An algorithm for store goods display taking and placing management based on passive RFID technology

Abstract

The application relates to the field of retail digitization management and asset supervision, and discloses a store goods display taking and placing management algorithm based on passive RFID technology, which comprises the following steps: acquiring radio frequency characteristic flow of a target asset and a neighborhood reference label; extracting neighborhood group feature distribution and mapping a suppression operator to offset field environment noise; stripping common mode disturbance in a phase evolution sequence according to the suppression operator to extract a single body space offset; and mapping the offset to a business state machine to determine asset interaction confidence. The application realizes deep decoupling of radio frequency physical fluctuation and commercial semantic state, filters out transient physical noise through a neighborhood beacon interference suppression mechanism, ensures data consistency in asset display and transfer migration, and improves asset mapping precision of a management system without adding sensing hardware.

Description

Technical Field

[0001] This invention relates to an algorithm for managing the display and retrieval of goods in stores based on passive RFID technology, belonging to the field of retail digital management and asset supervision technology. Background Technology

[0002] Current management systems typically use physical indicators such as the success rate of reading passive tags and the strength of received signals within a specific monitoring area to characterize the display status of goods. When the relevant physical indicators fluctuate or the signal disappears, the system automatically updates the goods location information in the store management database to support management functions such as replenishment reminders, display layout optimization, and business forecasting. The carrier phase reflected by passive RFID tags and the strength of received signals are constrained by both physical displacement and the channel environment. The displacement of objects causes the characteristic evolution of the phase sequence and the shift of the signal energy envelope. This objective physical relationship constitutes the data foundation of existing radio frequency sensing technology.

[0003] This logic, which directly links underlying physical parameters to high-level business states, produces data distortion in complex retail environments. Human occlusion, multipath reflections from metal shelves, and electromagnetic interference in real-world store environments cause transient jitter in radio frequency signals. Besides limitations in hardware deployment, the data processing strategies at the logic level are ill-suited to high-frequency interactive business scenarios. For example, Chinese invention patent application CN121036795A discloses a data reading method and system based on an RFID reader, which relies on analyzing the geometric characteristics of the signal phase trajectory curvature to estimate relative motion speed and switching modulation and coding methods to maintain communication reliability. The solution assumes that the signal phase drift is unidirectional and continuous. In actual retail store operations, customers frequently pick up and tumble, causing nonlinear and local transient common-mode disturbances. Based on the single-entity trajectory curvature recognition mechanism, it is easy to misjudge environmental noise as business displacement in scenarios where group background noise and business actions overlap. The physical layer features and business semantic state produce logical mismatch. In order to maintain the apparent stability of data feedback, the industry usually sets debouncing time or relaxes the misjudgment threshold in the management logic. Although such practices reduce the false alarm frequency, they cause a lag in management data feedback, resulting in the inventory system being unable to accurately capture the high-frequency real business picking actions.

[0004] Therefore, the technical problem to be solved by this invention is how to utilize the carrier phase evolution sequence collected in the existing system to build a business discrimination mechanism at the logic layer, realize the assessment of the confidence of business actions in complex environments, eliminate the impact of physical layer signal noise on the consistency of business management data, and establish a business logic processing framework that can automatically correct sensor deviations and support management decisions. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An algorithm for managing the display and retrieval of store merchandise based on passive RFID technology, comprising the following steps: Step S101: Obtain the radio frequency feature stream of the target tag corresponding to the managed target within a preset sampling window of the management field. The radio frequency feature stream includes phase evolution sequence and signal strength distribution data. Step S102: Identify a set of reference tags that are physically adjacent to the target tag within the display area, and extract the neighborhood group feature distribution corresponding to the reference tag set to characterize the background environmental noise benchmark of the management area. Step S103: Based on the background environmental noise reference mapping, a suppression operator characterizing the field disturbance intensity is generated. The suppression operator is aggregated from the phase dispersion weights of each label in the reference label set, and is used to lock the common physical disturbance characteristics of the managed field. Step S104: Perform differential decoupling calculation on the radio frequency characteristic flow and the suppression operator, remove the common mode disturbance component in the phase evolution sequence, and generate the individual spatial offset of the managed target relative to the display carrying area. Step S105: Map the spatial offset of the single unit to the preset business state machine, calculate the business confidence of the managed target in the management process from static display to dynamic flow. The business state machine is built based on the business logic flow of asset interaction, and is used to achieve accurate identification of real picking and placing behavior in store retail management business through logical extraction of physical features.

