RFID ranging model construction method and system based on one-dimensional convolutional neural network

Abstract

The invention provides an RFID distance measurement model construction method and system based on a one-dimensional convolutional neural network, and the method comprises the steps: obtaining signal intensity data, carrying out the preprocessing, and enabling an obtained effective RSSI value to serve as model input data; inputting the model input data into a one-dimensional convolutional neural network (CNN) model for training, and optimizing the CNN model by adopting a third-generation non-dominated sorting genetic algorithm to obtain an NSGAIII-1D-CNN model; and dividing the effective RSSI values and the corresponding actual distance values into a training set and a test set, training the NSGAIII-1D-CNN model by using the training set to obtain a trained RFID distance measurement model, and verifying the RFID distance measurement model through the test set to obtain a final RFID distance measurement model. The method further includes performing location exploration through the final RFID ranging model in an applicable environment. According to the invention, a new RFID ranging model of a multi-objective optimization one-dimensional convolutional neural network based on a non-dominated sorting genetic algorithm is designed, so that the precision and robustness of RFID ranging are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of geographic exploration technology, and in particular to a method and system for constructing an RFID ranging model based on a one-dimensional convolutional neural network. Background Art

[0002] As a non-contact short-distance data exchange technology, ultra-high frequency passive RFID uses low-power electronic tag chips, UHF antennas, anti-collision algorithms and test and certification systems. It is widely used in the earthquake prevention and disaster reduction industry, specifically in the fields of earthquake equipment management, earthquake emergency material management, safety monitoring of earthquake-hazardous areas, earthquake prediction systems, etc., especially in geophysical exploration, where it plays an important role in node instrument deployment and short-distance data interaction. In some special earthquake monitoring site conditions, such as underground pits, dense vegetation, and nighttime mountain roads, node-type seismic instruments have the disadvantages of fuzzy deployment positions and cumbersome manual exploration. Using RFID for ranging or positioning is a possible solution. In view of the problem that wireless node seismometers cannot accurately locate in the network in complex field exploration environments, and their deployment is hidden and easy to lose, it is of great significance and economic value to study ultra-high frequency RFID high-precision ranging and positioning. However, the existing RFID ranging or positioning technology mostly relies on empirical approximate formulas. Before calculation, it is necessary to measure two environment-related constants, such as site propagation factor and standard receiver power, according to each different monitoring environment. This is cumbersome and has large errors, which is not conducive to the automation and promotion of ranging model algorithms. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for constructing an RFID ranging model based on a one-dimensional convolutional neural network, aiming to solve the above-mentioned problems in the prior art.

[0004] The embodiment of the present invention provides a method for constructing an RFID ranging model based on a one-dimensional convolutional neural network, comprising:

[0005] Acquire signal strength data and preprocess it, and use the obtained effective RSSI value as model input data;

[0006] Inputting the model input data into a one-dimensional convolutional neural network (CNN) model for training, and optimizing the CNN model in training using a third-generation non-dominated sorting genetic algorithm to obtain a third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model; and

[0007] The effective RSSI values ​​and their corresponding actual distance values ​​are divided into a training set and a test set. The NSGAIII-1D-CNN model is trained using the training set to obtain a trained RFID ranging model. The RFID ranging model is verified using the test set to obtain a final RFID ranging model.

[0008] The embodiment of the present invention provides an RFID ranging model construction system based on a one-dimensional convolutional neural network, comprising:

[0009] The data module is used to obtain the signal strength data and perform preprocessing, and use the obtained effective RSSI value as the model input data;

[0010] An optimization module, used for inputting the model input data into a one-dimensional convolutional neural network (CNN) model for training, and optimizing the CNN model in training using a third-generation non-dominated sorting genetic algorithm to obtain a third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model; and

[0011] A construction module is used to divide the effective RSSI value and its corresponding actual distance value into a training set and a test set, use the training set to train the NSGAIII-1D-CNN model to obtain a trained RFID ranging model, and verify the RFID ranging model through the test set to obtain a final RFID ranging model.

[0012] An embodiment of the present invention also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the above-mentioned method for constructing an RFID ranging model based on a one-dimensional convolutional neural network are implemented.

