Model training method and device, positioning method and device, electronic equipment, storage medium and computer program product

Abstract

The invention discloses a model training method and device, a positioning method and device, electronic equipment, a storage medium and a computer program product, and the model training method comprises the steps: constructing a training sample set based on a first data set; wherein the first data set comprises signal quality parameters and phases of reflected signals of a plurality of passive RFID tags under the conditions that a moving body passes through the passive RFID tags and no moving body passes through the passive RFID tags; each training sample in the training sample set comprises first data corresponding to a plurality of passive RFID tags at the same moment; the first data comprises second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where the passive RFID tag through which the moving body passes is located; and training a first model based on the training sample set, wherein the first model is used for predicting an area where a moving body is located based on an input training sample.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a model training method and device, a positioning method and device, an electronic device, a storage medium, and a computer program product. Background Art

[0002] Radio Frequency Identification (RFID) technology is one of the mainstream wireless communication technologies. It has the characteristics of information carrying and information identification, which makes RFID-based positioning technology gain attention. Existing RFID-based positioning mostly uses RFID tags placed on mobile objects, or trains fixed reference RFID tags and performs positioning based on the analogy results between the reference RFID tags and the RFID tags to be tested. However, the mobile objects will interfere with the signal transmission process of the RFID tags, resulting in the problem of inaccurate dynamic positioning in these RFID-based positioning methods. Summary of the invention

[0003] To solve related technical problems, the embodiments of the present application provide a model training method and device, a positioning method and device, an electronic device, a storage medium, and a computer program product.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] The present application embodiment provides a model training method, which is applied to a first electronic device, and the method includes:

[0006] A training sample set is constructed based on a first data set; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located;

[0007] A first model is trained based on the training sample set, where the first model is used to predict the area where the moving object is located based on the input training samples.

[0008] In the above solution, the step of constructing a training sample set based on the first data set includes:

[0009] Determining a second data set and / or a third data set and / or a fourth data set based on the first data set, wherein one data in the second data set represents a difference between a signal quality parameter of a reflection signal of a passive RFID tag and a first value, the first value represents an average value of the signal quality parameter of the reflection signals of all passive RFID tags without a moving body passing through, one data in the third data set represents a difference between a phase of a reflection signal of a passive RFID tag and a second value, the second value represents an average value of the phase of the reflection signals of all passive RFID tags without a moving body passing through, and one data in the fourth data set represents a trigonometric function value of the phase of the reflection signal of a passive RFID tag;

[0010] The training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set.

[0011] In the above solution, constructing the training sample set based on the first data set, and based on the second data set and the third data set includes:

[0012] Based on the fitting function and the second data set, a fifth data set is obtained; the fitting function is obtained by fitting the second data set and the third data set;

[0013] Based on the data of the same passive RFID tag at the same time corresponding to the fifth data set and the third data set, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal;

[0014] constructing a sixth data set based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase;

[0015] The training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set.

[0016] In the above scheme, the sixth data set includes the seventh data set and / or the eighth data set; the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

[0017] In the above scheme, the determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal based on the data of the same passive RFID tag corresponding to the same time in the fifth data set and the third data set includes:

[0018] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold, determining that the phase of the reflection signal of the same passive RFID tag is abnormal;

[0019] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal; wherein,

[0020] The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

[0021] In the above solution, the corrected phase of the reflection signal of the same passive RFID tag with phase anomaly is obtained by subtracting 180 degrees from the phase of the reflection signal of the same passive RFID tag.

[0022] In the above scheme, the method further includes:

[0023] The parameters of the set layer of the trained first model are adjusted based on the set sample to obtain the final first model; the set sample is determined from the training sample set based on the loss value corresponding to each training sample in the training sample set.

[0024] In the above scheme, the method further includes:

[0025] Calculate the difference between the third value and the fourth value to obtain a fifth value; the third value represents the loss value corresponding to each training sample in the training sample set, and the fourth value is the average loss value corresponding to the training sample set;

[0026] When the fifth value corresponding to the training sample is smaller than the second threshold and larger than the third threshold, the training sample is determined as the set sample, and the second threshold is smaller than the third threshold.

[0027] The embodiment of the present application further provides a positioning method, which is applied to a second electronic device, and the method includes:

[0028] Processing signal quality parameters and phases of reflected signals of multiple passive RFID tags received by the terminal to obtain first information;

[0029] Inputting the first information into a second model to obtain second information, where the second information represents an area where the terminal is located, and the second model is trained based on any one of the methods on the first electronic device side;

[0030] The second information is output or displayed.

[0031] The present application also provides a model training device, which includes:

[0032] A sample construction module is used to construct a training sample set based on a first data set; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located;

[0033] The training module is used to train a first model based on the training sample set, wherein the first model is used to predict the area where the moving object is located based on the input training samples.

[0034] The present application also provides a positioning device, the device comprising:

[0035] A processing module, configured to process the signal quality parameters and phases of the reflected signals of the plurality of passive RFID tags received by the terminal to obtain first information;

[0036] An input module, configured to input the first information into a second model to obtain second information, wherein the second information represents an area where the terminal is located, and the second model is obtained by training based on any one of the methods on the first electronic device side;

[0037] An output module is used to output or display the second information.

[0038] The embodiment of the present application further provides a first electronic device, comprising: a first processor and a first communication interface; wherein,

[0039] The first processor is used to construct a training sample set based on a first data set; train a first model based on the training sample set, and the first model is used to predict the area where a moving body is located based on the input training samples; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tags and when no moving body passes through the passive RFID tags; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag through which a moving body passes is located.

[0040] The embodiment of the present application further provides a second electronic device, including: a second processor and a second communication interface; wherein,

[0041] The second processor is used to process the signal quality parameters and phases of the reflected signals of multiple passive RFID tags received by the terminal to obtain first information; input the first information into a second model to obtain second information; output or display the second information; the second information represents the area where the terminal is located, and the second model is trained based on any method on the first electronic device side mentioned above.

[0042] An embodiment of the present application also provides an electronic device, including a processor and a memory for storing a computer program that can be run on the processor, wherein the processor, when used to run the computer program, executes the steps of any method on the first electronic device side, or executes the steps of any method on the second electronic device side.

[0043] An embodiment of the present application further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program implements the steps of any method on the first electronic device side or implements the steps of any method on the second electronic device side.

[0044] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any method on the first electronic device side, or implements the steps of any method on the second electronic device side.