[0006] Step S106: When the business confidence level meets the preset judgment threshold, it is confirmed that the managed target has undergone business state migration, and the asset attribute database of the associated management system is synchronized according to the output result of the business state machine, so as to ensure the consistency between the inventory data and the physical display status in the store asset management system at the management level through the verification of the business logic layer.

[0007] Preferably, step S101 includes the following sub-steps: obtaining the original phase values ​​corresponding to multiple discrete sampling points of the target tag within the preset sampling window; performing smoothing filtering and unwinding processing on the original phase values ​​to eliminate phase jumps caused by abrupt changes in the carrier reflection path length of the target tag; constructing a phase evolution sequence characterizing the displacement continuity of the managed target in the physical space dimension by combining the sampling period of the preset sampling window; and simultaneously extracting the received signal strength corresponding to the discrete sampling points to form the signal strength distribution data reflecting the energy fluctuation characteristics of the target tag.

[0008] Preferably, step S102 specifically includes the following steps: querying the field configuration image table, extracting a set of candidate tags that have the same hierarchical attributes as the target tag and whose antenna coverage areas overlap; monitoring the sampling frequency consistency of each tag in the candidate tag set within the preset sampling window, and determining the tags whose sampling frequency consistency is higher than a preset threshold as the reference tag set.

[0009] Preferably, step S104 includes the following sub-steps: convolving the phase evolution sequence with the suppression operator to denoise, so as to offset the collective phase shift caused by the obstruction of people in the management area; and extracting the phase change slope and frequency center shift in the individual spatial shift based on the denoised result, so as to characterize the degree of physical stripping of the target label relative to the display area.

[0010] Preferably, step S105 specifically includes: inputting the individual spatial offset into the business state machine, using convolution kernels to extract the feature correlation between the individual spatial offset and the preset standard interaction behavior template; and calculating the business confidence level by combining the asset attribute parameters corresponding to the managed target and the business weight corresponding to the management field.

[0011] Preferably, after step S106, the following steps are also included: after confirming that the managed target has undergone a business status migration, retrieving the inventory status change log associated with the target tag; when the inventory status change log shows that the managed target has entered a preset delivery area or trial area, verifying the validity of the business status migration through the business status machine feedback.

[0012] Preferably, step S106 specifically includes: encapsulating the status synchronization request of the asset attribute database using a preset asymmetric encryption algorithm; and submitting the encapsulated status synchronization request to the inventory logic processing unit of the management system to update the physical status field of the asset in the asset attribute database.

[0013] Preferably, the method further includes the following steps: real-time monitoring of the dynamic evolution trend of the suppression operator within the preset sampling window; when the suppression operator continuously exceeds the preset abnormal fluctuation threshold and covers all tags in the reference tag set, determining that there is systematic electromagnetic interference in the current management field, and outputting a management strategy adjustment signal.

[0014] Preferably, the method further includes the following steps: after the asset attribute database completes state synchronization, a sampling frequency degradation strategy is initiated for the target tag; when the managed target re-enters the display area and triggers phase regression determination, the sampling frequency of the target tag is restored.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the management of merchandise display and retrieval in stores, by establishing a business confidence assessment mechanism, a deep decoupling of physical layer radio frequency fluctuations and business semantic states is achieved. This effectively eliminates the interference of environmental noise in the retail environment on the authenticity of management data. Traditional management systems rely excessively on the linear mapping logic between RFID tag reading status and asset business status, leading to the erroneous judgment of asset displacement due to the disappearance of transient signals caused by human body obstruction, metal reflection, or random flipping. This invention utilizes carrier phase evolution gradient and signal strength fluctuation characteristics to construct inertial constraints for business state migration. When physical characteristics undergo drastic changes, a business undetermined domain is introduced for verification, enabling the management system to identify and eliminate transient physical noise that does not conform to the logic of business behavior, ensuring that the output inventory change instructions accurately correspond to the actual business retrieval behavior.