[0013] An embodiment of the present invention also provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps of the above-mentioned method for constructing an RFID ranging model based on a one-dimensional convolutional neural network are implemented.

[0014] The use of the embodiments of the present invention can include the following beneficial effects: the NSGAIII-1D-CNN model proposed in the embodiments of the present invention is to improve the problem that the accurate position of wireless node seismometers is difficult to obtain and the instrument may be lost in special environments such as geophysical exploration and field earthquake monitoring. The embodiments of the present invention design a new RFID ranging model based on a non-dominated sorting genetic algorithm multi-objective optimization of a one-dimensional convolutional neural network, and use the determination coefficient and the average deviation error as the dependent variables of the multi-objective optimization function to obtain the desired network hyperparameter values. The new model is used to achieve the purpose of high accuracy and certain robustness of RFID ranging prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] Figure 1 It is a flow chart of a method for constructing an RFID ranging model based on a one-dimensional convolutional neural network according to an embodiment of the present invention;

[0017] Figure 2 It is a flow chart of ranging based on NSGAIII-1D-CNN model in an embodiment of the present invention;

[0018] Figure 3 Schematic diagram of a node seismograph RFID ranging system according to an embodiment of the present invention;

[0019] Figure 4 1 is a schematic diagram of a one-dimensional convolutional neural network structure according to an embodiment of the present invention;

[0020] Figure 5 Schematic diagram of RSSI values ​​at 800 different RFID scanning distances according to an embodiment of the present invention;

[0021] Figure 6 This is a comparison chart of RMSE results of 100 rounds of NSGAIII-1D-CNN ranging model predictions in an embodiment of the present invention;

[0022] Figure 7 It is a schematic diagram of a RFID ranging model construction system based on a one-dimensional convolutional neural network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0024] Method Embodiment

[0025] According to an embodiment of the present invention, a method for constructing an RFID ranging model based on a one-dimensional convolutional neural network is provided. Figure 1 is a flow chart of a method for constructing an RFID ranging model based on a one-dimensional convolutional neural network according to an embodiment of the present invention. Figure 1 As shown, the RFID ranging model construction method based on a one-dimensional convolutional neural network according to an embodiment of the present invention specifically includes:

[0026] Step S101, obtaining signal strength data and preprocessing it, and using the obtained effective RSSI value as model input data, specifically includes:

[0027] Collect signal strength data through the received signal strength indicator RSSI, obtain an approximate propagation factor according to the signal strength data using an approximate calculation formula as shown in Formula 1, and determine whether the approximate propagation factor is within a threshold range. If so, use the corresponding signal strength data as a valid RSSI value; if not, remove the abnormal value and re-collect the signal strength data until a valid RSSI value is obtained;

[0028] PR (dBm) ≈ rssi1-10·prop·lgdist Formula 1;

[0029] Wherein, PR represents the receiving frequency of the wireless signal, rssi1 represents the receiving signal frequency when the wireless signal transmission distance is 1 meter, prop represents the propagation factor related to the wireless signal propagation environment, and dist represents the absolute distance between the transceiver devices;

[0030] Step S102, inputting the model input data into a one-dimensional convolutional neural network CNN model for training, and optimizing the CNN model in training using a third-generation non-dominated sorting genetic algorithm to obtain a third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model, specifically comprising:

[0031] A third-generation non-dominated sorting genetic algorithm is used to establish a multi-objective optimization function with the initial learning rate, batch size and learning rate reduction factor of the CNN model as independent variables, and the dependent variables are the average deviation error and the determination coefficient, respectively. The expressions are shown in Formula 2, Formula 3 and Formula 4, respectively. The optimal network hyperparameters of the CNN model are obtained through the multi-objective optimization function;

[0032] minF={f1,f2} Formula 2;

[0033]

[0034] Among them, minF represents the solution model of NSGA-III multi-objective optimization problem, f1 is the calculation formula of average deviation error, f2 is the calculation formula of determination coefficient, yi is the predicted value of RFID scanning distance, is the actual value of the distance measurement, is the average value of the ranging prediction value set, and n is the total number of samples involved in the regression evaluation calculation;

[0035] Step S103, dividing the effective RSSI value and its corresponding actual distance value into a training set and a test set, using the training set to train the NSGAIII-1D-CNN model to obtain a trained RFID ranging model, and verifying the RFID ranging model through the test set to obtain a final RFID ranging model;

[0036] The method further comprises:

[0037] The final RFID ranging model is used to perform location exploration in an applicable environment.

