RFID positioning method, device, equipment, medium and program product
By obtaining the tag data of RFID tags and the trained positioning model for weighting and reconstruction, the problem of retraining the model after the environment changes is solved, and the effect of intelligent positioning and cost reduction is achieved.
Patent Information
- Application Number
- CN202411084565.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art requires retraining the positioning model after the environment changes, resulting in high application costs.
By obtaining the tag data of the RFID tag, determining the weighting function and its value parameters, and using the trained positioning model for weighting and reconstruction, realizing intelligent positioning without retraining the model.
The adaptive effect of no need to retrain the positioning model after the environment changes is achieved, which reduces application costs and solves the problem of insufficient technical resources of the end-side equipment.
Smart Images

Figure CN119095017A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of positioning technology, and in particular to an RFID positioning method, device, equipment, medium and program product. Background Art
[0002] Passive IoT systems have been widely used in the field of smart retail. All retail stores have fully launched passive IoT systems. All products are affixed with RFID (Radio Frequency Identification) tags, which are identified by handheld devices to improve work efficiency in warehousing, storage, and checkout.
[0003] The current passive indoor positioning method trains the positioning model by collecting a labeled data set Ds in the original environment; collecting a small amount of unlabeled data set Dt after the environment changes; using the labeled data set Ds of the original environment to calculate the coefficient matrix Ai; using the coefficient matrix Ai to assist in clustering Dt to obtain a labeled data set Dl; retraining the positioning model with Dl, and fine-tuning the model to obtain the positioning model in the new environment.
[0004] However, the existing positioning model cannot adapt to changes in the environment and needs to be retrained, resulting in high application costs. Summary of the invention
[0005] The technical solution of the present invention aims to provide an RFID positioning method, device, equipment, medium and program product, which are used to solve the problem in the prior art that the positioning model needs to be retrained after the environment changes, resulting in high application costs.
[0006] To achieve the above object, the present invention is achieved by:
[0007] In a first aspect, an embodiment of the present invention provides a radio frequency identification (RFID) positioning method, comprising:
[0008] When the RFID antenna sends a radio frequency signal, obtaining first tag data of the first RFID tag;
[0009] Determining a corresponding weighting function according to a first label feature vector corresponding to the first label data;
[0010] Performing a difference measurement between the weighting function and a second label feature vector to determine a value parameter of the weighting function, wherein the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model;
[0011] Determining a third label feature vector corresponding to the first label data according to the weighting function and the value parameter;
[0012] The location information of the first RFID tag is obtained according to the third tag feature vector and the positioning model.
[0013] Optionally, the RFID positioning method, wherein the weighting function and the second tag feature vector are subjected to difference measurement to determine the value parameters of the weighting function, comprises:
[0014] Obtaining a difference metric between the weighting function and a second label feature vector;
[0015] Taking minimization of the difference metric as an iteration goal, the weighting function is iterated to determine the value parameters of the weighting function.
[0016] Optionally, the RFID positioning method further comprises:
[0017] When the RFID antenna sends a radio frequency signal, obtaining third tag data of the reference tag;
[0018] A positioning model is trained according to the second label feature vector and the environment feature vector corresponding to the third label data.
[0019] Optionally, the RFID positioning method further comprises:
[0020] When the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag;
[0021] The second label feature vector is obtained according to a feature extraction rule and the second label data, wherein the feature extraction rule is determined based on the sensitivity of each potential feature vector in a potential feature vector set to noise, and the potential feature vector set is obtained based on the second label data and the third label data.
[0022] Optionally, the RFID positioning method further comprises:
[0023] Obtaining a first importance score for each potential feature vector in the potential feature vector set according to the first machine learning model and the potential feature vector set;
[0024] Performing noise processing on the potential feature vector set to obtain a processed potential feature vector set;
[0025] Obtaining a second importance score for each potential feature vector in the processed set of potential feature vectors based on the first machine learning model and the processed set of potential feature vectors;
[0026] Determining the sensitivity of each potential feature vector to noise according to the first importance score and the second importance score;
[0027] The feature extraction rule is determined based on the sensitivity of each potential feature vector to noise, and the feature extraction rule is used to indicate to remove the potential feature vectors with high sensitivity to noise.
[0028] Optionally, the RFID positioning method further comprises:
[0029] According to the second label data and the third label data, an original feature vector set is obtained, where the original feature vector set includes a feature vector corresponding to at least one of the following data:
[0030] Received signal strength indication RSSI;
[0031] Initial phase;
[0032] Arrival phase;
[0033] Frequency;
[0034] Timestamp;
[0035] Based on the original feature vector set, the potential feature vector set is obtained, and the potential feature vector set includes potential feature vectors corresponding to at least one of the following data within the inventory time:
[0036] Number of reads per RFID antenna per tag;
[0037] The average RSSI value of each tag single RFID antenna;
[0038] RSSI variance of a single RFID antenna per tag.
[0039] Optionally, the RFID positioning method, wherein training a positioning model according to the second tag feature vector and the environment feature vector corresponding to the third tag data, comprises:
[0040] Training a positioning model according to the environmental feature vector corresponding to the third label data to obtain a positioning model with a weight of a first value;
[0041] Training the positioning model with the first weight according to the second label feature vector to obtain a positioning model with the second weight;
[0042] Whether the positioning model has completed training is determined according to the positioning accuracy of the positioning model whose weight is the second value.