[0016] 2. Introducing neighborhood fluctuation suppression logic: Utilizing the group correlation of tag signals within the display space, this solution enhances the reliability of asset status determination using existing processors in the system without introducing additional sensing hardware. Based on the objective law that non-displacement flipping caused by customers selecting goods will lead to group disturbances in the local radio frequency field, the signal fluctuation of adjacent display tags is used as a suppression factor in the determination rule. The system only confirms the occurrence of commercial interaction when the physical characteristics of the target tag evolve and exhibit spatial separation characteristics from the surrounding environment. This multi-feature cross-validation mechanism enables the system to distinguish between group environmental background noise and individual commercial actions, solving the problem of management data distortion caused by physical artifacts in high-traffic scenarios. Attached Figure Description

[0017] Figure 1 This is a flowchart of the goods management algorithm based on radio frequency feature decoupling and suppression operator compensation of the present invention; Figure 2 This is the core of the business confidence quantification assessment and the three-level state machine transition logic diagram of this invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] An algorithm for managing the display and retrieval of merchandise in stores based on passive RFID technology includes the following steps: Step S101: Obtain the radio frequency feature stream of the target tag corresponding to the managed target within a preset sampling window of the management field. The radio frequency feature stream includes phase evolution sequence and signal strength distribution data. Step S102: Identify a set of reference tags that are physically adjacent to the target tag within the display area, and extract the neighborhood group feature distribution corresponding to the reference tag set to characterize the background environmental noise benchmark of the management area. Step S103: Based on the background environmental noise reference mapping, a suppression operator characterizing the field disturbance intensity is generated. The suppression operator is aggregated from the phase dispersion weights of each label in the reference label set, and is used to lock the common physical disturbance characteristics of the managed field. Step S104: Perform differential decoupling calculation on the radio frequency characteristic flow and the suppression operator, remove the common mode disturbance component in the phase evolution sequence, and generate the individual spatial offset of the managed target relative to the display carrying area. Step S105: Map the spatial offset of the single unit to the preset business state machine, calculate the business confidence of the managed target in the management process from static display to dynamic flow. The business state machine is built based on the business logic flow of asset interaction, and is used to achieve accurate identification of real picking and placing behavior in store retail management business through logical extraction of physical features.

[0021] Step S106: When the business confidence level meets the preset judgment threshold, it is confirmed that the managed target has undergone business state migration, and the asset attribute database of the associated management system is synchronized according to the output result of the business state machine, so as to ensure the consistency between the inventory data and the physical display status in the store asset management system at the management level through the verification of the business logic layer.

[0022] Preferably, step S101 includes the following sub-steps: obtaining the original phase values ​​corresponding to multiple discrete sampling points of the target tag within the preset sampling window; performing smoothing filtering and unwinding processing on the original phase values ​​to eliminate phase jumps caused by abrupt changes in the carrier reflection path length of the target tag; constructing a phase evolution sequence characterizing the displacement continuity of the managed target in the physical space dimension by combining the sampling period of the preset sampling window; and simultaneously extracting the received signal strength corresponding to the discrete sampling points to form the signal strength distribution data reflecting the energy fluctuation characteristics of the target tag.

[0023] Preferably, step S102 specifically includes the following steps: querying the field configuration image table, extracting a set of candidate tags that have the same hierarchical attributes as the target tag and whose antenna coverage areas overlap; monitoring the sampling frequency consistency of each tag in the candidate tag set within the preset sampling window, and determining the tags whose sampling frequency consistency is higher than a preset threshold as the reference tag set.

[0024] Preferably, step S104 includes the following sub-steps: convolving the phase evolution sequence with the suppression operator to denoise, so as to offset the collective phase shift caused by the obstruction of people in the management area; and extracting the phase change slope and frequency center shift in the individual spatial shift based on the denoised result, so as to characterize the degree of physical stripping of the target label relative to the display area.

[0025] Preferably, step S105 specifically includes: inputting the individual spatial offset into the business state machine, using convolution kernels to extract the feature correlation between the individual spatial offset and the preset standard interaction behavior template; and calculating the business confidence level by combining the asset attribute parameters corresponding to the managed target and the business weight corresponding to the management field.

