitoring module is responsible for generating trigger signals that initiate the autonomous diagnostic process, referring to... Figure 6 As shown, the event sources include two categories: health degradation events and insufficient target coverage events. The performance monitoring data includes key performance indicators (KPIs), which are quantitative indicators reflecting the operating performance of the RFID reader / writer; the coverage of the filtering rule set includes the target EPCs covered by the filtering rule set.

[0110] The step of generating a trigger signal based on performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set includes: If the degradation of the key performance indicators exceeds the corresponding degradation trigger threshold for multiple consecutive inventory cycles, a health degradation event is generated. The coverage rate is determined based on the target EPCs covered by the filtering rule set and the current target EPC set in the EPC database. If the coverage rate is lower than a first preset threshold, a target coverage insufficiency event is generated.

[0111] Specifically, it determines whether the degradation level of the key performance indicators (KPIs) in the current inventory cycle exceeds the corresponding degradation trigger threshold. For each KPI, a corresponding degradation trigger threshold is preset. If the degradation level of the KPIs in multiple consecutive inventory cycles exceeds the corresponding degradation trigger threshold, a health degradation event is generated. The degradation level refers to the degree of difference between the KPI value of the current inventory cycle and the historical baseline value.

[0112] The target insufficiency event is also called the target set change event. When the target EPC set in the EPC database changes, if the upper-layer business system has the ability to actively notify, the EPC database will directly send a change notification through a preset interface; this is the first triggering method. If the upper-layer business system does not have the ability to actively notify, the real-time inventory performance monitoring module will compare the EPC set covered by the current filtering rule set with the current target EPC set in the EPC database after each inventory. If the coverage rate is significantly lower than 100%, it is determined that the target EPC set has changed; this is the second triggering method.

[0113] Specifically, coverage is the ratio of the number of target EPCs covered by the filtering rule set to the total number of EPCs in the current target EPC set in the EPC database. The target EPCs covered by the filtering rule set are those that can be successfully identified and read by the current filtering rule set. The current target EPC set in the EPC database contains all the target EPCs corresponding to the items to be identified. The coverage is obtained by dividing the number of target EPCs covered by the filtering rule set by the total number of EPCs in the current target EPC set in the EPC database. .

[0114] in, This represents the set of target EPCs covered by the filtering rule set. This represents the current target EPC set in the EPC database. If a change to the target EPC set is detected through any of the above triggering methods, and the coverage rate falls below a first preset threshold, a target coverage insufficiency event is generated.

[0115] A trigger signal is formed by either or both of the following events: health degradation event and insufficient target coverage event. The trigger signal indicates that the current set of filtering rules is not performing correctly and a diagnostic and remediation process needs to be initiated.

[0116] This application generates a health degradation event when the degradation level of key performance indicators (KPIs) exceeds the corresponding degradation trigger threshold in multiple consecutive inventory cycles. The comparison between the real-time monitoring value of the KPI and the preset degradation trigger threshold directly determines the triggering of the health degradation event, thus achieving automatic identification of RFID reader performance degradation. Furthermore, by determining the coverage rate based on the target EPCs covered by the filtering rule set and the current target EPC set in the EPC database, a target coverage insufficiency event is generated when the coverage rate is lower than a first preset threshold. This allows the actual coverage of the target EPCs by the filtering rule set to be precisely quantified through the coverage rate value. The comparison between the coverage rate and the first preset threshold directly determines the triggering of the target coverage insufficiency event, achieving automatic identification of rule coverage insufficiency. The health degradation event and the target coverage insufficiency event together constitute a trigger signal, enabling the trigger signal to reflect two different types of anomalies: performance degradation and coverage insufficiency, providing event type information for subsequent diagnostic positioning.