[0043] Optionally, the RFID positioning method, wherein when the RFID antenna sends a radio frequency signal, obtaining the third tag data of the reference tag includes:
[0044] Based on the location information of the reference tag and the coverage of each RFID antenna, selecting an RFID antenna combination from the plurality of RFID antennas, wherein the coverage of any two RFID antennas in the RFID antenna combination does not overlap;
[0045] The third tag data of the reference tag is acquired when all RFID antennas in the RFID antenna combination simultaneously transmit radio frequency signals, wherein the radio frequency signals carry an inventory mode instruction.
[0046] Optionally, the RFID positioning method, wherein when the RFID antenna sends a radio frequency signal, obtaining the second tag data of the second RFID tag includes:
[0047] Determining a scheduling mode according to the number of the second RFID tags;
[0048] When the RFID antenna sends a radio frequency signal according to the scheduling mode, second tag data of the second RFID tag is acquired.
[0049] In a second aspect, an embodiment of the present invention further provides an RFID positioning device, comprising:
[0050] A first acquisition module, configured to acquire first tag data of a first RFID tag when the RFID antenna sends a radio frequency signal;
[0051] A first determining module, configured to determine a corresponding weighting function according to a first label feature vector corresponding to the first label data;
[0052] A second determination module is used to measure the difference between the weighting function and the second label feature vector to determine the value parameter of the weighting function, where the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model;
[0053] A third determination module, used to determine a third label feature vector corresponding to the first label data according to the weighting function and the value parameter;
[0054] The first obtaining module is used to obtain the location information of the first RFID tag according to the third tag feature vector and the positioning model.
[0055] In a third aspect, an embodiment of the present invention further provides an RFID positioning device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the RFID positioning method as described in the first aspect is implemented.
[0056] In a fourth aspect, an embodiment of the present invention further provides a readable storage medium, on which a program is stored, and when the program is executed by a processor, the RFID positioning method as described in the first aspect is implemented.
[0057] In a fifth aspect, an embodiment of the present invention further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the RFID positioning method as described in the first aspect.
[0058] The beneficial effects of the above technical solution of the present invention are as follows:
[0059] By adopting the RFID positioning method described in the embodiment of the present invention, when the RFID antenna sends a radio frequency signal, the first tag data of the first RFID tag is obtained; according to the first tag feature vector corresponding to the first tag data, the corresponding weighting function is determined; the weighting function and the second tag feature vector are measured for difference, and the value parameter of the weighting function is determined, and the second tag feature vector is the feature vector corresponding to the second tag data used when training the positioning model; according to the weighting function and the value parameter, the third tag feature vector corresponding to the first tag data is determined; according to the third tag feature vector and the positioning model, the location information of the first RFID tag is obtained. In this way, by measuring the difference between the weighting function corresponding to the first RFID tag and the second tag feature vector used when training the positioning model, the value parameter of the weighting function is determined, so that the feature vector corresponding to the first RFID tag is weighted and reconstructed, and the weighted and reconstructed third tag feature vector can directly apply the trained positioning model to realize intelligent positioning, without retraining the positioning model, to achieve adaptive effect and reduce application cost. Moreover, the positioning model does not need to be retrained, which solves the problem of insufficient technical resources of the terminal side equipment deployed by the positioning model. BRIEF DESCRIPTION OF THE DRAWINGS
[0060] Figure 1 A schematic diagram of the flow of the RFID positioning method according to an embodiment of the present invention;
[0061] Figure 2 A schematic diagram of a flow chart of one implementation method of the method according to an embodiment of the present invention;
[0062] Figure 3 A schematic diagram of a second implementation of the method according to an embodiment of the present invention;
[0063] Figure 4 This is a schematic diagram of the coverage of the RFID antenna according to an embodiment of the present invention;
[0064] Figure 5 A concurrent logic diagram of the RFID antenna according to the embodiment of the present invention;
[0065] Figure 6 This is a schematic diagram of the structure of the RFID positioning device according to an embodiment of the present invention;
[0066] Figure 7 A hardware structure diagram of the RFID positioning device is provided for an embodiment of the present invention. DETAILED DESCRIPTION
[0067] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.
[0068] The embodiment of the present invention, as Figure 1 As shown, a radio frequency identification (RFID) positioning method is provided, comprising:
[0069] Step 101: Acquire first tag data of a first RFID tag when an RFID antenna sends a radio frequency signal.
[0070] It should be noted that the first RFID tag is a newly added RFID tag in the current area to be located, resulting in a change in the environment. The so-called newly added can be understood as the first RFID tag being newly added compared to the second RFID tag used when training the positioning model, and can also be understood as the first RFID tag being different from the second RFID tag used when training the positioning model.
[0071] Optionally, the first RFID tag and the second RFID tag are set on commodities in the area to be located.
[0072] Step 102: Determine a corresponding weighting function according to the first label feature vector corresponding to the first label data.
[0073] In this step, the weighting function is determined using the following formula (1):
[0074] C Weighted =A·C New +B(1)
[0075] Among them, C New is the first label feature vector corresponding to the first label data; A is the weight value parameter; B is the bias value parameter.
[0076] Step 103, performing a difference measurement between the weighting function and the second label feature vector to determine the value parameters of the weighting function, wherein the second label feature vector is a feature vector corresponding to the second label data used in training the positioning model.
[0077] In the embodiment of the present invention, the second tag feature vector is a feature vector corresponding to the second tag data used when training the positioning model; the second tag data is tag data corresponding to the second RFID tag.
[0078] Determining the value parameters of the weighting function includes: determining the weight value parameter A and the bias value parameter B in the above formula (1).
[0079] Step 104: Determine a third label feature vector corresponding to the first label data according to the weighting function and the value parameter.