[0045] In the model training method and device, positioning method and device, electronic device, storage medium, and computer program product provided by the present application, a training sample set is constructed based on a first data set; the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a mobile body passes through the passive RFID tag and when no mobile body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a mobile body passing through is located; the first model is trained based on the training sample set, and the first model is used to predict the area where the mobile body is located based on the input training sample. In the above scheme, the training sample set includes a first data set, and the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag. The training sample set takes into account the influence of the passing of a moving body and the non-passing of a moving body on the reflection signal of the passive RFID tag, and determines the second data based on the data in the first data set, enriches the data included in the training samples in the training sample set, and trains the first model based on the training sample set, so that the accuracy of dynamic positioning of the trained first model can be higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of the model training method according to an embodiment of the present application;

[0047] Figure 2 This is a schematic diagram of the model structure of the embodiment of the present application;

[0048] Figure 3 A schematic diagram of a process for constructing a sixth data set for an embodiment of the present application;

[0049] Figure 4 A schematic diagram of the training process of the first model after training in an embodiment of the present application;

[0050] Figure 5 This is a flow chart of the positioning method according to an embodiment of the present application;

[0051] Figure 6 This is a schematic diagram of the layout of the passive RFID tag in the embodiment of the present application;

[0052] Figure 7 A schematic diagram of the overall model training and application process of the embodiment of the present application;

[0053] Figure 8This is a schematic diagram of the structure of a model training device according to an embodiment of the present application;

[0054] Fig. 9 This is a schematic diagram of the structure of a positioning device according to an embodiment of the present application;

[0055] Fig.10 This is a schematic diagram of the structure of a first electronic device according to an embodiment of the present application;

[0056] Fig.11 This is a schematic diagram of the structure of a second electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0057] As one of the mainstream wireless communication technologies, RFID technology has the characteristics of information carrying and information identification, and has been applied to many IoT scenarios. Among them, passive RFID tags generate power for the chip by coupling energy from electromagnetic waves emitted by the reader, without the need to carry their own power supply, and have the advantage of low cost, becoming one of the research hotspots in recent years.

[0058] Personnel positioning is a challenging technical issue in IoT scenarios. For example, in urban pipe corridor scenarios, water, gas and other pipelines are mostly deployed underground, and patrol personnel are required to check the pipelines regularly. However, since there is no signal coverage underground, it is difficult to get in touch with the outside world in real time when encountering security issues. At this time, RFID technology can be used to locate patrol personnel. For example, in confidential electronic equipment production scenarios, personnel need to be managed, but they cannot carry communication equipment to get in touch in real time. RFID technology can also be used to locate personnel.

[0059] There are methods that use computer vision to locate people, but these methods have large models, complex algorithms, and high requirements for device computing power. On the one hand, they lead to high costs, and on the other hand, they lead to high energy consumption, which is not conducive to the deployment of terminal equipment. There are four technical solutions for using RFID technology to locate people:

[0060] The positioning area is divided into multiple sub-areas, and the coordinate systems of the RFID positioning system and the visual positioning system are aligned to the same world coordinate system; based on the RFID positioning system, the RFID positioning of the pedestrian to be positioned is performed in real time; and based on the visual positioning system, the visual positioning of the pedestrian to be positioned is performed at set time intervals; it is determined whether the visual positioning result and the RFID positioning result at the same time are in the same sub-area: if so, the RFID positioning is determined to be valid, and the RFID positioning of the pedestrian to be positioned is performed in real time based on the RFID positioning system; if not, the RFID positioning is determined to be invalid, and the visual positioning of the pedestrian to be positioned is performed in real time based on the visual positioning system;

[0061] An RFID positioning transponder is arranged at intervals along the longitudinal direction of the tunnel; the location information of the RFID positioning transponder is stored in the chip of the RFID positioning transponder; an RFID reader is arranged on the underwater robot, the RFID reader transmits electromagnetic wave signals, and receives the location information of the RFID positioning transponder fed back from all RFID positioning transponders within the effective working area; data processing is performed based on the transmission signal data of the RFID reader and the received data, combined with the height data of the tunnel itself, to determine the location information of the underwater robot;

[0062] Read the card reader coordinate data, reference tag coordinate data and the real coordinate data of the test point, and obtain the measured coordinates and positioning error data of the test point by using the LANDMARC (Location Identification Based on Dynamic Active RFID Calibration) algorithm, BVIRE (Boundary Virtual Label Algorithm) algorithm and VIRE (Virtual Reference Elimination) algorithm; extract the timestamp and main error features from the measured coordinates and positioning error data of the test point to construct a training set; correct the minimum error position point coordinates and the minimum error position point time, and output the corrected value of the test point coordinates and the corresponding maximum possible position coordinates;

[0063] The positions of the reader and reference tags are arranged, and the distance from the reference tag to the reader is calculated; the reader sends an electromagnetic wave signal to the reference tag and receives the signal strength (RSSI, Received Signal Strength Indication) value to construct a training sample vector matrix; an adaptive deep belief network is constructed, with the RSSI value of each reference tag as the input value and the distance as the output value; the contrastive divergence algorithm is used to complete the pre-training of network parameters; the adaptive moment estimation method is used to adjust the weights of each layer of the deep learning network; the reader sends a signal to the tag to be tested and receives the RSSI value, and the deep belief network is used to predict the position of the tag to be tested.

[0064] There are three types of existing technical solutions for personnel positioning using RFID technology and the corresponding problems:

[0065] The first type of technical solution is to combine computer vision information and RFID tag information for hybrid positioning. By using the mature technology of visual positioning and RFID information to make joint decisions, the positioning accuracy can be improved. However, this type of technical solution introduces other technologies to assist in positioning, which will increase equipment costs, energy consumption, and computing power requirements, which is not conducive to terminal deployment and application.

[0066] The second technical solution is to place RFID tags on mobile objects such as robots, and analyze the changes in electromagnetic wave signals to locate the mobile objects. However, since RFID tags are very small, they are easily lost when placed on mobile objects (such as people and robots), which has a great impact on the positioning effect. At the same time, the structure of the mobile object itself is likely to interfere with the reflection, refraction and scattering of the signal during the propagation process, affecting the accuracy and stability of the returned data. Deploying RFID tags on mobile objects is prone to uncontrollable factors, which is not conducive to positioning accuracy.

[0067] The third type of technical solution is to place a reference tag in the target area, use the RSSI, phase, distance from the reader and other information generated by the electromagnetic wave signal of the reference tag to construct a training sample, and build a corresponding network model to achieve the positioning of the tag to be tested; however, this type of technical solution achieves the purpose of locating the tag to be tested by analogy with the reference tag to be tested. The environment and status of the reference tag and the tag to be tested should be consistent. It is aimed at the positioning of static tags and is not suitable for personnel positioning scenarios.

[0068] Based on this, in various embodiments of the present application, a training sample set is constructed based on a first data set; the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag where a moving body passes is located; a first model is trained based on the training sample set, and the first model is used to predict the area where the moving body is located based on the input training sample. In the above scheme, the training sample set includes a first data set, and the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag. The training sample set takes into account the influence of the passing of a moving body and the non-passing of a moving body on the reflection signal of the passive RFID tag, and determines the second data based on the data in the first data set, enriches the data included in the training samples in the training sample set, and trains the first model based on the training sample set, so that the accuracy of dynamic positioning of the trained first model can be higher.