[0117] In one embodiment, the repair strategy includes a full rule reconstruction strategy and a local fine-tuning strategy; refer to Figure 6 As shown, the problem diagnosis module starts upon receiving a trigger signal and locates the root cause of the problem in a phased manner. The problem diagnosis module includes an environmental change diagnosis submodule and a rule efficiency audit submodule. The first phase is environmental change diagnosis. The generation of a repair strategy based on the trigger signal includes: If the trigger signal is a target coverage insufficiency event, a full rule reconstruction strategy is generated. If the trigger signal is a health degradation event, collect EPC samples in the current environment to form a current environment EPC snapshot set, determine the similarity based on the current environment EPC snapshot set and the historical interference EPC set, and generate a full rule reconstruction strategy if the similarity is lower than the similarity threshold. If the similarity is not lower than the similarity threshold, then the rules in the set of filtering rules that meet the preset degradation conditions are marked as degradation rules, and a full rule reconstruction strategy or a local fine-tuning strategy is generated based on the number of degradation rules and the cause of degradation.

[0118] Specifically, the environmental change diagnosis submodule first determines the event type. If the trigger signal is a target coverage insufficiency event (i.e., a target set change event), it indicates that the root cause of the problem is a deterministic change in the target EPC set on the business side. The problem diagnosis module directly outputs a diagnostic conclusion suggesting a full rule reconstruction and terminates the subsequent diagnosis stages.

[0119] If the trigger signal is a health degradation event, environmental snapshot collection is initiated. The problem diagnosis module sends a temporary command to the RFID reader, causing it to enter a short-term listening mode. In this mode, the RFID reader indiscriminately collects EPC samples within the current electromagnetic field, forming a current environmental EPC snapshot set. .

[0120] Environment similarity comparison: compare the current environment EPC snapshot set With the historical interference EPC set recorded in the interference management library Perform a comparison and calculate the set similarity between the two. The Jaccard similarity coefficient is used, i.e.: ; in, This represents the size of the intersection between the current environment EPC snapshot set and the historical interference EPC set, i.e., the number of EPCs that appear in both sets simultaneously. This represents the size of the union of the current environment EPC snapshot set and the historical interference EPC set, i.e., the total number of all unique EPCs in the two sets. The similarity value ranges from [0,1]. The closer the value is to 1, the more similar the current environment is to the historical interference environment; the closer the value is to 0, the greater the change in the current environment relative to the historical state.

[0121] The system outputs a diagnostic conclusion. If the calculated similarity is lower than a preset similarity threshold, it determines that the current environment has undergone a drastic change relative to the historical state, and the historical data has low reference value. The problem diagnosis module then outputs a diagnostic conclusion suggesting full rule reconstruction and terminates the subsequent diagnostic stages. If the similarity is not lower than the similarity threshold, it indicates that the current environment is basically consistent with the historical interference environment, and the root cause of the problem may lie in the degradation of a specific rule itself, thus entering the second stage of diagnosis.

[0122] If the first stage determines that the environment has not changed drastically, it indicates that the performance degradation is more likely due to the inefficiency of the rules themselves, and at this point, a more refined second stage of diagnosis is initiated.

[0123] The rule efficiency auditing submodule extracts each rule from the currently running filtering rule set from the enhanced prefix tree. Corresponding node of and The attribute value, combined with the recent hit count of the rule recorded by the real-time inventory performance monitoring module.

[0124] Calculate the interference introduction ratio for each rule. and target contribution The interference introduction ratio and the target contribution are determined as follows: the target EPC coverage count and non-target EPC coverage count corresponding to each rule in the filtering rule set are extracted from the enhancement prefix tree; the interference introduction ratio is determined based on the non-target EPC coverage count corresponding to the rule and the current total non-target EPC misreading; and the target contribution is determined based on the target EPC coverage count corresponding to the rule and the total number of target EPCs.

[0125] ; ; in, For rules The corresponding non-target EPC coverage count value, For rules The corresponding target EPC coverage count value, This represents the current total non-target error reading. The target is the total number of EPCs. Rules that meet preset degradation conditions are marked as degradation rules. The preset degradation conditions include any one of the following: 1. If the interference introduction ratio exceeds a preset multiple (e.g., 2 times) of the average interference introduction ratio of other rules, it indicates that the interference introduction ratio is too large; 2. If the target contribution rate is lower than the preset percentage (e.g., 5%), it indicates that the target contribution rate is too low. 3. A consistently zero hit rate for rules in recent inventory counts indicates an excessively low recent hit rate. "Recent" refers to the consecutive inventory counts preceding the current moment.