[0080] That is, the label feature vector corresponding to the first label data is weightedly reconstructed using formula (1) according to the value parameters determined in step 103 to determine the third label feature vector.
[0081] Step 105: Obtain location information of the first RFID tag according to the third tag feature vector and the positioning model.
[0082] In the embodiment of the present invention, the weighted reconstructed third tag feature vector and the positioning model are used to obtain the location information of the first RFID tag, without the need to retrain the positioning model, thereby reducing the application cost.
[0083] The RFID positioning method described in the embodiment of the present invention measures the difference between the weighted function corresponding to the newly added first RFID tag and the second tag feature vector used in the positioning model training, determines the value parameters of the weighted function, and thus weighted reconstructs the feature vector corresponding to the first RFID tag. The weighted reconstructed feature vector can directly apply the trained positioning model to realize intelligent positioning without retraining, thereby achieving an adaptive effect and reducing application costs.
[0084] In one implementation manner, optionally, performing a difference measurement between the weighting function and the second label feature vector to determine a value parameter of the weighting function includes:
[0085] Obtaining a difference metric between the weighting function and a second label feature vector;
[0086] Taking minimization of the difference metric as an iteration goal, the weighting function is iterated to determine the value parameters of the weighting function.
[0087] It should be noted that the difference metric is used to measure the probability distribution difference between the weighted function and the second label feature vector. Optionally, the difference metric includes one of KL (Kullback-Leibler divergence) divergence, JS (Jensen-Shannon divergence) divergence, and EMD (Earth Mover's Distance).
[0088] Optionally, a gradient descent optimization algorithm is iterated for multiple rounds to learn the value parameters of the weighting function with the goal of minimizing the difference metric.
[0089] like Figure 2 The flowchart of one implementation of the RFID positioning method according to the embodiment of the present invention is shown in FIG. In one of the implementations, the RFID positioning method includes the following steps:
[0090] Step 201, obtaining first tag data of a first RFID tag;
[0091] Step 202: Obtain a first label feature vector corresponding to the first label data according to the feature extraction rule and the first label data.
[0092] Step 203: Determine a corresponding weighting function according to the first label feature vector.
[0093] Step 204: Obtain a second tag feature vector, where the second tag feature vector is a feature vector corresponding to the second tag data of the second RFID tag used when training the positioning model.
[0094] Step 205: Obtain a difference measurement index between the weighting function and the second label feature vector.
[0095] Step 206, determining whether the difference metric is minimized.
[0096] If the judgment result of step 206 is no, then the process proceeds to step 207 to iteratively optimize the weighting function, and then returns to step 203 .
[0097] If the judgment result of step 206 is yes, then proceed to step 208 to perform weighted reconstruction on the first label data according to the value parameters of the weighting function to obtain a third label feature vector.
[0098] Step 209: input the third label feature vector into the positioning model.
[0099] Step 210: output positioning information.
[0100] In one embodiment, optionally, the method further comprises:
[0101] When the RFID antenna sends a radio frequency signal, obtaining third tag data of the reference tag;
[0102] A positioning model is trained according to the second label feature vector and the environment feature vector corresponding to the third label data.
[0103] In an embodiment of the present invention, a passive Internet of Things system is deployed in the area to be positioned, and reference tags are deployed in the area to be positioned. The specific deployment location is related to the scenario requirements. If the scenario requires shelf-level positioning, m reference tags can be deployed on the left and right sides of the shelf respectively.
[0104] In order to save system resources, the regional level, antenna level or subdivided levels such as shelf level, shelf left and right level, meter level, etc. are configured to meet the needs of different scenarios. Different data will be obtained for different scenario needs, thereby reducing resource waste.
[0105] The RFID positioning method described in the embodiment of the present invention is aimed at the problem that in retail store scenarios, due to the presence of obstructions or interferences such as metal shelves and walls, the signal undergoes multiple reflections, refractions and scattering during the propagation process, resulting in multiple arrivals of the signal, and the returned data information lacks accuracy and stability, making it difficult to support accurate judgment by the positioning algorithm. A reference tag is used to measure the environment of the area to be positioned, and the third tag data returned can be understood as environmental data, thereby realizing intelligent regional-level positioning in a complex interference environment.
[0106] In one embodiment, optionally, the method further comprises:
[0107] When the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag;
[0108] The second label feature vector is obtained according to a feature extraction rule and the second label data, wherein the feature extraction rule is determined based on the sensitivity of each potential feature vector in a potential feature vector set to noise, and the potential feature vector set is obtained based on the second label data and the third label data.
[0109] In the embodiment of the present invention, a feature extraction rule is determined based on the sensitivity of the potential feature vector to noise, and the second label feature vector is screened out according to the feature extraction rule, which can improve the robustness of the positioning model.
[0110] In one embodiment, optionally, the method further comprises:
[0111] Obtaining a first importance score for each potential feature vector in the potential feature vector set according to the first machine learning model and the potential feature vector set;
[0112] Performing noise processing on the potential feature vector set to obtain a processed potential feature vector set;
[0113] Obtaining a second importance score for each potential feature vector in the processed set of potential feature vectors based on the first machine learning model and the processed set of potential feature vectors;
[0114] Determining the sensitivity of each potential feature vector to noise according to the first importance score and the second importance score;
[0115] The feature extraction rule is determined based on the sensitivity of each potential feature vector to noise, and the feature extraction rule is used to indicate to remove the potential feature vectors with high sensitivity to noise.