[0038] The following is a specific example of the method for constructing an RFID ranging model based on a one-dimensional convolutional neural network according to an embodiment of the present invention. Figure 2 As shown, the above technical solution of the embodiment of the present invention is described in detail.

[0039] The embodiment of the present invention designs a one-dimensional convolutional neural network RFID ranging method, constructs network structures such as an input layer, a convolution kernel, a batch normalization layer, a Relu activation layer, a fully connected layer, a regression layer, and a Dropout layer, and uses the third-generation non-dominated sorting genetic algorithm (NSGA-III) to generate a learning rate reduction factor, an initial learning rate, and a batch size as function independent variables, and constructs two different optimization functions with a determination coefficient and an average deviation error as targets, respectively. The optimal hyperparameters of the adaptive prediction environment are obtained by combining a partial training set and a test set, and a NSGAIII-1D-CNN model is formed; at the same time, a sample outlier removal model is constructed using the RSSI classical ranging principle to ensure the reliability of the network input data matrix.

[0040] Among them, convolutional neural network (CNN) is a feedforward deep learning model that is relatively popular in many industries. It can use translation invariance and convolution calculation to extract local features and overall features. It has been successfully applied to earthquake warning magnitude estimation, earthquake data denoising, and seismic wave and phase identification. Therefore, the embodiment of the present invention takes into account the advanced recognition and complex feature processing capabilities of the CNN model, and applies neural networks to RFID ranging as a new attempt, with predictable experimental results.

[0041] The process of the node seismograph RFID ranging method of optimizing a one-dimensional convolutional neural network using a non-dominated sorting genetic algorithm proposed in an embodiment of the present invention is as follows:

[0042] Step 1: Signal acquisition and preprocessing

[0043] Use the received signal strength indicator RSSI to receive the signal strength (such as Figure 3 As shown in the figure, the approximate propagation factor prop of the collected original data is judged by the approximate calculation formula (Formula (1)) to see whether it meets the threshold range. If an abnormal value appears, it means that the sample error is large and is greatly disturbed by the test environment. It needs to be re-collected to ensure that the effective RSSI value (Received Signal Strength Indication) of each type of tag value is 50.

[0044] PR(dBm)≈rssi1-10·prop·lgdist(1);

[0045] In the formula, PR represents the receiving frequency of the wireless signal / dBm, rssi1 represents the receiving signal frequency when the wireless signal transmission distance is 1 meter, prop represents the propagation factor related to the wireless signal propagation environment, which is generally between 2 and 4, and dist represents the absolute distance between the transceiver and the transmitter, in meters.

[0046] However, the propagation process of radio waves is easily affected by various environmental factors, such as metal object obstruction, multipath effect, long-distance signal attenuation and electromagnetic interference. Therefore, the traditional RFID ranging formula cannot be used directly, and complex automated algorithms and exception handling methods are needed to improve the ranging accuracy.

[0047] Step 2: Optimize the one-dimensional convolutional neural network using the third-generation non-dominated sorting genetic algorithm

[0048] The third generation non-dominated sorting genetic algorithm is used to construct an optimization objective function with two independent variables (learning rate reduction factor, initial learning rate, batch size, etc.) and the same dependent variables, namely, the discrimination coefficient and the mean deviation error MBE. The main body of the function is a one-dimensional convolutional neural network (such as Figure 4 shown).

[0049] The solution model of NSGA-III multi-objective optimization problem is generally:

[0050] minF = {f1, f2} (2);

[0051]

[0052] Where f1 is the mean bias error (MBE), which can determine whether the prediction model has positive or negative bias; f2 is the calculation formula of the coefficient of determination (R2 score), which is used to evaluate the degree of explanation of the independent variable on the dependent variable in the regression analysis; y i is the predicted value of RFID scanning distance; is the actual value of the distance measurement; is the average value of the ranging prediction value set; n is the total number of samples involved in the regression evaluation calculation.