[0069] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0070] The present application embodiment provides a model training method, which is applied to a first electronic device. Figure 1 As shown, the method includes:

[0071] Step 101: construct a training sample set based on a first data set.

[0072] Among them, the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located.

[0073] Here, the first data set includes the signal quality parameters and phases of the reflected signals of multiple passive RFID tags when there is a moving body passing by the passive RFID tag and when there is no moving body passing by the passive RFID tag; the moving body can be a person or an object, without specific limitation; the signal quality parameter can be an RSSI value or other parameters reflecting the signal quality, without specific limitation. It is worth noting that the passive RFID tag is fixed to the wall of the area to be located or other places in the area to be located that are easy to fix the passive RFID tag; each passive RFID tag corresponds to an area in the area to be located, and different passive RFID tags correspond to different areas.

[0074] A training sample set is constructed based on the first data set, each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time, and the first data includes second data and / or data in the first data set; the second data is determined based on the data in the first data set, specifically, data of reflected signals of passive RFID tags without moving objects passing through can be used to correct the data in the first data set to obtain the second data; trigonometric function processing can be performed on the phase data in the first data set to obtain the second data; and phase abnormality data in the first data set can also be determined and corrected to obtain the second data.

[0075] It should be noted that each training sample in the training sample set includes the first data corresponding to multiple passive RFID tags at the same time, and each training sample can be regarded as a class image sample. Each row of each training sample can be a different feature of the same passive RFID tag, namely, the first data, which can be compared to a patch in an image, and each column can be the same feature of different passive RFID tags, namely, the same class of first data, which can be compared to a channel in an image, which can help to effectively extract the features of the training samples.

[0076] Step 102: Train the first model based on the training sample set.

[0077] The first model is used to predict the area where the moving object is located based on the input training samples.

[0078] Here, the input of the first model is the training sample, and the output is the area where the moving object is located; the first model is a classification model, that is, the input training sample is classified into an area where a passive RFID tag is located, and the area where the moving object is located is output as the predicted area. The first model can be a fully connected neural network MLP-Mixer model, and the model structure diagram is as follows: Figure 2 shown. Figure 2 Each structure in the dotted box constitutes a basic module, and there are Q basic modules in total. The "C×D" before the basic module represents the training sample, C represents the number of rows of each training sample corresponding to the number of passive RFID tags, and D represents the number of columns of each training sample corresponding to the number of categories of the first data. The basic module includes the pooling layer Pool and M fully connected layers FC in sequence, and finally outputs the classification result, that is, the predicted area where the moving object is located. Among them, Q is greater than or equal to 1, and M is greater than or equal to 1; two fusions are performed in a basic module, namely the fusion between different rows of training samples and the fusion between different columns of training samples, and each fusion passes through the normalization layer LN, the fully connected layer FC, the activation function layer, the fully connected layer FC and the fusion layer; the fusion between different rows of training samples is to change C to C', and the fusion between different columns of training samples is to change D to D'. Specifically, Q, M, C', D' and the activation function can be set according to actual needs.

[0079] The first model is trained based on the training sample set until a first set condition is met to obtain a trained first model; the first set condition may be reaching a first set number of iterations, the loss function value may be less than or equal to a first set threshold, or the accuracy of the first model may be greater than or equal to a second set threshold.

[0080] In this embodiment, the training sample set includes a first data set, which includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag. The training sample set takes into account the impact of the passing of a moving body and the absence of a moving body on the reflection signal of the passive RFID tag, and determines the second data based on the data in the first data set, enriches the data included in the training samples in the training sample set, and trains the first model based on the training sample set, so that the accuracy of dynamic positioning of the trained first model can be higher.

[0081] In one embodiment, constructing a training sample set based on the first data set includes:

[0082] Determining a second data set and / or a third data set and / or a fourth data set based on the first data set, wherein one data in the second data set represents a difference between a signal quality parameter of a reflection signal of a passive RFID tag and a first value, the first value represents an average value of the signal quality parameter of the reflection signals of all passive RFID tags without a moving body passing through, one data in the third data set represents a difference between a phase of a reflection signal of a passive RFID tag and a second value, the second value represents an average value of the phase of the reflection signals of all passive RFID tags without a moving body passing through, and one data in the fourth data set represents a trigonometric function value of the phase of the reflection signal of a passive RFID tag;

[0083] The training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set.

[0084] Here, the first value may be the average value of the signal quality parameter of the reflected signals of all passive RFID tags that no moving body passes through in the first set continuous time period, and the second value may be the average value of the phase of the reflected signals of all passive RFID tags that no moving body passes through in the second set continuous time period. One data in the fourth data set may represent the phase cosine value and / or phase sine value of the reflected signal of a passive RFID tag.

[0085] A training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set. Specifically, each training sample in the training sample set includes the first data corresponding to multiple passive RFID tags at the same time, the first data includes the second data and the data in the first data set, and the second data includes the data in the second data set and / or the data in the third data set and / or the data in the fourth data set.

[0086] In this embodiment, the data of the reflected signal without a moving object passing through the passive RFID tag, i.e., the first value and the second value, are used to correct the data in the first data set to obtain the data in the second data set and the data in the third data set, which can enrich the data included in the training samples in the training sample set and improve the accuracy of the training samples. The data in the fourth data set compresses the phase mapping range, which can reduce the error caused by phase instability.

[0087] In one embodiment, constructing the training sample set based on the first data set, and based on the second data set and the third data set includes:

[0088] Based on the fitting function and the second data set, a fifth data set is obtained; the fitting function is obtained by fitting the second data set and the third data set;

[0089] Based on the data of the same passive RFID tag at the same time corresponding to the fifth data set and the third data set, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal;

[0090] constructing a sixth data set based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase;

[0091] The training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set.

[0092] Here, the fitting function is obtained by fitting the second data set and the third data set. There is no limitation on the way of fitting the second data set and the third data set, for example, linear function fitting, nonlinear function fitting, or fully connected network fitting. In practical applications, constructing a multi-layer fully connected network for fitting can ensure a good fitting effect. It should be understood that the change in the signal quality parameter of the reflected signal of the same passive RFID tag at the same time is related to the change in the phase, so the phase abnormality or normal phase is determined by fitting the function.

[0093] Based on the fitting function and the second data set, a fifth data set is obtained. Specifically, the data in the second data set can be sequentially input into the fitting function, and the output data can be used to construct the fifth data set. The data in the second data set can also be sequentially input into the fitting function, and the fifth data set can be constructed by multiplying the output data and the first set deviation.