[0116] Among them, the first machine learning model includes but is not limited to RF (Random Forest) model, SVM (Support Vector Machine), XGBoost (Extreme Gradient Boosting) model, etc.
[0117] In an embodiment of the present invention, taking the first machine learning model as an RF model as an example, the RF model is trained based on a potential feature vector set to obtain a first importance score for each potential feature vector in the potential feature vector set.
[0118] Noise processing is performed on the latent feature vector set, for example, the order of the latent feature vectors in the latent feature vector set is randomly disrupted, and a noise vector is added to the randomly disrupted latent feature vector set.
[0119] Furthermore, the RF model is trained based on the noise-processed latent feature vector set to obtain the second importance score of each latent feature vector in the latent feature vector set.
[0120] According to the first importance score and the second importance score, the sensitivity of each potential feature vector to noise is determined. If the second importance score decreases more than the first importance score, the corresponding potential feature vector is more sensitive to noise. Conversely, if the second importance score decreases less than the first importance score, the corresponding potential feature vector is less sensitive to noise.
[0121] The latent feature vectors with high sensitivity to noise are eliminated to determine the feature extraction rule. Here, the latent feature vectors with high sensitivity to noise can be understood as the difference between the first importance score and the second importance score corresponding to the latent feature vector is greater than the score threshold.
[0122] In the embodiment of the present invention, the first label feature vector, the second label feature vector and the environment feature vector can all be obtained according to the above-mentioned feature extraction rules.
[0123] In one embodiment, optionally, the method further comprises:
[0124] According to the second label data and the third label data, an original feature vector set is obtained, where the original feature vector set includes a feature vector corresponding to at least one of the following data:
[0125] RSSI (Received Signal Strength Indication);
[0126] Initial phase;
[0127] Arrival phase;
[0128] Frequency;
[0129] Timestamp;
[0130] Based on the original feature vector set, the potential feature vector set is obtained, and the potential feature vector set includes potential feature vectors corresponding to at least one of the following data within the inventory time:
[0131] Number of reads per RFID antenna per tag;
[0132] The average RSSI value of each tag single RFID antenna;
[0133] RSSI variance of a single RFID antenna per tag.
[0134] In the embodiment of the present invention, the original feature vector set can be directly obtained from the second label data and the third label data, and further calculations are performed based on the original feature vector set to deeply mine potential feature vectors and obtain the potential feature vector set.
[0135] Specifically, the number of reads per RFID antenna for each tag is calculated as follows:
[0136] Assume that in a passive IoT system, there are n RFID antennas and m tags. Let the number of times the jth tag is read by the ith RFID antenna be N. ij , where i = 1…n, j = 1…m. By counting the number of times each tag is read by a single RFID antenna, it is possible to determine the RFID antenna to which the tag is closer, thereby assisting in achieving more accurate positioning.
[0137] The calculation method of the RSSI average of each tag single RFID antenna is as follows:
[0138] For the case where the jth tag is read multiple times by the ith RFID antenna, the RSSI values read multiple times are recorded as RSSI ij1 ,RSSI ij2 ,…,RSSI ijk , where k represents the number of reads. Then the RSSI average of the jth tag read by the i-th RFID antenna is It is expressed by the following formula (2):
[0139]
[0140] The RSSI variance of each tag single antenna is calculated as follows:
[0141] For the case where the jth tag is read multiple times by the i-th RFID antenna, the RSSI variance It is expressed by the following formula (3):
[0142]
[0143] In one implementation manner, optionally, training a positioning model according to the second label feature vector and the environment feature vector corresponding to the third label data includes:
[0144] Training a positioning model according to the environmental feature vector corresponding to the third label data to obtain a positioning model with a weight of a first value;
[0145] Training the positioning model with the first weight according to the second label feature vector to obtain a positioning model with the second weight;
[0146] Whether the positioning model has completed training is determined according to the positioning accuracy of the positioning model whose weight is the second value.
[0147] like Figure 3 The flowchart of the second implementation method of the RFID positioning method according to the embodiment of the present invention is shown in FIG. In the second implementation method, the RFID positioning method includes the following steps:
[0148] Step 301, building a model framework: setting the initial parameters of the model. Here, the positioning model includes but is not limited to RF model, SVM, XGBoost model, etc.
[0149] Step 302, model pre-training: train the positioning model according to the environmental feature vector until the performance converges, and obtain a positioning model with a weight of the first value W1, so that the positioning model learns the environmental characteristics of the area to be positioned, and reduces the adverse effects of environmental interference on positioning accuracy.
[0150] Step 303, model retraining: according to the second label feature vector, the positioning model with the weight of the first value W1 is trained until the performance converges, and the positioning model with the weight of the second value W2 is obtained, so that the positioning model learns the positioning task.
[0151] Step 304, model optimization: test the positioning accuracy of the positioning model with the weight of the second value W2 on the test data. If the positioning accuracy does not meet the standard, return to step 301, modify the model parameters or model structure until the positioning accuracy meets the standard, and complete the positioning model training.
[0152] Step 305, model deployment: deploy the trained positioning model to the cloud or end-side device.
[0153] It should be noted that the RFID positioning method described in the embodiment of the present invention does not need to be retrained when the model is applied, which solves the problem in the prior art that the model is deployed to the terminal device and the computing resources of the terminal device are limited and cannot meet the model retraining requirements.