[0053] Step 3: Use the new model to estimate RFID distance prediction

[0054] The CNN optimal learning rate reduction factor, optimal initial learning rate, optimal batch size and other hyperparameter values ​​obtained in the previous step are used to construct a new prediction model NSGAIII-1D-CNN adapted to the sample data of the embodiment of the present invention. The ratio of the training set to the test set is 4:1, the training algorithm adopts the gradient descent method, the maximum number of training times is 100, and the network structure is designed as 1 input layer, 2 convolution layers, 2 batch normalization layers, 2 Relu activation layers, 1 Dropout layer, 1 fully connected layer and 1 regression layer. The RSSI test value and the actual distance are used as network input, and the predicted distance vector is used as output.

[0055] The model is evaluated based on the determination coefficient, mean absolute error (MAE), and MBE between the ranging prediction value vector and the actual value vector.

[0056] In summary, the purpose of the embodiment of the present invention is to provide a node seismograph RFID ranging method based on a non-dominated sorting genetic algorithm to optimize a one-dimensional convolutional neural network, which changes the previous node seismograph RFID ranging or positioning method that mainly relies on empirical formulas or manual auxiliary measurement. The implementation process is exemplified as follows:

[0057] Firstly, the third generation non-dominated sorting genetic algorithm (NSGA-III) was introduced to establish two multi-objective optimization functions with initial learning rate, batch size and learning rate reduction factor as independent variables and mean bias error (MBE) and determination coefficient (R2 score) as dependent variables, so as to optimize the one-dimensional convolutional neural network to obtain the optimal network hyperparameters and build the NSGAIII-1D-CNN ranging model; then, 800 valid RSSI values ​​(such as Figure 5 There are 16 types of measured distances, which are distributed between 0.1-1.6m in sequence, with a distance step of 0.1m. 50 RSSI values ​​are collected for each distance measurement. The results are shown in Table 1.

[0058] Table 1 Prediction results of NSGAIII-1D-CNN ranging model under different ranging conditions in a single round

[0059]

[0060] Finally, NSGAIII-1D-CNN has a significant improvement in prediction performance compared to the unoptimized original 1D-CNN model, with better regression goodness of fit and model robustness (e.g. Figure 6 As shown in the figure, it performs well in multiple indicators such as determination coefficient (R2), RMSE, MAE, MBE and their mean, standard deviation, etc., and their means are 0.9779, 0.0586, 0.0472, and -0.0013, respectively, which proves that the model has certain value and novelty in the field of RFID ranging.

[0061] System Example

[0062] According to an embodiment of the present invention, a RFID ranging model construction system based on a one-dimensional convolutional neural network is provided. Figure 7 is a schematic diagram of a RFID ranging model construction system based on a one-dimensional convolutional neural network according to an embodiment of the present invention, such as Figure 7 As shown, the RFID ranging model construction system based on a one-dimensional convolutional neural network according to an embodiment of the present invention specifically includes:

[0063] The data module 70 is used to obtain signal strength data and perform preprocessing, and use the obtained effective RSSI value as model input data, specifically for:

[0064] Collect signal strength data through the received signal strength indicator RSSI, obtain an approximate propagation factor according to the signal strength data using an approximate calculation formula as shown in Formula 1, and determine whether the approximate propagation factor is within a threshold range. If so, use the corresponding signal strength data as a valid RSSI value; if not, remove the abnormal value and re-collect the signal strength data until a valid RSSI value is obtained;

[0065] PR (dBm) ≈ rssi1-10·prop·lgdist Formula 1;

[0066] Wherein, PR represents the receiving frequency of the wireless signal, rssi1 represents the receiving signal frequency when the wireless signal transmission distance is 1 meter, prop represents the propagation factor related to the wireless signal propagation environment, and dist represents the absolute distance between the transceiver devices;

[0067] The optimization module 72 is used to input the model input data into the one-dimensional convolutional neural network CNN model for training, and use the third-generation non-dominated sorting genetic algorithm to optimize the CNN model in training to obtain the third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model, which is specifically used for:

[0068] A third-generation non-dominated sorting genetic algorithm is used to establish a multi-objective optimization function with the initial learning rate, batch size and learning rate reduction factor of the CNN model as independent variables, and the dependent variables are the average deviation error and the determination coefficient, respectively. The expressions are shown in Formula 2, Formula 3 and Formula 4, respectively. The optimal network hyperparameters of the CNN model are obtained through the multi-objective optimization function;