[0094] Based on the data of the same passive RFID tag at the same time in the fifth data set and the third data set corresponding to the same passive RFID tag, it is determined whether the phase of the reflected signal of the same passive RFID tag is normal or abnormal. Specifically, it can be determined whether the phase of the reflected signal of the same passive RFID tag is normal or abnormal based on the difference between the data of the same passive RFID tag at the same time in the fifth data set and the third data set corresponding to the same passive RFID tag; it can also be determined whether the phase of the reflected signal of the same passive RFID tag is normal or abnormal based on the product of the difference between the data of the same passive RFID tag at the same time in the fifth data set and the third data set corresponding to the same passive RFID tag and the second set deviation.

[0095] For example, when the fifth data corresponding to the same passive RFID tag is less than or equal to the fourth threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal;

[0096] When the fifth data corresponding to the same passive RFID tag is greater than the fourth threshold, the phase of the reflected signal of the same passive RFID tag is determined to be abnormal; wherein,

[0097] The fifth data represents the product of the difference between the data of the same passive RFID tag in the fifth data set and the data of the third data set at the same time and the second set deviation. The fourth threshold value can be set according to actual needs.

[0098] In one embodiment, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal based on the data of the same passive RFID tag at the same time in the fifth data set and the third data set includes:

[0099] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold, determining that the phase of the reflection signal of the same passive RFID tag is abnormal;

[0100] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal; wherein,

[0101] The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

[0102] It is worth noting that the first threshold can be set according to actual needs.

[0103] In this embodiment, the conditions for whether the phase of the reflected signal of the same passive RFID tag is normal or abnormal are clearly determined, and the efficiency of determining whether the phase of the reflected signal of the same passive RFID tag is normal or abnormal is improved, so as to help improve the accuracy of the data in the training samples.

[0104] Since there is phase jump noise in the hardware, that is, the phase change in the reflected signal of the passive RFID tag may not be caused by the passing of a moving object, but by the hardware causing the phase to jump by 180 degrees. These phases caused by the hardware causing the phase to jump by 180 degrees are abnormal phases, which will seriously affect the accuracy of the first model. In the case of determining that the phase of the reflected signal of the same passive RFID tag is abnormal, the phase of the reflected signal of the same passive RFID tag with abnormal phase needs to be corrected. Based on this, in one embodiment, the corrected phase of the reflected signal of the same passive RFID tag with abnormal phase is obtained by subtracting 180 degrees from the phase of the reflected signal of the same passive RFID tag.

[0105] In this embodiment, the correction method of the phase of the reflected signal of the same passive RFID tag with phase abnormality is clarified to improve the phase accuracy, thereby helping to improve the accuracy of the first model.

[0106] Based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase, a sixth data set is constructed. Specifically, when the phases of the reflection signals of all passive RFID tags are normal, the sixth data set includes the phase of the reflection signal of the same passive RFID tag with normal phase and / or the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value. When the phases of the reflection signals of all passive RFID tags are abnormal, the sixth data set includes the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value. When the reflection signal of the passive RFID tag has abnormal phase and normal phase, the sixth data set includes the phase of the reflection signal of the same passive RFID tag with normal phase and / or the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, as well as the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value. The second value represents the average value of the phases of the reflected signals of all passive RFID tags through which no moving body passes.

[0107] A training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set. Specifically, each training sample in the training sample set includes the first data corresponding to multiple passive RFID tags at the same time, the first data includes the second data and the data in the first data set, and the second data includes the data in the second data set, the data in the third data set and the data in the sixth data set.

[0108] In this embodiment, based on the fitting function obtained by fitting the second data set and the third data set, the data with phase anomaly is determined, and the data with phase anomaly is corrected to construct a sixth data set. A training sample set is constructed based on the sixth data set, which can effectively reduce the impact of noise on the model and help improve the accuracy of the first model.

[0109] In one embodiment, the sixth data set includes the seventh data set and / or the eighth data set; the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

[0110] Specifically, the sixth data set includes the seventh data set, or includes the eighth data set, or includes the seventh data set and the eighth data set. When the phases of the reflection signals of all passive RFID tags are normal, the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase, and the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value. When the phases of the reflection signals of all passive RFID tags are abnormal, the seventh data set is constructed based on the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase, and the eighth data set is constructed based on the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value. When the reflection signal of the passive RFID tag has abnormal phase and normal phase, the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or, the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value. The second value represents the average value of the phases of the reflected signals of all passive RFID tags through which no moving body passes.

[0111] Here, a flow chart of constructing the sixth data set is shown as follows: Figure 3 As shown, the following steps are included:

[0112] Step 1: Determine the second data set and the third data set.

[0113] For example, one of the data in the second data set represents the difference between the RSSI value of the reflected signal of a passive RFID tag and the first value; the data in the second data set includes ΔRSSI i , the data in the third data set include Δphase i ; i represents the amount of data in the data set.

[0114] Step 2: Determine the fitting function f.

[0115] Specifically, the second data set and the third data set are fitted to obtain a fitting function. The method of fitting the second data set and the third data set is not limited, for example, linear function fitting, nonlinear function fitting, or fully connected network fitting. In practical applications, constructing a multi-layer fully connected network for fitting can ensure a better fitting effect.

[0116] Step 3: Based on the fitting function and the second data set, the fifth data set f(ΔRSSI i ).

[0117] Specifically, the data in the second data set are sequentially input into the fitting function, and the fifth data set is constructed based on the output data.

[0118] Step 4: Determine whether the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold ε. If the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, execute step 5; if the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold, determine that the phase of the reflected signal of the same passive RFID tag is abnormal, and execute step 6.

[0119] The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

[0120] Step 5: Determine that the phase of the reflected signal from the same passive RFID tag is normal.

[0121] Step 6: Construct the seventh and eighth data sets.

[0122] Specifically, the corrected phase of the reflection signal of the same passive RFID tag with phase abnormality is determined, and the seventh data set is constructed using the phase of the reflection signal of the same passive RFID tag with normal phase and the corrected phase of the reflection signal of the same passive RFID tag with phase abnormality.

[0123] The data in the seventh data set include phase i ', as follows:

[0124]

[0125] Among them, phase i -π is the corrected phase of the reflected signal of the same passive RFID tag with phase anomaly, phase i This is the phase of the reflected signal from the same passive RFID tag with normal phase.

[0126] The eighth data set is constructed using the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

[0127] The data in the eighth data set include Δphase i ', as follows:

[0128]

[0129] in, is the second value.

[0130] Step seven: Determine the seventh data set, the eighth data set and the data of phase anomaly.

[0131] In this embodiment, the accuracy of the sixth data set is relatively high, and the sixth data set includes the seventh data set and / or the eighth data set, further enriching the data included in the training samples in the training sample set, and helping to improve the accuracy of the first model.

[0132] After determining the trained first model, in order to further improve the model accuracy, in one embodiment, the model training method further includes:

[0133] The parameters of the set layer of the trained first model are adjusted based on the set sample to obtain the final first model; the set sample is determined from the training sample set based on the loss value corresponding to each training sample in the training sample set.