[0154] In one embodiment, optionally, when the RFID antenna sends a radio frequency signal, obtaining the third tag data of the reference tag includes:
[0155] Based on the location information of the reference tag and the coverage of each RFID antenna, selecting an RFID antenna combination from the plurality of RFID antennas, wherein the coverage of any two RFID antennas in the RFID antenna combination does not overlap;
[0156] The third tag data of the reference tag is acquired when all RFID antennas in the RFID antenna combination simultaneously transmit radio frequency signals, wherein the radio frequency signals carry an inventory mode instruction.
[0157] In an embodiment of the present invention, the method for selecting the inventory mode includes one of the following:
[0158] The same as the actual business execution mode, for example, if the actual business execution mode is a polling inventory mode, then the inventory mode described in the embodiment of the present invention also adopts the polling inventory mode;
[0159] For a small number of reference tags (e.g., several or dozens) and a small space, such as less than 10 readers or RFID antennas, inventory mode 1 can be used, which is a polling inventory mode;
[0160] For low positioning requirements, such as regional level, inventory mode 2 may be used. Inventory mode 2 is a concurrent inventory mode for all RFID antennas. Here, the embodiment of the present invention uses inventory mode 2.
[0161] For situations with high positioning requirements or multi-positioning business combinations and large spaces during actual business execution, inventory mode 3 can be used. Inventory mode 3 is the execution of inventory mode 1 and inventory mode 2 N times respectively.
[0162] It should be noted that the coverage of the RFID antenna is related to the RFID antenna model, such as Figure 4 As shown, numbers 1 to 8 represent the coverage of the RFID antenna. The coverage of the RFID antenna is a circular area. Adjacent RFID antennas may have overlapping coverage. The small black circle represents the reference tag.
[0163] The RFID antenna combination described in the embodiment of the present invention needs to meet the non-interference target. It should be noted that it is a non-interference target when one of the following conditions is met:
[0164] All reference tags within the coverage range of each RFID antenna in the RFID antenna combination can be detected;
[0165] All reference tags within the coverage range of each RFID antenna in the RFID antenna combination can be detected, and the RSSI is greater than the RSSI threshold;
[0166] The reference tags within the coverage range of each RFID antenna in the RFID antenna combination can be retrieved, and the number of times the reference tags are retrieved within a preset time is greater than a number threshold.
[0167] Further, Figure 4 The RFID antenna 1 in the center meets the above non-interference goal and starts concurrently from the adjacent RFID antennas, such as Figure 5 The RFID antenna concurrency logic shown determines an RFID antenna combination, and the coverage ranges of any two RFID antennas in the RFID antenna combination do not overlap.
[0168] Furthermore, if Figure 5 The RFID antenna concurrency logic shown includes the following steps:
[0169] Step 501, n RFID antennas concurrently;
[0170] Step 502, taking RFID antenna 1 as the center, starting from adjacent RFID antennas, record the combination: 1+2+…+n, where the number represents the ID of the RFID antenna;
[0171] Step 503, if the concurrent RFID antennas meet the above non-interference target, record the combination, which is the non-interference target combination;
[0172] Step 504, if the concurrent RFID antennas do not meet the above non-interference target, record the combination, which is an interference target combination;
[0173] Step 505, determining whether n RFID antennas have completed the concurrency;
[0174] If the result of the judgment in step 505 is no, the process proceeds to step 506, where the concurrency is started with the RFID antenna 2 that has not completed the concurrency, and the record combination is: 1+2+…+n-1+n+1;
[0175] If the result of the judgment in step 505 is no, the process proceeds to step 507, and the concurrency is started with the RFID antenna 3 that has not completed the concurrency, and the record combination is: 1+3+…+n+n+1;
[0176] If the determination result of step 505 is yes, the process proceeds to step 508 to end the concurrent logic and determine the RFID antenna combination.
[0177] In one embodiment, optionally, when the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag includes:
[0178] Determining a scheduling mode according to the number of the second RFID tags;
[0179] When the RFID antenna sends a radio frequency signal according to the scheduling mode, second tag data of the second RFID tag is acquired.
[0180] In the embodiment of the present invention, determining the scheduling mode according to the number of the second RFID tags includes:
[0181] The number of the second RFID tags is the full number, that is, all the second RFID tags known in the area to be located, corresponding to scheduling mode 1;
[0182] The number of the second RFID tags is partial, that is, the number of the second RFID tags is greater than one, corresponding to scheduling mode 2;
[0183] The number of the second RFID tag is one, corresponding to scheduling mode three.
[0184] Furthermore, the scheduling mode 1 includes:
[0185] Polling is performed in sequence based on the identification of the reader and the identification of the RFID antenna. For example, if readers 1 to 10 are deployed and RFID antennas 1A 1B 1C to 10A 10B 10C are deployed, then polling starts from RFID antenna 1A and then to RFID antenna 10C in sequence.
[0186] Polling starts from the RFID antenna that last completed the inventory;
[0187] Concurrency is based on the non-interference principle.
[0188] Scheduling mode 2 includes:
[0189] Polling is performed first according to the last recorded position of the target tag; if the disk is not complete, other RFID antennas are polled;
[0190] According to the last recorded position of the target tag, the maximum intersection in the non-interfering record table is used for concurrency, and then the tags without work are polled;
[0191] According to the last recorded position of the target tag, the interference record table is used for partition concurrency. For example, if the last recorded position of the target tag is 1A 1B 2C 4C, if the interference list contains 1A+1B, 2C+4C, concurrency is not possible; therefore, 1A+2C concurrency or 1B+4C concurrency is used.
[0192] Scheduling mode three includes:
[0193] Based on the last recorded position of the target tag, an inventory is taken (using the select command). If it is not found, polling starts from the adjacent RFID antenna.