[0126] The rule efficiency audit submodule outputs diagnostic conclusions, specifically generating a rule efficiency audit report. The report clearly identifies the rules marked as "degraded" and their quantitative basis. Simultaneously, based on the number of degraded rules and the reasons for the degradation, it outputs tiered diagnostic conclusions. If the number of degraded rules exceeds a preset threshold (e.g., exceeding 30% of the total number of rules), or if the degradation points to multiple unrelated interference sources, it outputs "Recommend full rule reconstruction." If only individual rules are degraded, and the increase in their interference introduction ratio can be traced back to changes in the activity of a specific EPC in the interference management library, indicating a localized problem, it outputs "Recommend local fine-tuning," along with a list of interference EPCs to be adjusted.

[0127] This application uses a preset degradation condition, where the interference introduction ratio exceeds the average interference introduction ratio of other rules by a preset multiple. This marks rules with significantly higher interference introduction ratios than the average as degraded rules, meaning rules that contribute abnormally to false coverage can be automatically identified. Similarly, it uses a preset degradation condition, where the target contribution is lower than a preset ratio. This marks rules that contribute insufficiently to target coverage. Furthermore, it uses a preset degradation condition, where the hit rate of a rule in recent inventory remains zero. This marks rules that have not been hit for a long time as degraded rules, meaning rules that have lost their practical function in the current environment can be automatically identified. When any of the above preset degradation conditions is met, the corresponding rule is marked as a degraded rule. This allows the identification of degraded rules to simultaneously cover three different degradation types: abnormal false coverage, insufficient target coverage, and a persistently zero hit rate. This provides a quantitative basis for subsequent generation of full rule reconstruction strategies or local fine-tuning strategies based on the number and causes of degraded rules.

[0128] The repair execution module is configured to select and execute the most suitable repair strategy based on the diagnostic conclusions output by the problem diagnosis module.

[0129] In one embodiment, reference is made to Figure 6 As shown, if the diagnostic conclusion is local fine-tuning, then the local fine-tuning strategy is executed. The local fine-tuning strategy includes: updating the activity weight of a specific interference EPC in the interference management library and adjusting any item in the list category to which the specific interference EPC belongs; Wherein, the interference EPC is the EPC that causes the increase in the interference introduction ratio of the degradation rule.

[0130] Specifically, the node corresponding to the degradation rule is located in the enhancement prefix tree. Starting from this node, the process traces downwards layer by layer along the child node branch with the fastest growth in non-target EPC coverage count until a leaf node is reached. The EPC corresponding to this leaf node is the interfering EPC that caused the rule's interference introduction ratio to increase. After identifying the interfering EPC, its activity weight is updated in the interference management library, or its list category is adjusted (e.g., moving the interfering EPC from the observation list to the confirmed interference list). The activity weight is used to quantify the impact of the interfering EPC in the current environment. During full rule reconstruction, interfering EPCs with activity weights greater than or equal to a preset weight threshold are selected from the interference management library and used as a set of non-target EPCs in rule generation.

[0131] In one embodiment, if the diagnostic conclusion is a drastic environmental change, or a full rule reconstruction is recommended, or the trigger signal is an event of insufficient target coverage, then a full rule reconstruction strategy is executed. The full rule reconstruction strategy includes: obtaining the current target EPC set and the current non-target EPC set, and regenerating a new set of filtering rules based on the current target EPC set and the current non-target EPC set.

[0132] Specifically, the repair execution module prepares data by retrieving the latest target EPC set from the EPC database (i.e., the aforementioned current target EPC set) and filtering out interfering EPCs with activity weights greater than or equal to the preset weight threshold from the interference management library as the current non-target EPC set. If it is the first deployment or a reconstruction after a change in the target EPC set, and the interference management library is empty or needs to be reset, a full inventory of non-target EPC data in the current environment is performed. The repair execution module packages the latest target EPC set and the filtered current non-target EPC set as input data for the rule generation layer, calls the rule generation layer, and executes a complete rule generation and iterative merging process based on the current target EPC set, the current non-target EPC set, and the device's maximum rule count constraint parameter to generate a new set of filtering rules. After the repair execution module executes the full rule reconstruction strategy, the verification update module verifies the new filtering rule set. Once the verification is successful, the new filtering rule set is deployed to the RFID reader.