[0134] Here, the set samples are determined from the training sample set based on the loss value corresponding to each training sample in the training sample set. Specifically, the training samples in the training sample set whose loss values ​​corresponding to the training samples in the training sample set are within a set range can be used as the set samples; or the training samples in the training sample set can be sorted from large to small according to the corresponding loss values, and a set number of training samples arranged in front are selected as the set samples.

[0135] The parameters of the setting layer of the first model after training are adjusted based on the setting sample, and the setting layer can be set according to actual needs. For example, the setting layer is one or more FC layers in the M fully connected layers FC after the basic module, or the setting network layer in one or more basic modules close to the output in the Q basic modules and one or more FC layers in the M fully connected layers FC after the basic module. The setting network layer can be an LN layer, an FC layer, an activation function layer, and a fusion layer.

[0136] Based on the set samples, the parameters of the set layer of the trained first model are adjusted until the second set condition is met to obtain the final first model; the second set condition may be reaching a second set number of iterations, the loss function value may be less than or equal to a third set threshold, or the accuracy of the first model may be greater than or equal to a fourth set threshold.

[0137] In this embodiment, by adjusting the parameters of the setting layer of the trained first model through setting samples, the accuracy of the final first model can be improved.

[0138] In one embodiment, the model training method further includes:

[0139] Calculate the difference between the third value and the fourth value to obtain a fifth value; the third value represents the loss value corresponding to each training sample in the training sample set, and the fourth value is the average loss value corresponding to the training sample set;

[0140] When the fifth value corresponding to the training sample is smaller than the second threshold and larger than the third threshold, the training sample is determined as the set sample, and the second threshold is smaller than the third threshold.

[0141] The second threshold and the third threshold can be set according to actual needs.

[0142] Here, the training process diagram of the first model after training is shown as Figure 4 As shown, the following steps are included:

[0143] Step 1: Determine the third value and the fourth value.

[0144] For example, the third value is recorded as L j , the fourth value is recorded as L j Represents the loss value corresponding to the jth training sample in the training sample set.

[0145] Step 2: Determine the fifth value.

[0146] Specifically, the fifth value is

[0147] Step 3: Determine whether the fifth value corresponding to the training sample is greater than or equal to the second threshold β. If the fifth value corresponding to the training sample is greater than or equal to the second threshold, execute step 4; if the fifth value corresponding to the training sample is less than the second threshold, execute step 5;

[0148] Step 4: Determine that the training sample is an outlier sample. Forced fitting will reduce the generalization ability of the first model.

[0149] Step 5: Determine whether the fifth data corresponding to the training sample is less than or equal to the third threshold α. If the fifth data corresponding to the training sample is less than or equal to the third threshold, execute step 6; if the fifth data corresponding to the training sample is greater than the third threshold, execute step 7.

[0150] Step 6: Determine that the training sample is an easy-to-classify sample, and continuing to fit will waste computing resources.

[0151] Step 7: Determine the set samples, which are difficult-to-classify samples that do not meet the classification expectations of the model.

[0152] Specifically, when the fifth numerical value corresponding to the training sample is smaller than the second threshold value and larger than the third threshold value, the training sample is determined as the set sample.

[0153] Step 8: Use the set samples to adjust the parameters of the set layer of the trained first model.

[0154] Step 9: Determine whether the second setting condition is met. If the second setting condition is not met, return to step 8 to continue execution; if the second setting condition is met, execute step 10.

[0155] Step 10: Determine the final first model.

[0156] In this embodiment, the set samples are determined by the fifth value to ensure that the determined set samples have a high degree of importance, which can help improve the accuracy and generalization of the first model and avoid the catastrophic forgetting of the information of easy-to-classify samples by the first model.

[0157] Corresponding to the model training method on the first electronic device side described above, the embodiment of the present application further provides a positioning method, which is applied to a second electronic device, the second electronic device being a terminal with a display screen or display function; the second electronic device may be the same device as the first electronic device. Figure 5 As shown, the method includes:

[0158] Step 501: Process the signal quality parameters and phases of the reflected signals of multiple passive RFID tags received by the terminal to obtain first information.

[0159] Here, the signal quality parameter may be an RSSI value, or may be other parameters reflecting signal quality, which are not specifically limited. The first information includes data in the first data set and / or the second data, and the second data includes data in the second data set and / or data in the third data set and / or data in the fourth data set and / or data in the sixth data set. The specific processing process here can be found in the above embodiment, which will not be repeated here.

[0160] It is worth noting that the passive RFID tags here are the same as those used during model training.

[0161] Taking the positioning accuracy of k meters and the antenna coverage of K meters as an example, the layout diagram of passive RFID tags is as follows: Figure 6 W passive RFID tags need to be arranged between every two antennas, namely, passive RFID tag 1, passive RFID tag 2, passive RFID tag 3, passive RFID tag 4, ..., passive RFID tag W, where W = K / k. Figure 6 The layout of the passive RFID tags between antenna 1 and antenna 2 is shown in FIG. With each passive RFID tag as the center and k / 2 meters on the left and right as the boundaries, the class area corresponding to each passive RFID tag is determined. The area between antenna 1 and antenna 2 includes class0, class1, class2, class3, ..., classW-1.

[0162] Step 502: Input the first information into a second model to obtain second information.

[0163] The second information represents the area where the terminal is located, and the second model is trained based on any one of the methods on the first electronic device side.

[0164] The second model may be the trained first model or the final first model. The specific model training process here can be found in the above embodiment, which will not be described in detail here.

[0165] by Figure 6 For example, the second information is class3.

[0166] Step 503: Output or display the second information.

[0167] The positioning method provided in this embodiment uses passive RFID tags to achieve real-time and accurate positioning of mobile objects without the need to introduce auxiliary equipment such as cameras and GPS. Passive RFID tags can effectively avoid the problems of high equipment cost and high power consumption. Constructing a second model that can be deployed in new computing architectures such as storage and computing integration can weaken the problem of high energy consumption. That is, while ensuring good positioning accuracy, it can meet the requirements of low cost, low energy consumption and easy terminal deployment, which is conducive to the practical application of the positioning method.

[0168] The following is a further detailed description of this application in conjunction with application examples. Figure 7 As shown, the following steps are included:

[0169] Step 701: Convert the complex positioning problem into a simple classification problem.

[0170] Specifically, it is determined to deploy passive RFID tags on the wall, it is determined that a training sample corresponds to an area where a passive RFID tag with a moving body passes, and it is determined to convert the moving body positioning problem into an area classification problem corresponding to each passive RFID tag.

[0171] Step 702: Obtain a first data set.