[0026] Preferably, after step S106, the following steps are also included: after confirming that the managed target has undergone a business status migration, retrieving the inventory status change log associated with the target tag; when the inventory status change log shows that the managed target has entered a preset delivery area or trial area, verifying the validity of the business status migration through the business status machine feedback.

[0027] Preferably, step S106 specifically includes: encapsulating the status synchronization request of the asset attribute database using a preset asymmetric encryption algorithm; and submitting the encapsulated status synchronization request to the inventory logic processing unit of the management system to update the physical status field of the asset in the asset attribute database.

[0028] Preferably, the method further includes the following steps: real-time monitoring of the dynamic evolution trend of the suppression operator within the preset sampling window; when the suppression operator continuously exceeds the preset abnormal fluctuation threshold and covers all tags in the reference tag set, determining that there is systematic electromagnetic interference in the current management field, and outputting a management strategy adjustment signal.

[0029] Preferably, the method further includes the following steps: after the asset attribute database completes state synchronization, a sampling frequency degradation strategy is initiated for the target tag; when the managed target re-enters the display area and triggers phase regression determination, the sampling frequency of the target tag is restored.

[0030] Example 1: In a retail store application scenario with high-density garment racks, frequent foot traffic and product handling cause transient fluctuations in the signals collected by the passive RFID system. These fluctuations are physically similar to the movement of goods being removed from the racks. Traditional methods based on a single signal strength threshold struggle to distinguish between electromagnetic shielding and actual asset displacement in such high-traffic environments, leading to erroneous warnings from the inventory management system and distortions between display data and actual product status. The data processing unit in the retail management system continuously collects RFID feature streams, including the target tag's original phase value within a 200-millisecond sampling window and the received signal strength RSS data. These features are then processed through smoothing filtering and dewinding to eliminate... In addition to phase jumps caused by abrupt changes in carrier reflection path length, a phase evolution sequence P(t) reflecting the continuity of physical displacement and signal strength distribution data RSS(t) reflecting energy fluctuation characteristics are constructed. To eliminate pseudo-displacement interference caused by the movement of people in the environment, the data processing unit queries the field configuration image table, identifies the set of reference tags adjacent to the target tag in physical space, and extracts the neighborhood group feature distribution corresponding to the reference tag set. The data processing unit aggregates and generates a suppression operator according to the phase dispersion weight of each reference tag, and uses the suppression operator to perform differential decoupling calculation with the radio frequency feature flow of the target tag. By stripping the common-mode disturbance component in the phase evolution sequence, the single spatial offset of the target product relative to the display position at the physical level is determined.

[0031] The data processing unit maps the calculated individual spatial offset to the business state machine, and calculates the business confidence level of the product's migration from static display to dynamic flow. This business state machine, through logical-level inertial constraints, introduces business rules for verification when physical characteristics undergo abrupt changes, and when the business confidence level... When the preset judgment threshold is met, the data processing unit confirms that the target product has undergone a business state migration, and accordingly synchronizes the asset attribute database of the associated management system to ensure the consistency between the inventory data and the physical display status in the store asset management system at the management level. When performing differential decoupling calculation and feature correlation extraction, the processor constructs a time series array with a preset sampling window length based on the management field clock frequency as a suppression operator. It aligns the suppression operator array and the phase evolution sequence with the timestamp corresponding to the discrete sampling point of the target label as the synchronization reference, and performs element-wise subtraction operation along the time axis to generate a single spatial offset time series. The business state machine has a built-in standard interactive behavior template with a 50-sampling-period monotonically increasing phase change rate discrete sequence. The one-dimensional discrete feature mask is a standard interactive behavior template with a fixed phase change rate. The quasi-interactive behavior template uses a consistent numerical vector. In actual construction, this mask sequence is derived by statistically modeling the phase gradients of a large number of real off-shelf actions. Each element represents the significance weight of the phase change rate to the real business action at a specific sampling moment. The mask has a higher weight value in the middle stage (corresponding to the acceleration stage when the goods are completely removed from the shelf), while the weight value is lower in the beginning and end stages. This weighted calculation improves the sensitivity of the recognition of typical business feature waveforms and suppresses the interference of edge noise. A one-dimensional discrete feature mask is used as the convolution kernel, and the sliding step is set to a single sampling period. The discrete inner product is calculated by sliding along the time series of individual spatial offsets. The arithmetic mean is obtained based on the discrete inner product to determine the feature correlation between the individual spatial offset and the standard interactive behavior template.