[0133] This application refines the management of interference EPCs into two granularities: adjusting the weight of specific interference EPCs and adjusting the category of the list when executing a local fine-tuning strategy. This provides two optional operation methods for the dynamic updating of the interference management library. By defining interference EPCs as EPCs that cause an increase in the interference introduction ratio of deteriorating rules, the adjustment target of local fine-tuning is directly located on the specific interference EPCs that cause an increase in false coverage, avoiding overall rule adjustments that may affect the normal coverage of other rules. By obtaining the current target EPC set and the current non-target EPC set when executing a full rule reconstruction strategy, and regenerating a new set of filtering rules based on the current target EPC set and the current non-target EPC set, the input data for full reconstruction includes the latest target EPC set and non-target EPC set, ensuring that the regenerated set of filtering rules can adapt to changes in the current environment.

[0134] In one embodiment, reference is made to Figure 6 As shown, after the repair operation is completed, the verification and update module performs security verification on the new filtering rule set. Upon successful verification, it executes the complete security update process from verification to deployment. Updating the RFID reader's filtering rule set according to the repair strategy includes: S1401, Control the RFID reader to perform a screening task using the currently effective screening rule set and receive tag response signals; S1402. The new set of filtering rules obtained after executing the repair strategy is sent to the buffer area of ​​the RFID reader. S1403. Perform a simulated match between the tag response signal and the new filtering rule set to obtain a simulated match result; S1404. If the simulated matching result indicates that the new filtering rule set meets the verification conditions, then send an atomic switching command to the RFID reader during the interval between the two inventory tasks to switch the currently effective filtering rule set to the new filtering rule set.

[0135] Specifically, the RFID reader is controlled to use the currently effective set of filtering rules (referred to as the old set of filtering rules). The RFID reader executes a filtering task and receives tag response signals. The currently effective filtering rule set refers to the set of rules that the RFID reader is currently using to execute the inventory filtering task. The new filtering rule set obtained after implementing the repair strategy is sent to the RFID reader's buffer area, putting the new filtering rule set into shadow mode. In shadow mode, the RFID reader still uses the currently effective filtering rule set to execute the actual filtering task, but simultaneously performs simulated matching of the received tag response signals with the new filtering rule set in the background. Simulated matching of the received tag response signals with the new filtering rule set in the background means that in each inventory task, the RFID reader receives a response signal returned by the tag, which carries the tag's EPC data. The high-order bit sequence of the EPC in each received tag response signal is compared one by one with the prefix corresponding to each rule in the new filtering rule set. If the high-order bit sequence of the EPC is completely consistent with the prefix corresponding to a certain rule, then the EPC is determined to be matched by that rule. After traversing all rules in the new filtering rule set, the set of target EPCs matched by the new filtering rule set is counted, obtaining the simulated matching result.

[0136] The system collects and compares the simulated matching results of the old filtering rule set (i.e., the currently effective filtering rule set) and the new filtering rule set. If the simulated matching result indicates that the new filtering rule set meets the verification conditions, the system determines that the new filtering rule set is authorized to go live. If the new filtering rule set meets the verification conditions, it is authorized to go live; if it does not meet the verification conditions, the new filtering rule set is not deployed, and an alarm message is generated.

[0137] After successful verification, the verification update module sends an atomic switching command to the RFID reader during the interval between two inventory tasks. The RFID reader responds to the command, switching the currently effective old filtering rule set to the new filtering rule set in the cache area. The entire process is transparent to upper-layer business applications, achieving seamless hot updates without business interruption.

[0138] This autonomous action, starting from the triggering of an anomaly, proceeds through diagnosis, decision-making, remediation, and verification updates. The entire lifecycle of operational data is integrated into a structured, timestamped, and digitally signed self-healing event report. This report includes at least: the triggering event type and data, intermediate data and conclusions during the diagnosis process, the basis for the decision, copies of the old and new filtering rule sets, and the results of the shadow run verification. This report is securely stored in the autonomous event log repository within the data and infrastructure layer as technical evidence of the system's self-healing capabilities.