[0172] Specifically, RSSI values ​​and phases of reflected signals of multiple passive RFID tags are obtained when there is a moving body passing through the passive RFID tag and when there is no moving body passing through the passive RFID tag. It should be noted that the RSSI values ​​and phases of reflected signals of multiple passive RFID tags without a moving body passing through the passive RFID tag are obtained in order to eliminate external interference factors and are used in step 703.

[0173] Step 703: Process the data in the first data set.

[0174] Specifically, since the feature dimension of the data in the first data set is low, the data in the first data set is processed to obtain the second data set, the third data set, the fourth data set, the seventh data set and the eighth data set.

[0175] Step 704: Construct a training sample set.

[0176] Specifically, in order to retain all valuable information and avoid information loss, each training sample in the training sample set includes the first data corresponding to multiple passive RFID tags at the same time. The first data includes feature_1, feature_2, feature_3, feature_4, feature_5, feature_6, feature_7 and feature_8. ​​Among them, feature_1 represents the RSSI value of the reflected signal of multiple passive RFID tags when there is a moving body passing through the passive RFID tag and when there is no moving body passing through the passive RFID tag; feature_2 represents the phase of the reflected signal of multiple passive RFID tags when there is a moving body passing through the passive RFID tag and when there is no moving body passing through the passive RFID tag; feature_3 represents the data in the second data set; feature_4 represents the data in the third data set; feature_5 represents the data in the seventh data set; feature_6 represents the data in the eighth data set; feature_7 represents the phase cosine value of the reflected signal of a passive RFID tag in the fourth data set; feature_8 represents the phase sine value of the reflected signal of a passive RFID tag in the fourth data set.

[0177] Step 705: Train the first model.

[0178] Specifically, the first model is trained based on the training sample set to obtain a trained first model.

[0179] Step 706: Adjust the trained first model.

[0180] Specifically, when the fifth value corresponding to the training sample is less than the second threshold value and greater than the third threshold value, the training sample is determined as the set sample; the parameters of the set layer of the trained first model are adjusted based on the set sample to obtain the final first model. Here, the reason for selecting the parameters of the set layer of the trained first model for adjustment is that, first, the first few layers of the model learn common features, and continuing training is not meaningful and will increase the amount of calculation; second, the model adjustment uses set samples, that is, difficult-to-classify samples. Retaining the parameters of the first few layers of the model can avoid catastrophic forgetting of information on easy-to-classify samples, so that the entire model can learn the features of difficult-to-classify samples and easy-to-classify samples at the same time.

[0181] Step 707: Use the final first model for positioning.

[0182] Specifically, the signal quality parameters and phases of the received reflection signals of the plurality of passive RFID tags are processed to obtain the first information; the first information is input into the final first model to obtain the second information; and the second information is output or displayed.

[0183] The model training method and positioning method provided in the present application, by deploying multiple passive RFID tags at fixed positions, each passive RFID tag is divided into an area corresponding to a category, thereby converting a complex positioning problem into a simple classification problem; processing the data in the first data set, enriching the training sample features, and each training sample in the training sample set is an image-like sample, constructing a first model, that is, a fully connected neural network model for classification, and training and adjusting the first model, can not only achieve real-time and high-precision positioning, but also meet the requirements of low cost, low energy consumption and easy terminal deployment; compared with the existing technology of placing RFID tags on moving objects such as robots and using changes in electromagnetic wave signals to achieve the positioning of moving objects, The present application deploys passive RFID tags on walls or other places where it is easy to fix tags, monitors the data fluctuations generated by the relative position of the moving object and the passive RFID tags to locate the moving object, which can avoid the interference of the structure of the moving object on the reflection, refraction and scattering of the reflected signal during the propagation process, which is beneficial to the accuracy and stability of the data. Compared with the existing technology of placing reference tags in the target area and using the electromagnetic wave signals of the reference tags to locate the tags to be tested, the present application does not set reference tags and tags to be tested, but uses the reflected signals of fixed passive RFID tags and the relative position of the moving object and the fixed passive RFID tags to convert the complex positioning problem into a simple classification problem, which is more suitable for the scenario of mobile object positioning.

[0184] In order to implement the model training method on the first electronic device side of the embodiment of the present application, the embodiment of the present application also provides a model training device, which is arranged on the first electronic device, such as Figure 8 As shown, the device comprises:

[0185] The sample construction module 801 is used to construct a training sample set based on a first data set; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located;

[0186] The training module 802 is used to train a first model based on the training sample set, where the first model is used to predict the area where the moving object is located based on the input training samples.

[0187] In one embodiment, the sample construction module 801 is specifically used to determine a second data set and / or a third data set and / or a fourth data set based on the first data set, wherein one data in the second data set represents a difference between a signal quality parameter of a reflection signal of a passive RFID tag and a first value, the first value represents an average value of the signal quality parameter of the reflection signals of all passive RFID tags through which no moving body passes, one data in the third data set represents a difference between a phase of a reflection signal of a passive RFID tag and a second value, the second value represents an average value of the phase of the reflection signals of all passive RFID tags through which no moving body passes, and one data in the fourth data set represents a trigonometric function value of the phase of the reflection signal of a passive RFID tag;

[0188] The training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set.

[0189] In one embodiment, the sample-based construction module 801 is specifically used to obtain a fifth data set based on a fitting function and the second data set; the fitting function is obtained by fitting the second data set and the third data set;

[0190] Based on the data of the same passive RFID tag at the same time corresponding to the fifth data set and the third data set, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal;

[0191] constructing a sixth data set based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase;

[0192] The training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set.

[0193] In one embodiment, the sixth data set includes the seventh data set and / or the eighth data set; the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

[0194] In one embodiment, the sample construction module 801 is specifically used to determine the phase anomaly of the reflection signal of the same passive RFID tag when the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold;

[0195] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal; wherein,

[0196] The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

[0197] In one embodiment, the corrected phase of the reflection signal of the same passive RFID tag with phase anomaly is obtained by subtracting 180 degrees from the phase of the reflection signal of the same passive RFID tag.

[0198] In one embodiment, the device further comprises:

[0199] An adjustment module is used to adjust the parameters of the set layer of the trained first model based on the set samples to obtain the final first model; the set samples are determined from the training sample set based on the loss value corresponding to each training sample in the training sample set.

[0200] In one embodiment, the device further comprises:

[0201] A calculation module, used to calculate the difference between the third value and the fourth value to obtain a fifth value; the third value represents the loss value corresponding to each training sample in the training sample set, and the fourth value is the average loss value corresponding to the training sample set;

[0202] The sample determination module is used to determine the training sample as a set sample when the fifth value corresponding to the training sample is less than the second threshold and greater than the third threshold, and the second threshold is less than the third threshold.

[0203] In actual application, the sample construction module 801, training module 802, adjustment module, calculation module and sample determination module can be implemented by a processor of the model training device.