[0194] Based on the last recorded position of the target tag, the maximum intersection of the recorded position in the non-interference list is concurrently counted. If it is not counted, polling or concurrency is performed through the RFID antenna that is not in the interference list.
[0195] It should be noted that the above-mentioned target tag is the second RFID tag to be located.
[0196] Therefore, the RFID antenna inventory can be performed according to the above-determined scheduling mode, and the radio frequency signal can be sent, so that the second tag data returned by the second RFID tag can be obtained.
[0197] In summary, the RFID positioning method described in the embodiment of the present invention deeply mines the potential feature vector set and eliminates the potential feature vectors that are more sensitive to noise, determines the feature extraction rules, and obtains the environmental feature vector corresponding to the reference tag according to the feature extraction rules, and obtains the second tag feature vector corresponding to the second RFID tag, trains the positioning model, and realizes intelligent regional level positioning in a complex interference environment; moreover, by measuring the difference between the weighting function corresponding to the first RFID tag and the second tag feature vector used in the positioning model training, the value parameters of the weighting function are determined, and the feature vector corresponding to the first RFID tag is weighted and reconstructed. The weighted and reconstructed third tag feature vector can directly apply the trained positioning model to realize intelligent positioning, without retraining the positioning model, to achieve adaptive effect and reduce application cost. Moreover, the positioning model does not need to be retrained, which solves the problem of insufficient technical resources of the terminal equipment deployed by the positioning model.
[0198] like Figure 6As shown, an embodiment of the present invention further provides an RFID positioning device, comprising:
[0199] A first acquisition module 601 is used to acquire first tag data of a first RFID tag when the RFID antenna sends a radio frequency signal;
[0200] A first determination module 602, configured to determine a corresponding weighting function according to a first label feature vector corresponding to the first label data;
[0201] A second determination module 603 is used to measure the difference between the weighting function and the second label feature vector to determine the value parameter of the weighting function, where the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model;
[0202] A third determination module 604 is used to determine a third label feature vector corresponding to the first label data according to the weighting function and the value parameter;
[0203] The first obtaining module 605 is used to obtain the location information of the first RFID tag according to the third tag feature vector and the positioning model.
[0204] Optionally, in the RFID positioning device, the second determining module 603 is specifically configured to:
[0205] Obtaining a difference metric between the weighting function and a second label feature vector;
[0206] Taking minimization of the difference metric as an iteration goal, the weighting function is iterated to determine the value parameters of the weighting function.
[0207] Optionally, the RFID positioning device further comprises:
[0208] A second acquisition module, used to acquire third tag data of the reference tag when the RFID antenna sends a radio frequency signal;
[0209] A training module is used to train a positioning model according to the second label feature vector and the environment feature vector corresponding to the third label data.
[0210] Optionally, the RFID positioning device further comprises:
[0211] A third acquisition module, used for acquiring the second tag data of the second RFID tag when the RFID antenna sends a radio frequency signal;
[0212] The fourth acquisition module is used to obtain the second label feature vector according to the feature extraction rule and the second label data, wherein the feature extraction rule is determined based on the sensitivity of each potential feature vector in the potential feature vector set to noise, and the potential feature vector set is obtained based on the second label data and the third label data.
[0213] Optionally, the RFID positioning device further comprises:
[0214] A second obtaining module is used to obtain a first importance score of each potential feature vector in the potential feature vector set according to the first machine learning model and the potential feature vector set;
[0215] a fifth acquisition module, configured to perform noise processing on the potential feature vector set to obtain a processed potential feature vector set;
[0216] A third obtaining module, configured to obtain a second importance score for each potential feature vector in the processed potential feature vector set according to the first machine learning model and the processed potential feature vector set;
[0217] a fourth determination module, configured to determine the sensitivity of each potential feature vector to noise according to the first importance score and the second importance score;
[0218] The fifth determination module is used to determine the feature extraction rule based on the sensitivity of each potential feature vector to noise, and the feature extraction rule is used to indicate to eliminate the potential feature vector with high sensitivity to noise.
[0219] Optionally, the RFID positioning device further comprises:
[0220] A sixth acquisition module is used to acquire an original feature vector set according to the second label data and the third label data, wherein the original feature vector set includes a feature vector corresponding to at least one of the following data:
[0221] Received signal strength indication RSSI;
[0222] Initial phase;
[0223] Arrival phase;
[0224] Frequency;
[0225] Timestamp;
[0226] A seventh acquisition module is used to acquire the potential feature vector set based on the original feature vector set, wherein the potential feature vector set includes potential feature vectors corresponding to at least one of the following data within the inventory time:
[0227] Number of reads per RFID antenna per tag;
[0228] The average RSSI value of each tag single RFID antenna;
[0229] RSSI variance of a single RFID antenna per tag.
[0230] Optionally, in the RFID positioning device, the training module is specifically used for:
[0231] Training a positioning model according to the environmental feature vector corresponding to the third label data to obtain a positioning model with a weight of a first value;
[0232] Training the positioning model with the first weight according to the second label feature vector to obtain a positioning model with the second weight;
[0233] Whether the positioning model has completed training is determined according to the positioning accuracy of the positioning model whose weight is the second value.
[0234] Optionally, in the RFID positioning device, the second acquisition module is specifically used to:
[0235] Based on the location information of the reference tag and the coverage of each RFID antenna, selecting an RFID antenna combination from the plurality of RFID antennas, wherein the coverage of any two RFID antennas in the RFID antenna combination does not overlap;
[0236] The third tag data of the reference tag is acquired when all RFID antennas in the RFID antenna combination simultaneously transmit radio frequency signals, wherein the radio frequency signals carry an inventory mode instruction.