[0139] This application controls an RFID reader to perform screening tasks using the currently effective screening rule set and receive tag response signals. After sending a new screening rule set to the RFID reader's buffer area, it simulates a match between the tag response signal and the new screening rule set to obtain a simulated match result. This allows the verification process of the new screening rule set to be performed in the background, without affecting the screening tasks performed by the RFID reader using the currently effective screening rule set. When the simulated match result indicates that the new screening rule set meets the verification conditions, an atomic switching command is sent to the RFID reader during the interval between two inventory tasks to switch the currently effective screening rule set to the new screening rule set. This allows the rule switching operation to be completed during the inventory interval without affecting the ongoing inventory task, achieving seamless switching of inventory operations without interruption.

[0140] In one embodiment, the verification condition includes: the target EPC set matched by the new filtering rule set. With the current target EPC set in the EPC database The size of the intersection is equal to the total number of target EPCs. ,Right now This indicates that the new filtering rule set can cover all target EPCs, with a target coverage rate of 100%; and the number of rules in the new filtering rule set... No more than the maximum number of rules for the device ,Right now .

[0141] This application verifies that the intersection of the target EPC set matched by the new filtering rule set and the current target EPC set in the EPC database is equal to the total number of target EPCs, ensuring that the new filtering rule set can cover all target EPCs. It also verifies that the number of rules in the new filtering rule set does not exceed the device's rule limit, ensuring that the number of rules in the new filtering rule set complies with device hardware limitations before deployment, thus avoiding deployment failure due to exceeding the rule limit. These two verification conditions together constitute the basis for granting the new filtering rule set permission to go online. The new filtering rule set is only granted permission to go online when both conditions are met simultaneously; if either condition is not met, the new filtering rule set will not be deployed, making the deployment decision based on the joint verification results of two dimensions: full target coverage and the device's rule limit.

[0142] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call a computer program in the memory 330 to execute an EPC screening rule determination method, which may include, for example: Coverage count information is generated based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tags; Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set; A trigger signal is generated based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; A repair strategy is generated based on the trigger signal, and the filtering rule set of the RFID reader is updated according to the repair strategy.

[0143] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0144] On the other hand, embodiments of this application also provide a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the EPC screening rule determination method provided in the above embodiments, for example including: Coverage count information is generated based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tags; Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set; A trigger signal is generated based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; A repair strategy is generated based on the trigger signal, and the filtering rule set of the RFID reader is updated according to the repair strategy.

[0145] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program, the computer program being used to cause a processor to execute the EPC screening rule determination method provided in the above embodiments, for example including: Coverage count information is generated based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tags; Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set; A trigger signal is generated based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; A repair strategy is generated based on the trigger signal, and the filtering rule set of the RFID reader is updated according to the repair strategy.

[0146] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0147] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0148] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for determining Electronic Product Code (EPC) screening rules, characterized in that, include: Coverage count information is generated based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tags; Based on the coverage count information, the initial rule set is iteratively merged to generate a filtering rule set; A trigger signal is generated based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; A repair strategy is generated based on the trigger signal, and the filtering rule set of the RFID reader is updated according to the repair strategy.

2. The method for determining EPC screening rules according to claim 1, characterized in that, The coverage count information is represented by an enhanced prefix tree, where each node of the enhanced prefix tree contains a target EPC coverage count value and a non-target EPC coverage count value. The step of generating coverage count information based on the target EPC set and the non-target EPC set includes: Initialize the root node of the enhanced prefix tree; Traverse each target EPC in the target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the target EPC, and increment the target EPC coverage count value of all nodes on the path. Traverse each non-target EPC in the non-target EPC set, starting from the root node, expand the child nodes layer by layer according to the binary bit sequence of the non-target EPC, and increment the non-target EPC coverage count value of all nodes on the path.

3. The method for determining EPC screening rules according to claim 2, characterized in that, Each node of the enhanced prefix tree also contains a node depth, a binary prefix sequence, a parent node reference, and a child node dictionary; Wherein, the target EPC coverage count value corresponding to each binary prefix sequence represents the number of all target EPCs prefixed by the binary prefix sequence, and the non-target EPC coverage count value represents the number of all non-target EPCs prefixed by the binary prefix sequence.