[0032] For the business confidence quantification assessment process, the processor sets the management area business weight based on the reciprocal of the customer flow heat distribution statistics of the store display area over the past 72 hours. The customer flow heat distribution statistics are defined by the arithmetic mean of the trigger reading frequency of the reference label set in the corresponding area. When the individual spatial offset time series exceeds the preset offset threshold of 0.1 rad for 5 consecutive sampling points, the peak envelope integral value of the individual spatial offset time series that is higher than the environmental baseline offset threshold within a 200 ms sampling window is extracted. The peak envelope integral value is then multiplied and compared with the preset asset weight coefficient in the area configuration mapping table and the management area business weight. The business confidence score is obtained by multiplying the weights. The asset weight coefficient is calibrated based on the measured value of radio frequency loss of the metal shielding material of the specific goods. When the business confidence score is in the range of 0.8 to 1.0, the corresponding asset status synchronization command is output. This technical approach based on neighborhood beacon interference suppression treats complex physical interference as background noise with common characteristics of the group. By extracting business features at the logic layer instead of simply judging radio frequency strength, the system resolves the core conflict between physical environmental noise and commercial supervision accuracy without adding sensing hardware, and realizes the architectural evolution of the retail asset management system from physical signal tracking to business status recognition.

[0033] Example 2: In a test field simulating a high-density clothing display, an RF reading unit with a sampling frequency of 50Hz and a phase detection accuracy better than 0.001rad was deployed to acquire the passive tag signal attached to the merchandise. To evaluate the stability of the method under actual engineering noise, an electromagnetic noise floor with a signal-to-noise ratio of 18dB was introduced into the field, and dynamic human body occlusion interference at different frequencies was simulated. The determination of the value is constrained by the data processing load and the real-time nature of recognition. This is achieved by calculating the rate of change of multipath reflection paths under different display densities. The test was conducted within the 30Hz to 100Hz range to ensure that the complete displacement envelope was obtained under different occlusion frequencies. The test was divided into a control group A using a single intensity threshold, a control group B lacking neighborhood suppression logic, and an experimental group using the method of this invention. During the 600-second test, it was observed that when a person approached the shelf but did not move, the RSS of the received signal in control group A dropped by more than the preset 10dB threshold, causing the management system to judge the environmental occlusion as a taking action. Although control group B used phase sequence to identify displacement, under the influence of the collective phase drift caused by the simulated large-scale flow of people, its phase difference value fluctuated drastically due to the lack of decoupling effect of the suppression operator, and it could not maintain a stable business confidence output.

[0034] The experimental sample data exhibited anti-interference characteristics. Under strong interference gradients simulating high-frequency occlusion, the recognition accuracy of control sample A rapidly decreased from 82.5% in the low-interference state to 54.2%. The experimental sample improved its performance by extracting the phase dispersion weights of the neighborhood reference label set. Aggregation generation suppression operator By differentially decoupling the suppression operator with the radio frequency characteristic flow of the target tag, the mean residual of the output single-unit spatial offset under occlusion interference is maintained below 0.05 rad. For the first Phase dispersion weights of each reference label, To suppress the operator, and when the sampling window length is set within the range of 150 milliseconds to 250 milliseconds, the business confidence level is... The correlation between the growth slope and the actual displacement reaches its peak. However, when the window length exceeds 300 milliseconds, the rate of improvement in recognition accuracy tends to level off due to the increased weight of accumulated electromagnetic multipath noise. By using the neighborhood beacon interference suppression mechanism to remove common-mode physical disturbances, the determination of the product status changes from relying on electromagnetic energy to relying on deterministic single displacement features. Under the condition of a 3-fold increase in interference intensity, the consistency between the business confidence determination result and the actual status of the experimental sample group remains above 98.6%, which confirms the adaptability of the present invention to high-density interaction scenarios and provides business decision-making logic support for retail management systems based on the decoupling of radio frequency underlying parameters.