[0204] It should be noted that the model training device provided in the above embodiment only uses the division of the above program modules as an example when performing model training. In actual applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device is divided into different program modules to complete all or part of the processing described above. In addition, the model training device provided in the above embodiment and the model training method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0205] In order to implement the positioning method on the second electronic device side of the embodiment of the present application, the embodiment of the present application also provides a positioning device, which is arranged on the second electronic device, such as Fig. 9 As shown, the device comprises:

[0206] The processing module 901 is used to process the signal quality parameters and phases of the reflected signals of the multiple passive RFID tags received by the terminal to obtain the first information;

[0207] An input module 902 is used to input the first information into a second model to obtain second information, where the second information represents the area where the terminal is located, and the second model is trained based on any method on the first electronic device side;

[0208] The output module 903 is used to output or display the second information.

[0209] In actual application, the processing module 901, input module 902 and output module 903 can be implemented by a processor of the positioning device.

[0210] It should be noted that the positioning device provided in the above embodiment is only illustrated by the division of the above program modules when performing positioning. In actual applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device is divided into different program modules to complete all or part of the processing described above. In addition, the positioning device provided in the above embodiment and the positioning method embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0211] Based on the hardware implementation of the above program modules, and in order to implement the method of the first electronic device side of the embodiment of the present application, the embodiment of the present application also provides a first electronic device, such as Fig.10 As shown, the first electronic device 1000 includes:

[0212] The first communication interface 1001 is capable of exchanging information with other network nodes;

[0213] The first processor 1002 is connected to the first communication interface 1001 to implement information exchange with other network nodes, and is used to execute the method provided by one or more technical solutions of the first electronic device side when running a computer program. The computer program is stored in the first memory 1003.

[0214] Specifically, the first processor 1002 is used to construct a training sample set based on a first data set; train a first model based on the training sample set, and the first model is used to predict the area where the moving body is located based on the input training samples; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag through which a moving body passes is located.

[0215] In one embodiment, the first processor 1002 is specifically configured to determine a second data set and / or a third data set and / or a fourth data set based on the first data set, wherein one data in the second data set represents a difference between a signal quality parameter of a reflection signal of a passive RFID tag and a first value, the first value represents an average value of the signal quality parameter of the reflection signals of all passive RFID tags through which no moving body passes, one data in the third data set represents a difference between a phase of a reflection signal of a passive RFID tag and a second value, the second value represents an average value of the phase of the reflection signals of all passive RFID tags through which no moving body passes, and one data in the fourth data set represents a trigonometric function value of the phase of the reflection signal of a passive RFID tag;

[0216] The training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set.

[0217] In one embodiment, the first processor 1002 is specifically configured to obtain a fifth data set based on a fitting function and the second data set; the fitting function is obtained by fitting the second data set and the third data set;

[0218] Based on the data of the same passive RFID tag at the same time corresponding to the fifth data set and the third data set, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal;

[0219] constructing a sixth data set based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase;

[0220] The training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set.

[0221] In one embodiment, the sixth data set includes the seventh data set and / or the eighth data set; the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

[0222] In one embodiment, the first processor 1002 is specifically configured to determine that the phase of the reflection signal of the same passive RFID tag is abnormal when the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold;

[0223] When the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal; wherein,

[0224] The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

[0225] In one embodiment, the corrected phase of the reflection signal of the same passive RFID tag with phase anomaly is obtained by subtracting 180 degrees from the phase of the reflection signal of the same passive RFID tag.

[0226] In one embodiment, the first processor 1002 is further used to adjust the parameters of the set layer of the trained first model based on the set sample to obtain the final first model; the set sample is determined from the training sample set based on the loss value corresponding to each training sample in the training sample set.

[0227] In one embodiment, the first processor 1002 is further configured to calculate a difference between a third value and a fourth value to obtain a fifth value; the third value represents a loss value corresponding to each training sample in the training sample set, and the fourth value is an average loss value corresponding to the training sample set;

[0228] When the fifth value corresponding to the training sample is smaller than the second threshold and larger than the third threshold, the training sample is determined as the set sample, and the second threshold is smaller than the third threshold.

[0229] It should be noted that the specific processing process of the first processor 1002 can be understood by referring to the above method.

[0230] Of course, in actual application, the various components in the first electronic device 1000 are coupled together through the bus system 1004. It can be understood that the bus system 1004 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1004 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, Fig.10 Various buses are labeled as bus system 1004 .

[0231] The first memory 1003 in the embodiment of the present application is used to store various types of data to support the operation of the first electronic device 1000. Examples of such data include: any computer program used to operate on the first electronic device 1000.

[0232] The method disclosed in the above embodiment of the present application can be applied to the first processor 1002, or implemented by the first processor 1002. The first processor 1002 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the first processor 1002. The above-mentioned first processor 1002 may be a general-purpose processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The first processor 1002 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor may be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiment of the present application, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in the first memory 1003, and the first processor 1002 reads the information in the first memory 1003 and completes the steps of the above method in combination with its hardware.

[0233] In an exemplary embodiment, the first electronic device 1000 can be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), field programmable gate array (FPGA), general processor, controller, microcontroller (MCU), microprocessor, or other electronic components to execute the aforementioned method.

[0234] Based on the hardware implementation of the above program modules, and in order to implement the method of the second electronic device side of the embodiment of the present application, the embodiment of the present application also provides a second electronic device, such as Fig.11 As shown, the second electronic device 1100 includes:

[0235] The second communication interface 1101 is capable of exchanging information with other network nodes;

[0236] The second processor 1102 is connected to the second communication interface 1101 to implement information exchange with other network nodes, and is used to execute the method provided by one or more technical solutions of the second electronic device side when running a computer program. The computer program is stored in the second memory 1103.

[0237] Specifically, the second processor 1102 is used to process the signal quality parameters and phases of the reflected signals of multiple passive RFID tags received by the terminal to obtain first information; input the first information into a second model to obtain second information; output or display the second information; the second information represents the area where the terminal is located, and the second model is trained based on any method on the first electronic device side mentioned above.

[0238] It should be noted that the specific processing process of the second processor 1102 can be understood by referring to the above method.

[0239] Of course, in actual application, the various components in the second electronic device 1100 are coupled together through the bus system 1104. It can be understood that the bus system 1104 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1104 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, Fig.11 Various buses are labeled as bus system 1104 .

[0240] The second memory 1103 in the embodiment of the present application is used to store various types of data to support the operation of the second electronic device 1100. Examples of such data include: any computer program used to operate on the second electronic device 1100.

[0241] The method disclosed in the above embodiment of the present application can be applied to the second processor 1102, or implemented by the second processor 1102. The second processor 1102 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of the hardware in the second processor 1102 or an instruction in the form of software. The above-mentioned second processor 1102 may be a general-purpose processor, a DSP, or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The second processor 1102 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present application. A general-purpose processor may be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiment of the present application, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in the second memory 1103, and the second processor 1102 reads the information in the second memory 1103 and completes the steps of the above method in combination with its hardware.