4. The method for determining EPC screening rules according to claim 1, characterized in that, The step of iteratively merging the initial rule set based on the coverage count information to generate a filtering rule set includes: Starting with the initial rule set, repeat the following steps until the termination condition is met to obtain the filtering rule set: Select one or more merge candidates from the current rule set, each of which includes an ancestor node and a group of rules to be merged; Calculate the comprehensive reward value for each merger candidate separately; Based on the comprehensive reward value, target merging candidates are determined, and a merging operation is performed on the target merging candidates. The merging operation includes deleting the rule group to be merged from the current rule set and adding a prefix rule corresponding to the ancestor node.

5. The method for determining EPC screening rules according to claim 4, characterized in that, The comprehensive reward value is determined by a multi-dimensional reward function, which includes a rule reduction benefit item, a non-target coverage cost penalty item, and a cross-level risk penalty item. The rule reduction benefit item is determined based on the number of rules reduced by the merger operation, the non-target coverage cost penalty item is determined based on the number of non-target EPC coverages increased by the merger operation, and the cross-level risk penalty item is determined based on the depth span of the merger operation.

6. The method for determining EPC screening rules according to claim 4, characterized in that, The step of selecting one or more merge candidates from the current rule set includes: For any leaf node in the current rule set, traverse upwards along the parent node reference until a target ancestor node that satisfies the first condition is found. The first condition includes that the target EPC coverage count of the target ancestor node is greater than the target EPC coverage count of the leaf node and the depth of the target ancestor node is less than the depth of the leaf node. If the target ancestor node has at least two child nodes, and both of the at least two child nodes have corresponding rules in the current rule set, and the target EPC coverage count of the target ancestor node is greater than or equal to the sum of the target EPC coverage counts covered by the rules corresponding to the at least two child nodes, then the rule group to be merged, which is composed of the rules corresponding to the target ancestor node and the at least two child nodes, is taken as a merging candidate.

7. The method for determining EPC screening rules according to claim 4, characterized in that, The iterative merging satisfies the first hard constraint and the second hard constraint; The first hard constraint includes that at any stage of the merging process, the union of the target EPCs covered by the current rule set is equal to the target EPC set; The second hard constraint includes that the number of rules in the final output filtering rule set does not exceed the upper limit of the number of rules in the device; The optimization objective of the iterative merging includes minimizing the total number of non-target EPCs covered by the current rule set, while satisfying the first hard constraint and the second hard constraint.

8. The method for determining EPC screening rules according to claim 4, characterized in that, After performing the merge operation, the method further includes: In the mapping table, the prefix rules newly added for the target merging candidate are mapped to the corresponding ancestor nodes, and the mapping relationships corresponding to the deleted rule groups to be merged are removed; The mapping table is used to maintain the mapping relationship between rules and nodes in the enhanced prefix tree.

9. The method for determining EPC screening rules according to claim 4, characterized in that, The termination conditions include either the number of rules in the current rule set being less than or equal to the upper limit of the number of rules in the device, or no merge candidate being selected.

10. The method for determining EPC screening rules according to claim 1, characterized in that, The performance monitoring data includes key performance indicators; the coverage of the filtering rule set includes the target EPCs covered by the filtering rule set; The step of generating a trigger signal based on performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set includes: If the degradation of the key performance indicators exceeds the corresponding degradation trigger threshold for multiple consecutive inventory cycles, a health degradation event is generated. The coverage rate is determined based on the target EPCs covered by the filtering rule set and the current target EPC set in the EPC database. If the coverage rate is lower than a first preset threshold, a target coverage insufficiency event is generated.

11. The method for determining EPC screening rules according to claim 10, characterized in that, The key performance indicators include any one of the following: inventory rate, target EPC read rate, non-target EPC misread rate, and average rule hit rate.

12. The method for determining EPC screening rules according to claim 1, characterized in that, The repair strategy includes a full rule reconstruction strategy and a local fine-tuning strategy; The repair strategy generated based on the trigger signal includes: If the trigger signal is a target coverage insufficiency event, a full rule reconstruction strategy is generated. If the trigger signal is a health degradation event, collect EPC samples in the current environment to form a current environment EPC snapshot set, determine the similarity based on the current environment EPC snapshot set and the historical interference EPC set, and generate a full rule reconstruction strategy if the similarity is lower than the similarity threshold. If the similarity is not lower than the similarity threshold, then the rules in the set of filtering rules that meet the preset degradation conditions are marked as degradation rules, and a full rule reconstruction strategy or a local fine-tuning strategy is generated based on the number of degradation rules and the cause of degradation.