[0237] Optionally, in the RFID positioning device, the third acquisition module is specifically used to:
[0238] Determining a scheduling mode according to the number of the second RFID tags;
[0239] When the RFID antenna sends a radio frequency signal according to the scheduling mode, second tag data of the second RFID tag is acquired.
[0240] The RFID positioning device provided in the embodiment of the present invention can execute the above-mentioned RFID positioning method embodiment, and its implementation principle and technical effect are similar, which will not be described in detail in this embodiment.
[0241] like Figure 7As shown, the RFID positioning device of an embodiment of the present invention includes: a processor 701; and a memory 702 connected to the processor 701 through a bus interface, the memory 702 is used to store programs and data used by the processor 701 when performing operations, and the processor 701 calls and executes the programs and data stored in the memory 702.
[0242] The processor 701 is used to read the program in the memory 702 and execute the following process:
[0243] When the RFID antenna sends a radio frequency signal, obtaining first tag data of the first RFID tag;
[0244] Determining a corresponding weighting function according to a first label feature vector corresponding to the first label data;
[0245] Performing a difference measurement between the weighting function and a second label feature vector to determine a value parameter of the weighting function, wherein the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model;
[0246] Determining a third label feature vector corresponding to the first label data according to the weighting function and the value parameter;
[0247] The location information of the first RFID tag is obtained according to the third tag feature vector and the positioning model.
[0248] The transceiver 703 is used to receive and send data in the controller of the processor 701.
[0249] Among them, Figure 7 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 701 and various circuits of memory represented by memory 702 are linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 703 may be a plurality of components, namely, a transmitter and a transceiver, providing a unit for communicating with various other devices on a transmission medium. For different user devices, the user interface 704 may also be an interface capable of externally and internally connecting required devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.
[0250] The processor 701 is responsible for managing the bus architecture and general processing, and the memory 702 can store data used by the processor 701 when performing operations.
[0251] Optionally, the processor 701 is specifically configured to read the computer program and execute the following steps:
[0252] Obtaining a difference metric between the weighting function and a second label feature vector;
[0253] Taking minimization of the difference metric as an iteration goal, the weighting function is iterated to determine the value parameters of the weighting function.
[0254] Optionally, the processor 701 is further configured to read the computer program and execute the following steps:
[0255] When the RFID antenna sends a radio frequency signal, obtaining third tag data of the reference tag;
[0256] A positioning model is trained according to the second label feature vector and the environment feature vector corresponding to the third label data.
[0257] Optionally, the processor 701 is further configured to read the computer program and execute the following steps:
[0258] When the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag;
[0259] The second label feature vector is obtained according to a feature extraction rule and the second label data, wherein the feature extraction rule is determined based on the sensitivity of each potential feature vector in a potential feature vector set to noise, and the potential feature vector set is obtained based on the second label data and the third label data.
[0260] Optionally, the processor 701 is further configured to read the computer program and execute the following steps:
[0261] Obtaining a first importance score for each potential feature vector in the potential feature vector set according to the first machine learning model and the potential feature vector set;
[0262] Performing noise processing on the potential feature vector set to obtain a processed potential feature vector set;
[0263] Obtaining a second importance score for each potential feature vector in the processed set of potential feature vectors based on the first machine learning model and the processed set of potential feature vectors;
[0264] Determining the sensitivity of each potential feature vector to noise according to the first importance score and the second importance score;
[0265] The feature extraction rule is determined based on the sensitivity of each potential feature vector to noise, and the feature extraction rule is used to indicate to remove the potential feature vectors with high sensitivity to noise.
[0266] Optionally, the processor 701 is further configured to read the computer program and execute the following steps:
[0267] According to the second label data and the third label data, an original feature vector set is obtained, where the original feature vector set includes a feature vector corresponding to at least one of the following data:
[0268] Received signal strength indication RSSI;
[0269] Initial phase;
[0270] Arrival phase;
[0271] Frequency;
[0272] Timestamp;
[0273] Based on the original feature vector set, the potential feature vector set is obtained, and the potential feature vector set includes potential feature vectors corresponding to at least one of the following data within the inventory time:
[0274] Number of reads per RFID antenna per tag;
[0275] The average RSSI value of each tag single RFID antenna;
[0276] RSSI variance of a single RFID antenna per tag.
[0277] Optionally, the processor 701 is specifically configured to read the computer program and execute the following steps:
[0278] Training a positioning model according to the environmental feature vector corresponding to the third label data to obtain a positioning model with a weight of a first value;
[0279] Training the positioning model with the first weight according to the second label feature vector to obtain a positioning model with the second weight;
[0280] Whether the positioning model has completed training is determined according to the positioning accuracy of the positioning model whose weight is the second value.
[0281] Optionally, the processor 701 is specifically configured to read the computer program and execute the following steps:
[0282] Based on the location information of the reference tag and the coverage of each RFID antenna, selecting an RFID antenna combination from the plurality of RFID antennas, wherein the coverage of any two RFID antennas in the RFID antenna combination does not overlap;
[0283] The third tag data of the reference tag is acquired when all RFID antennas in the RFID antenna combination simultaneously transmit radio frequency signals, wherein the radio frequency signals carry an inventory mode instruction.
[0284] Optionally, the processor 701 is specifically configured to read the computer program and execute the following steps:
[0285] Determining a scheduling mode according to the number of the second RFID tags;
[0286] When the RFID antenna sends a radio frequency signal according to the scheduling mode, second tag data of the second RFID tag is acquired.