[0035] Example 3: When changes to the store display structure lead to electromagnetic field reconstruction, the data processing unit initiates the field initialization calibration procedure. During a static reference period without personnel interaction, the procedure controls the RFID reader to continuously initiate 500 cycles of signal acquisition from all tags within the field. The data processing unit calculates the phase correlation coefficient between any two tag signals. The formula is as follows: ,in, For tags With tags The phase correlation coefficient, For tags With tags The covariance of the phase evolution sequence, For tags The phase standard deviation, For tags The phase standard deviation; the data processing unit compares the phase correlation coefficients between each label, and... Tag pairs with a value greater than 0.95 are identified as belonging to the same local multipath interference cluster covered by the same physical antenna, and a field configuration mapping table is constructed accordingly. This provides definite physical coordinate anchor points for the reference tag set used for subsequent real-time identification of target tags. To quantify and suppress field background noise, the data processing unit extracts the phase variance value of each reference tag in the reference tag set within the sampling window. And calculate the phase dispersion weight accordingly. , The calculation formula is as follows: ,in, For the first Phase dispersion weights of each reference label, To refer to the total number of tags in the tag set, For the first The phase variance of each reference tag; the data processing unit compares the phase offset of each tag in the reference tag set with the corresponding... Perform weighted summation to aggregate and generate suppression operators. The suppression operator Real-time characterization of the common-mode disturbance intensity caused by overall environmental electromagnetic fluctuations, and when a reference tag is accidentally obstructed... When it increases, the corresponding It automatically decreases, thereby maintaining the stability of the suppression operator's representation of global noise.

[0036] The data processing unit will process the phase evolution sequence of the target tag. Subtract the suppression operator Obtain the spatial offset of a single unit. and the individual spatial offset Input a preset business state machine, which contains three logical levels: static bit, pending state, and transition state. The initial state is static bit; when a single-unit spatial offset is detected... When five consecutive sampling points exceed the first preset threshold of 0.1 rad, the service state machine transitions from a static state to an undetermined state; in the undetermined state, the data processing unit calculates... The business confidence score is obtained by integrating the values ​​within a 200-millisecond sampling window and multiplying them by the asset weight coefficient of the item. The asset weighting coefficient is determined by the initial RF link loss baseline value in the field configuration mapping table of the goods, when the business confidence level is... If the value continuously increases within a 100-millisecond duration and the peak value exceeds the second preset threshold of 0.8, the business state machine transitions from the pending state to the transition state, confirming that a real removal action of the goods has occurred; if in the pending state... If the second preset threshold is not exceeded or a downward trend is observed, the business state machine reverts to a static position and marks the fluctuation as environmental transient interference. This logic architecture based on three-level state determination limits the random radio frequency jitter of the physical layer to the inertial constraints of the business logic. Through the dynamic coupling of physical quantity calibration and business threshold, it eliminates the uncertainty caused by relying on empirical values ​​in traditional algorithms, ensuring the logical rigor and reproducibility of inventory data updates in the retail asset supervision system.

[0037] Example 4: In a continuously operating store display area, due to minor physical deformations of the shelves or slow drift caused by environmental electromagnetic characteristics, the data processing unit adopts a dynamic maintenance method for the field reference based on a sliding window. During real-time monitoring, it extracts the phase median value of all reference tags in a static display state over the past 72 hours to construct a time-domain reference sequence. And calculate the current instantaneous observation value and Cumulative deviation value ;when When the drift tolerance threshold of 0.05 rad is exceeded, the data processing unit updates the background noise reference by using a time weighting coefficient that decreases based on the historical reference value. The initial RF link loss baseline value in the field configuration mapping table is incrementally corrected, wherein... As a time-domain reference sequence, This is the cumulative deviation value. The time weighting coefficient is set to 0.98 to ensure that the management system maintains consistency in extracting individual displacement features even when environmental parameters experience second-order slow loss.