[0069] minF={f1,f2} Formula 2;

[0070]

[0071] Among them, minF represents the solution model of NSGA-III multi-objective optimization problem, f1 is the calculation formula of average deviation error, f2 is the calculation formula of determination coefficient, y i is the predicted value of RFID scanning distance, is the actual value of the distance measurement, is the average value of the ranging prediction value set, and n is the total number of samples involved in the regression evaluation calculation;

[0072] A construction module 74 is used to divide the effective RSSI values ​​and the corresponding actual distance values ​​into a training set and a test set, use the training set to train the NSGAIII-1D-CNN model to obtain a trained RFID ranging model, and verify the RFID ranging model through the test set to obtain a final RFID ranging model;

[0073] The system further comprises:

[0074] The application module is used for performing location exploration by using the final RFID ranging model in an applicable environment.

[0075] The embodiment of the present invention is a system embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0076] In summary, compared with the prior art, the NSGAIII-1D-CNN model proposed in the embodiment of the present invention is to improve the problem that the accurate position of wireless node seismometers is difficult to obtain and the instrument may be lost in special environments in geophysical exploration and field earthquake monitoring. The embodiment of the present invention designs a new RFID ranging model based on the multi-objective optimization of one-dimensional convolutional neural network by non-dominated sorting genetic algorithm, and uses the determination coefficient and the average deviation error as the dependent variables of the multi-objective optimization function to obtain the desired network hyperparameter value, and uses the new model to achieve the purpose of RFID ranging prediction with high accuracy and certain robustness. Specifically including the following beneficial effects:

[0077] 1. The embodiment of the present invention applies the concepts of multi-objective optimization and deep learning model to the RFID ranging method for the first time, which is novel.

[0078] 2. Two sets of optimization objective functions were designed using the third-generation non-dominated sorting genetic algorithm, with the discrimination coefficient R2 and the mean deviation error MBE as dependent variables, and the three independent variables were all network parameters of the one-dimensional convolutional neural network (1D-CNN) (learning rate reduction factor, initial learning rate, batch size). After obtaining the optimal hyperparameters, a new model was constructed to improve the RFID ranging prediction accuracy performance.

[0079] 3. A method for removing outliers specifically for collecting RFID RSSI values ​​is designed to obtain a stable and reliable neural network input data set. This method uses the classic RSSI calculation formula, substitutes the RSSI value received by the RFID terminal into the formula, and uses whether the obtained environmental propagation factor value is within the threshold range to judge the validity of the RSSI value. After the anomaly removal is completed, it is necessary to collect additional RSSI valid values ​​to meet the demand for the number of neural network training.

[0080] Device Example 1

[0081] An embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps described in the method embodiment when executed by the processor.

[0082] Device Example 2

[0083] An embodiment of the present invention provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps described in the method embodiment are implemented.

[0084] The computer-readable storage medium in this embodiment includes, but is not limited to, ROM, RAM, magnetic disk or optical disk, etc.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing an RFID ranging model based on a one-dimensional convolutional neural network, characterized in that include: Acquire signal strength data and preprocess it, and use the obtained effective RSSI value as model input data; Inputting the model input data into a one-dimensional convolutional neural network (CNN) model for training, and optimizing the CNN model in training using a third-generation non-dominated sorting genetic algorithm to obtain a third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model; and The effective RSSI values ​​and their corresponding actual distance values ​​are divided into a training set and a test set. The NSGAIII-1D-CNN model is trained using the training set to obtain a trained RFID ranging model. The RFID ranging model is verified using the test set to obtain a final RFID ranging model.

2. The method according to claim 1, characterized in that The method further comprises: The final RFID ranging model is used to perform location exploration in an applicable environment.