[0287] A specific embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the above-mentioned RFID positioning method are implemented and the same technical effect can be achieved. To avoid repetition, it will not be described here.
[0288] In addition, an embodiment of the present invention further provides a computer program product, including computer instructions, which, when executed by a processor, implement the above Figure 1 The various processes of the method embodiment shown can achieve the same technical effect, and will not be described again here to avoid repetition.
[0289] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0290] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may be physically included separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.
[0291] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform some steps of the sending and receiving methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.
[0292] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A radio frequency identification (RFID) positioning method, characterized in that: include: When the RFID antenna sends a radio frequency signal, obtaining first tag data of the first RFID tag; Determining a corresponding weighting function according to a first label feature vector corresponding to the first label data; Performing a difference measurement between the weighting function and a second label feature vector to determine a value parameter of the weighting function, wherein the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model; Determining a third label feature vector corresponding to the first label data according to the weighting function and the value parameter; The location information of the first RFID tag is obtained according to the third tag feature vector and the positioning model.
2. The RFID positioning method according to claim 1, characterized in that: The weighting function and the second label feature vector are measured for difference to determine the value parameters of the weighting function, including: Obtaining a difference metric between the weighting function and a second label feature vector; Taking minimization of the difference metric as an iteration goal, the weighting function is iterated to determine the value parameters of the weighting function.
3. The RFID positioning method according to claim 1, characterized in that: The method further comprises: When the RFID antenna sends a radio frequency signal, obtaining third tag data of the reference tag; A positioning model is trained according to the second label feature vector and the environment feature vector corresponding to the third label data.
4. The RFID positioning method according to claim 3, characterized in that: The method further comprises: When the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag; The second label feature vector is obtained according to a feature extraction rule and the second label data, wherein the feature extraction rule is determined based on the sensitivity of each potential feature vector in a potential feature vector set to noise, and the potential feature vector set is obtained based on the second label data and the third label data.
5. The RFID positioning method according to claim 4, characterized in that: The method further comprises: Obtaining a first importance score for each potential feature vector in the potential feature vector set according to the first machine learning model and the potential feature vector set; Performing noise processing on the potential feature vector set to obtain a processed potential feature vector set; Obtaining a second importance score for each potential feature vector in the processed set of potential feature vectors based on the first machine learning model and the processed set of potential feature vectors; Determining the sensitivity of each potential feature vector to noise according to the first importance score and the second importance score; The feature extraction rule is determined based on the sensitivity of each potential feature vector to noise, and the feature extraction rule is used to indicate to remove the potential feature vectors with high sensitivity to noise.
6. The RFID positioning method according to claim 4, characterized in that: The method further comprises: According to the second label data and the third label data, an original feature vector set is obtained, where the original feature vector set includes a feature vector corresponding to at least one of the following data: Received signal strength indication RSSI; Initial phase; Arrival phase; Frequency; Timestamp; Based on the original feature vector set, the potential feature vector set is obtained, and the potential feature vector set includes potential feature vectors corresponding to at least one of the following data within the inventory time: Number of reads per RFID antenna per tag; The average RSSI value of each tag single RFID antenna; RSSI variance of a single RFID antenna per tag.
7. The RFID positioning method according to claim 3, characterized in that: Training a positioning model according to the second label feature vector and the environment feature vector corresponding to the third label data includes: Training a positioning model according to the environmental feature vector corresponding to the third label data to obtain a positioning model with a weight of a first value; Training the positioning model with the first weight according to the second label feature vector to obtain a positioning model with the second weight; Whether the positioning model has completed training is determined according to the positioning accuracy of the positioning model whose weight is the second value.
8. The RFID positioning method according to claim 3, characterized in that: When the RFID antenna sends a radio frequency signal, obtaining third tag data of the reference tag includes: Based on the location information of the reference tag and the coverage of each RFID antenna, selecting an RFID antenna combination from the plurality of RFID antennas, wherein the coverage of any two RFID antennas in the RFID antenna combination does not overlap; The third tag data of the reference tag is acquired when all RFID antennas in the RFID antenna combination simultaneously transmit radio frequency signals, wherein the radio frequency signals carry an inventory mode instruction.
9. The RFID positioning method according to claim 4, characterized in that: When the RFID antenna sends a radio frequency signal, obtaining second tag data of the second RFID tag includes: Determining a scheduling mode according to the number of the second RFID tags; When the RFID antenna sends a radio frequency signal according to the scheduling mode, second tag data of the second RFID tag is acquired.
10. An RFID positioning device, characterized in that: include: A first acquisition module, configured to acquire first tag data of a first RFID tag when the RFID antenna sends a radio frequency signal; A first determining module, configured to determine a corresponding weighting function according to a first label feature vector corresponding to the first label data; A second determination module is used to measure the difference between the weighting function and the second label feature vector to determine the value parameter of the weighting function, where the second label feature vector is a feature vector corresponding to the second label data used when training the positioning model; A third determination module, used to determine a third label feature vector corresponding to the first label data according to the weighting function and the value parameter; The first obtaining module is used to obtain the location information of the first RFID tag according to the third tag feature vector and the positioning model.
11. An RFID positioning device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the RFID positioning method according to any one of claims 1 to 9 is implemented.
12. A readable storage medium, characterized in that: The readable storage medium stores a program, and when the program is executed by a processor, the RFID positioning method according to any one of claims 1 to 9 is implemented.
13. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the RFID positioning method according to any one of claims 1 to 9.