[0242] In an exemplary embodiment, the second electronic device 1100 may be implemented by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general purpose processors, controllers, MCUs, Microprocessors, or other electronic components to perform the aforementioned methods.

[0243] It can be understood that the memory (first memory 1003, second memory 1103) of the embodiment of the present application can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a ferromagnetic random access memory, a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM, SyncLink Dynamic Random Access Memory), and direct RAM bus random access memory (DRRAM, Direct Rambus Random Access Memory).The memories described in the embodiments of the present application are intended to include, but are not limited to, these and any other suitable types of memories.

[0244] An embodiment of the present application also provides an electronic device, including a processor and a memory for storing a computer program that can be run on the processor, wherein the processor, when used to run the computer program, executes the steps of any method on the first electronic device side, or executes the steps of any method on the second electronic device side.

[0245] An embodiment of the present application further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program implements the steps of any method on the first electronic device side or implements the steps of any method on the second electronic device side.

[0246] In an exemplary embodiment, the embodiment of the present application further provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, for example, including a first memory 1003 storing a computer program, and the computer program can be executed by the first processor 1002 of the first electronic device 1000 to complete the steps of the method on the first electronic device side. For another example, a second memory 1103 storing a computer program can be executed by the second processor 1102 of the second electronic device 1100 to complete the steps of the method on the second electronic device side. The computer-readable storage medium can be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface storage, optical disk, or CD-ROM.

[0247] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any method on the first electronic device side, or implements the steps of any method on the second electronic device side.

[0248] It should be noted that "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. The term "and / or" in this article is only a description of the association relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0249] In addition, the technical solutions described in the embodiments of the present application can be combined arbitrarily without conflict.

[0250] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application.

Claims

1. A model training method, characterized in that: The method comprises: A training sample set is constructed based on a first data set; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located; A first model is trained based on the training sample set, where the first model is used to predict the area where the moving object is located based on the input training samples.

2. The method according to claim 1, characterized in that: The step of constructing a training sample set based on the first data set includes: Determining a second data set and / or a third data set and / or a fourth data set based on the first data set, wherein one data in the second data set represents a difference between a signal quality parameter of a reflection signal of a passive RFID tag and a first value, the first value represents an average value of the signal quality parameter of the reflection signals of all passive RFID tags without a moving body passing through, one data in the third data set represents a difference between a phase of a reflection signal of a passive RFID tag and a second value, the second value represents an average value of the phase of the reflection signals of all passive RFID tags without a moving body passing through, and one data in the fourth data set represents a trigonometric function value of the phase of the reflection signal of a passive RFID tag; The training sample set is constructed based on the first data set, and based on the second data set and / or the third data set and / or the fourth data set.

3. The method according to claim 2, characterized in that Constructing the training sample set based on the first data set, and based on the second data set and the third data set includes: Based on the fitting function and the second data set, a fifth data set is obtained; the fitting function is obtained by fitting the second data set and the third data set; Based on the data of the same passive RFID tag at the same time corresponding to the fifth data set and the third data set, determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal; constructing a sixth data set based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; The training sample set is constructed based on the first data set, the second data set, the third data set and the sixth data set.

4. The method according to claim 3, characterized in that The sixth data set includes the seventh data set and / or the eighth data set; the seventh data set is constructed based on the phase of the reflection signal of the same passive RFID tag with normal phase and / or the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase; the eighth data set is constructed based on the difference between the phase of the reflection signal of the same passive RFID tag with normal phase and the second value, and / or the difference between the corrected phase of the reflection signal of the same passive RFID tag with abnormal phase and the second value.

5. The method according to claim 3 or 4, characterized in that: The determining whether the phase of the reflection signal of the same passive RFID tag is normal or abnormal based on the data of the same passive RFID tag at the same time in the fifth data set and the third data set includes: When the difference between the third data and the fourth data corresponding to the same passive RFID tag is greater than the first threshold, determining that the phase of the reflection signal of the same passive RFID tag is abnormal; When the difference between the third data and the fourth data corresponding to the same passive RFID tag is less than or equal to the first threshold, it is determined that the phase of the reflected signal of the same passive RFID tag is normal; wherein, The third data represents the data in the fifth data set, the fourth data represents the data in the third data set, and the third data and the fourth data correspond to the same time.

6. The method according to claim 3 or 4, characterized in that: The corrected phase of the reflection signal of the same passive RFID tag with phase anomaly is obtained by subtracting 180 degrees from the phase of the reflection signal of the same passive RFID tag.

7. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: The parameters of the set layer of the trained first model are adjusted based on the set sample to obtain the final first model; the set sample is determined from the training sample set based on the loss value corresponding to each training sample in the training sample set.

8. The method according to claim 7, characterized in that The method further comprises: Calculate the difference between the third value and the fourth value to obtain a fifth value; the third value represents the loss value corresponding to each training sample in the training sample set, and the fourth value is the average loss value corresponding to the training sample set; When the fifth value corresponding to the training sample is smaller than the second threshold and larger than the third threshold, the training sample is determined as the set sample, and the second threshold is smaller than the third threshold.

9. A positioning method, characterized in that: The method comprises: Processing signal quality parameters and phases of reflected signals of multiple passive RFID tags received by the terminal to obtain first information; Inputting the first information into a second model to obtain second information, where the second information represents the area where the terminal is located, and the second model is trained based on the method according to any one of claims 1 to 8; The second information is output or displayed.

10. A model training device, characterized in that: The device comprises: A sample construction module is used to construct a training sample set based on a first data set; wherein the first data set includes signal quality parameters and phases of reflection signals of multiple passive RFID tags when a moving body passes through the passive RFID tag and when no moving body passes through the passive RFID tag; each training sample in the training sample set includes first data corresponding to multiple passive RFID tags at the same time; the first data includes second data and / or data in the first data set, and the second data is determined based on the data in the first data set; one training sample corresponds to an area where a passive RFID tag with a moving body passing through is located; The training module is used to train a first model based on the training sample set, wherein the first model is used to predict the area where the moving object is located based on the input training samples.

11. A positioning device, characterized in that: The device comprises: A processing module, configured to process the signal quality parameters and phases of the reflected signals of the plurality of passive RFID tags received by the terminal to obtain first information; An input module, used for inputting the first information into a second model to obtain second information, wherein the second information represents the area where the terminal is located, and the second model is trained based on the method according to any one of claims 1 to 8; An output module is used to output or display the second information.

12. An electronic device, characterized in that: comprising a processor and a memory for storing a computer program capable of being executed on the processor, Wherein, when the processor is used to run the computer program, it executes the steps of the method described in any one of claims 1 to 8, or executes the steps of the method described in claim 9.

13. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented, or the steps of the method according to claim 9 are implemented.

14. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented, or the steps of the method according to claim 9 are implemented.

Images