[0038] When the system faces the deployment of heterogeneous reading terminals or the mixing of passive tags from different batches, in order to determine the system normalization constant... The data processing unit employs a hardware gain flatness calibration method. Within a controlled, shielded test chamber, a standard sample tag is placed at a fixed position 2 meters from the antenna's central axis. The RF reader is controlled to transmit in steps with a power gradient ranging from 30dBm to 33dBm. The raw values ​​of the received signal strength at different transmit powers are recorded, and the corresponding link gain components are extracted. The data processing unit, according to The proportional relationship with the theoretical response threshold of the tag chip is used to determine the system normalization constant. ,in Let be the system normalization constant. It is the link gain component, and The value of makes the asset weight coefficient under standard operating conditions Within the range of 0.8 to 1.2, this quantitative feedback based on hardware physical response eliminates the bias in business confidence judgment caused by differences in perception layer devices, and achieves stable output of retail asset management accuracy in large-scale deployment scenarios.

[0039] Example 5: In a heterogeneous display verification scenario involving metal mesh shelves and enclosed counters, the data processing unit initiates a field adaptability parameter optimization program. This involves collecting 50 sets of phase evolution sequence samples containing standard picking and placing actions at the target display location, while simultaneously recording the background phase shift under controlled pedestrian flow interference. The data processing unit initiates an iterative step search for a first preset threshold within the range of 0.05 rad to 0.3 rad. By comparing the number of false alarms and false negatives at different thresholds, the value that minimizes the recognition error rate is selected as the first preset threshold for the current physical sub-region. This offline calibration process based on measured samples enables the business state machine to complete logical alignment with the specific physical display environment before real-time monitoring is initiated. When the system faces signal drift caused by minor settlement of the shelf structure or offset of label attachment position, the data processing unit initiates an online consistency self-check program. By continuously monitoring the sampling frequency consistency deviation of each label in the reference label set, the evolution entropy value of the environmental background noise benchmark is calculated. This is used to characterize the stability of the electromagnetic features of the field, when the evolution entropy value... When the drift tolerance threshold of 0.15 is exceeded for 12 consecutive hours, the data processing unit automatically issues a local incremental reconstruction instruction for the field configuration image table. By performing aggregation calculations on the suppression operator during idle periods, the phase residual bias caused by the slow deformation of the physical environment is corrected, thereby maintaining the mirror consistency between the product display status and the digital records in the retail management system.

[0040] In high-density data flow processing of people and goods interactions, to quantify the inertial constraints within the business state machine, the system establishes a statistical discrimination method based on a sliding time window. The data processing unit determines the business confidence level of the target label. First, divide the preset sampling window into The observation sub-intervals are divided into several equal-length sub-intervals, and the mean local confidence score is calculated for each sub-interval. When continuous Observation sub-intervals When all parameters exhibit a monotonically increasing trend and their slope gradient is greater than the preset kinetic energy compensation coefficient, the service state machine initiates a state accumulation operation. This decision logic, based on multi-point resonance in timing, eliminates the oscillations in the system's discrimination logic caused by RF multipath jitter at a single discrete sampling point. This represents the total number of observed sub-intervals, and its value is set between 5 and 10. Accumulate the step length for each state, and Less than To ensure the management system accurately updates the asset attribute database after confirming a business state migration of goods, the inventory logic processing unit maintains data by executing a transaction-consistent state synchronization procedure. Before issuing an inventory deduction instruction to the asset attribute database, the system retrieves the real-time activity score of the managed target in the field configuration image table. If the score is lower than the preset offline judgment threshold and corresponds to the business confidence level of the transition state, the system will proceed accordingly. If the value remains above 0.9, the management system initiates database transaction locking, changing the physical status field of the product from static display state to dynamic pending inspection state. The final inventory write-off operation is only executed after the POS terminal or anti-theft gate node returns an identity verification signal. This multi-node interlocking supervision strategy avoids inconsistencies between financial accounts and physical assets caused by misjudgments due to physical obstruction, and realizes a closed loop of supervision logic in the digital operation of stores.