3. The method according to claim 1, characterized in that Acquiring signal strength data and preprocessing specifically include: Collect signal strength data through the received signal strength indicator RSSI, obtain an approximate propagation factor according to the signal strength data using an approximate calculation formula as shown in Formula 1, and determine whether the approximate propagation factor is within a threshold range. If so, use the corresponding signal strength data as a valid RSSI value; if not, remove the abnormal value and re-collect the signal strength data until a valid RSSI value is obtained; PR (dBm) ≈ rssi1-10·prop·lgdist Formula 1; Among them, PR represents the receiving frequency of the wireless signal, rssi1 represents the receiving signal frequency when the wireless signal transmission distance is 1 meter, prop represents the propagation factor related to the wireless signal propagation environment, and dist represents the absolute distance between the transceiver.

4. The method according to claim 1, characterized in that: The third-generation non-dominated sorting genetic algorithm is used to optimize the CNN model in training, including: A third-generation non-dominated sorting genetic algorithm is used to establish a multi-objective optimization function with the initial learning rate, batch size and learning rate reduction factor of the CNN model as independent variables, and the dependent variables are the average deviation error and the determination coefficient, respectively. The expressions are shown in Formula 2, Formula 3 and Formula 4, respectively. The optimal network hyperparameters of the CNN model are obtained through the multi-objective optimization function; minF={f1,f2} Formula 2; Among them, minF represents the solution model of NSGA-III multi-objective optimization problem, f1 is the calculation formula of average deviation error, f2 is the calculation formula of determination coefficient, y i is the predicted value of RFID scanning distance, is the actual value of the distance measurement, is the average value of the ranging prediction value set, and n is the total number of samples involved in the regression evaluation calculation.

5. A RFID ranging model construction system based on one-dimensional convolutional neural network, characterized in that include: The data module is used to obtain the signal strength data and perform preprocessing, and use the obtained effective RSSI value as the model input data; An optimization module is used to input the model input data into a one-dimensional convolutional neural network (CNN) model for training, and optimize the CNN model in training using a third-generation non-dominated sorting genetic algorithm to obtain a third-generation non-dominated sorting genetic algorithm optimized one-dimensional convolutional neural network NSGAIII-1D-CNN model; as well as A construction module is used to divide the effective RSSI value and its corresponding actual distance value into a training set and a test set, use the training set to train the NSGAIII-1D-CNN model to obtain a trained RFID ranging model, and verify the RFID ranging model through the test set to obtain a final RFID ranging model.

6. The system according to claim 5, characterized in that The system further comprises: The application module is used for performing location exploration by using the final RFID ranging model in an applicable environment.

7. The system according to claim 5, characterized in that The data module is specifically used for: Collect signal strength data through the received signal strength indicator RSSI, obtain an approximate propagation factor according to the signal strength data using an approximate calculation formula as shown in Formula 1, and determine whether the approximate propagation factor is within a threshold range. If so, use the corresponding signal strength data as a valid RSSI value; if not, remove the abnormal value and re-collect the signal strength data until a valid RSSI value is obtained; PR (dBm) ≈ rssi1-10·prop·lgdist Formula 1; Among them, PR represents the receiving frequency of the wireless signal, rssi1 represents the receiving signal frequency when the wireless signal transmission distance is 1 meter, prop represents the propagation factor related to the wireless signal propagation environment, and dist represents the absolute distance between the transceiver.

8. The system according to claim 5, characterized in that The optimization module is specifically used for: A third-generation non-dominated sorting genetic algorithm is used to establish a multi-objective optimization function with the initial learning rate, batch size and learning rate reduction factor of the CNN model as independent variables, and the dependent variables are the average deviation error and the determination coefficient, respectively. The expressions are shown in Formula 2, Formula 3 and Formula 4, respectively. The optimal network hyperparameters of the CNN model are obtained through the multi-objective optimization function; minF={f1,f2} Formula 2; Among them, minF represents the solution model of NSGA-III multi-objective optimization problem, f1 is the calculation formula of average deviation error, f2 is the calculation formula of determination coefficient, y i is the predicted value of RFID scanning distance, is the actual value of the distance measurement, is the average value of the ranging prediction value set, and n is the total number of samples involved in the regression evaluation calculation.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method for constructing an RFID ranging model based on a one-dimensional convolutional neural network as described in any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an implementation program for information transmission, and when the program is executed by the processor, the steps of the method for constructing an RFID ranging model based on a one-dimensional convolutional neural network as described in any one of claims 1 to 4 are implemented.

Images