13. The method for determining EPC screening rules according to claim 12, characterized in that, The local fine-tuning strategy includes: updating the activity weight of a specific interference EPC in the interference management library and adjusting any one of the categories to which the specific interference EPC belongs; Wherein, the interference EPC is the EPC that causes the increase in the interference introduction ratio of the degradation rule.

14. The method for determining EPC screening rules according to claim 12, characterized in that, The full rule reconstruction strategy includes: obtaining the current target EPC set and the current non-target EPC set, and regenerating a new set of filtering rules based on the current target EPC set and the current non-target EPC set.

15. The method for determining EPC screening rules according to claim 12, characterized in that, The preset degradation conditions include any one of the following: The interference introduction ratio exceeds the preset multiple of the average interference introduction ratio of other rules; The target contribution rate is lower than the preset ratio; The rule hit rate in recent inventory counts has remained at zero. The interference introduction ratio and the target contribution are determined as follows: the target EPC coverage count and non-target EPC coverage count corresponding to each rule in the filtering rule set are extracted from the enhanced prefix tree; the interference introduction ratio is determined based on the non-target EPC coverage count corresponding to the rule and the current total non-target EPC misreading; and the target contribution is determined based on the target EPC coverage count corresponding to the rule and the total number of target EPCs.

16. The method for determining EPC screening rules according to claim 1, characterized in that, The step of updating the filtering rule set of the RFID reader according to the repair strategy includes: The RFID reader is controlled to perform a screening task using the currently effective set of screening rules and to receive tag response signals; The new set of filtering rules obtained after executing the repair strategy will be sent to the cache area of ​​the RFID reader. The tag response signal is simulated and matched with the new set of filtering rules to obtain a simulated matching result; If the simulated matching result indicates that the new filtering rule set meets the verification conditions, then an atomic switching command is sent to the RFID reader during the interval between the two inventory tasks to switch the currently effective filtering rule set to the new filtering rule set.

17. The method for determining EPC screening rules according to claim 16, characterized in that, The verification conditions include: the size of the intersection between the target EPC set matched by the new filtering rule set and the current target EPC set in the EPC database is equal to the total number of target EPCs, and the number of rules in the new filtering rule set does not exceed the upper limit of the number of device rules.

18. An EPC screening rule determination system, characterized in that, include: Rule generation layer, autonomous management layer, and infrastructure layer; The rule generation layer includes an enhanced prefix tree construction module and a merging optimization module; The autonomous management layer includes a real-time inventory performance monitoring module, a problem diagnosis module, a repair execution module, and a verification and update module; The infrastructure layer includes an RFID reader and an antenna. The RFID reader is used to read EPC data of multiple RFID tags within the coverage area through the antenna to obtain a target EPC set and a non-target EPC set, and to run a set of filtering rules. The enhanced prefix tree construction module is used to generate coverage count information based on the target EPC set and the non-target EPC set; wherein, the target EPC set is the EPC set corresponding to the item to be identified, and the non-target EPC set is the EPC set corresponding to the interference tag; The merging and optimization module is used to iteratively merge the initial rule set based on the coverage count information to generate a filtering rule set; The real-time inventory performance monitoring module is used to monitor the performance monitoring data of the RFID reader when running the filtering rule set and the coverage of the filtering rule set; The problem diagnosis module is used to generate a trigger signal based on the performance monitoring data of the RFID reader running the filtering rule set and the coverage of the filtering rule set; and to generate a repair strategy based on the trigger signal. The repair execution module is used to execute the repair strategy; The verification update module is used to update the filtering rule set of the RFID reader according to the repair strategy after the repair execution module executes the repair strategy.

19. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the EPC screening rule determination method according to any one of claims 1 to 17.

20. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the EPC screening rule determination method according to any one of claims 1 to 17.

Images