[0041] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An algorithm for managing the display and retrieval of merchandise in stores based on passive RFID technology, characterized in that, Includes the following steps: Step S101: Obtain the radio frequency feature stream of the target tag corresponding to the managed target within a preset sampling window of the management field. The radio frequency feature stream includes phase evolution sequence and signal strength distribution data. Step S102: Identify a set of reference tags that are physically adjacent to the target tag within the display area, and extract the neighborhood group feature distribution corresponding to the reference tag set to characterize the background environmental noise benchmark of the management area; wherein, step S102 specifically includes the following steps: query the field configuration image table to extract a set of candidate tags that have the same hierarchical attributes as the target tag and whose antenna coverage areas overlap; monitor the sampling frequency consistency of each tag in the candidate tag set within the preset sampling window, and determine the tags whose sampling frequency consistency is higher than a preset threshold as the reference tag set; Step S103: Based on the background environmental noise reference mapping, a suppression operator characterizing the field disturbance intensity is generated. The suppression operator is aggregated from the phase dispersion weights of each label in the reference label set, and is used to lock the common physical disturbance characteristics of the managed field. Step S104: Differential decoupling calculation is performed on the radio frequency feature stream and the suppression operator to remove the common-mode disturbance component in the phase evolution sequence and generate the individual spatial offset of the managed target relative to the display area; wherein, step S104 includes the following sub-steps: convolving the phase evolution sequence and the suppression operator to denoise, so as to cancel the collective phase offset caused by the blockage of people in the management area; extracting the phase change slope and frequency center offset in the individual spatial offset based on the denoising result, so as to characterize the degree of physical stripping of the target label relative to the display area; Step S105: Map the spatial offset of the single unit to the preset business state machine, calculate the business confidence of the managed target in the management process from static display to dynamic flow. The business state machine is built based on the business logic flow of asset interaction, and is used to achieve accurate identification of real picking and placing behavior in store retail management business through logical extraction of physical features. Step S106: When the business confidence level meets the preset judgment threshold, it is confirmed that the managed target has undergone business state migration, and the asset attribute database of the associated management system is synchronized according to the output result of the business state machine, so as to ensure the consistency between the inventory data and the physical display status in the store asset management system at the management level through the verification of the business logic layer.

2. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, Step S101 includes the following sub-steps: obtaining the original phase values ​​corresponding to multiple discrete sampling points of the target tag within the preset sampling window; performing smoothing filtering and unwinding processing on the original phase values ​​to eliminate phase jumps caused by abrupt changes in the carrier reflection path length of the target tag; constructing a phase evolution sequence characterizing the displacement continuity of the managed target in the physical space dimension by combining the sampling period of the preset sampling window; and simultaneously extracting the received signal strength corresponding to the discrete sampling points to form the signal strength distribution data reflecting the energy fluctuation characteristics of the target tag.

3. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, Step S105 specifically includes: inputting the spatial offset of the individual unit into the business state machine, using convolution kernels to extract the feature correlation between the spatial offset of the individual unit and the preset standard interaction behavior template; and calculating the business confidence level by combining the asset attribute parameters corresponding to the managed target and the business weight corresponding to the management field.

4. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, After step S106, the following steps are also included: after confirming that the managed target has undergone a business status migration, the inventory status change log associated with the target tag is retrieved; when the inventory status change log shows that the managed target has entered a preset delivery area or trial area, the validity of the business status migration is verified through the business status machine feedback.

5. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, Step S106 specifically includes: encapsulating the status synchronization request of the asset attribute database using a preset asymmetric encryption algorithm; and submitting the encapsulated status synchronization request to the inventory logic processing unit of the management system to update the physical status field of the asset in the asset attribute database.

6. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, It also includes the following steps: Real-time monitoring of the dynamic evolution trend of the suppression operator within the preset sampling window; When the suppression operator continuously exceeds the preset abnormal fluctuation threshold and covers all tags in the reference tag set, it is determined that there is systematic electromagnetic interference in the current management field, and a management strategy adjustment signal is output.

7. The algorithm for managing the display and retrieval of store merchandise based on passive RFID technology according to claim 1, characterized in that, It also includes the following steps: After the asset attribute database completes state synchronization, a sampling frequency degradation strategy is initiated for the target tag; when the managed target re-enters the display area and triggers phase regression determination, the sampling frequency of the target tag is restored.

Images