Ultra-wideband signal classification method and electronic equipment

Abstract

The invention provides an ultra wide band (UWB) signal classification method and electronic equipment, and the method comprises the following steps: carrying out the principal component analysis of an original UWB data set, and obtaining a new UWB data set; respectively inputting the new UWB data set into a back propagation BP neural network classification model, a genetic algorithm and back propagation GA-BP neural network classification model, a support vector machine SVM classification model and a random forest algorithm RFA classification model to obtain output results of the models; and constructing an enhanced UWB signal classification model through maximum ratio combination according to the output result of each model, and determining the classification result of the UWB signal. According to the method, a BP neural network classification model, a GA-BP neural network classification model, an SVM classification model and an RFA classification model are combined, a principal component analysis method is adopted at the input end, a maximum ratio merging method is adopted at the output end, and finally the classification accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication and artificial intelligence, and in particular to a classification method and electronic equipment for ultra-wideband signals. Background Art

[0002] With the continuous growth of urban population and the increasingly serious traffic congestion problem, subways, as a fast, efficient and environmentally friendly public transportation tool, play an increasingly significant role in people's daily travel. However, there are complex scenes such as tunnels and high-rise buildings in the subway lines, which leads to low positioning accuracy or even failure of the global navigation satellite system, and other positioning technologies are needed. Among the many wireless communication technologies, ultra-wideband (UWB) positioning technology is a more suitable choice, which has the following advantages: 1) High precision: Compared with indoor positioning technologies such as infrared (positioning accuracy of 5m~10m) and Bluetooth (positioning accuracy of 1m~3m), its positioning accuracy can reach 10cm~15cm; 2) Strong anti-interference ability: The duty cycle of UWB is generally 0.01~0.001, and the anti-interference processing gain can reach 50dB; 3) Strong reliability: The UWB positioning system adopts multi-path transmission technology to improve the signal coverage and transmission reliability. In addition, the radiation of UWB signals is very low, so it is not easily interfered when other devices work together, which is suitable for complex environments with multiple devices.

[0003] However, there are also many interference factors in UWB positioning that will affect the accuracy of UWB positioning. Especially in the subway positioning environment, UWB base stations are arranged along the narrow and long track area, and the effective ranging range of UWB signals is far. Therefore, the NLOS (Non Line of Sight) transmission error has a great impact on the accuracy of UWB positioning, and a small ranging error will lead to a large positioning error. At present, there are four main methods for classifying UWB signals transmitted in LOS (Line of Sight) and NLOS situations. The first method is the distance estimation method. This method can determine a specific NLOS channel and combine the identified channel information to formulate a rule to select a suitable NLOS range to achieve position estimation, and then complete the classification. The second method is the position estimation method. This method measures the arrival time of the radio frequency pulse transmitted between the tags to obtain the probability function of the arrival time measurement error, and then uses the tail of the probability function to calculate the position, and then complete the classification. The third method is the signal-based channel impulse response (CIR) estimation method. This method introduces an efficient node selection framework to identify LOS and NLOS transmission data based on the fluctuation and attenuation of the signal in the channel reflected by CIR, and then completes the classification. The fourth method is the LOS and NLOS recognition method based on a multi-layer convolutional neural network (CNN) and a channel attention module. This method embeds a channel attention module in a multi-layer CNN to extract the time domain data features of the original CIR, and then uses a global average pooling layer instead of a fully connected layer to integrate features and classify the output to complete the classification. These methods all extract different quantities from the LOS and NLOS channel features for comparison, and then use this as the basis for classification, but they all have their own limitations, so the classification accuracy is not high. Summary of the invention

[0004] The embodiments of the present application provide a classification method and electronic device for ultra-wideband signals, which integrate the output results of different classification models and greatly improve the classification accuracy of ultra-wideband signals for LOS and NLOS classification.

[0005] The embodiment of the present application provides a method for constructing an ultra-wideband signal classification model, comprising the following steps: performing principal component analysis on an original UWB data set to obtain a new UWB data set; inputting the new UWB data set into a BP neural network classification model, a GA-BP neural network classification model, a SVM classification model, and an RFA classification model respectively to obtain output results of each model; and constructing an enhanced UWB signal classification model according to the output results of each model through maximum ratio merging to determine the classification result of the UWB signal.

[0006] In one embodiment, the principal component analysis of the original UWB data set to obtain a new UWB data set includes: obtaining p principal components according to the characteristics of the UWB signal; dividing the original UWB data set into p UWB data subsets according to the p principal components; using each UWB data subset as a column of a matrix to form an original UWB data set matrix; solving the correlation coefficient matrix of the original UWB data set matrix, and solving the eigenvalues ​​corresponding to the p eigenroots of the correlation coefficient matrix; and obtaining a new UWB data set according to the eigenvalues ​​corresponding to the p eigenroots.

[0007] In one embodiment, the characteristics of the UWB signal include: signal flight time, total channel impulse response power, first path element detected in the channel impulse response accumulator, maximum noise value, received receive preamble sequence number, frame length, preamble length, bit rate, pulse repetition frequency rate, and absolute value of the channel impulse response.

[0008] In one embodiment, using each UWB data subset as a column of a matrix to form an original UWB data set matrix includes: standardizing each UWB data subset, and using each standardized UWB data subset as a column of a matrix to form an original UWB data set matrix.

[0009] In one embodiment, obtaining a new UWB data set based on the p characteristic roots includes: determining contribution rates of the p principal components respectively according to the eigenvalues ​​corresponding to the p characteristic roots; arranging the contribution rates of the p principal components in descending order, and selecting m UWB data subsets corresponding to the first m principal components; and merging the m UWB data subsets to form the new UWB data set.

[0010] In one embodiment, the cumulative contribution rate of the first m principal components is greater than a preset contribution rate.

[0011] In one embodiment, the new UWB data set is respectively input into a BP neural network classification model, a GA-BP neural network classification model, an SVM classification model, and an RFA classification model to obtain output results of each model, including: dividing the new UWB data set into a training set and a test set; respectively inputting the training set into the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model, and the RFA classification model for model training to obtain each trained model; inputting the test set into each trained model for model testing to obtain output results of each model; wherein the output result is used to characterize that the UWB signal belongs to line-of-sight LOS or the UWB signal belongs to non-line-of-sight NLOS.

[0012] In one embodiment, the enhanced UWB signal classification model is constructed through maximum ratio merging according to the output results of each model to determine the classification result of the UWB signal, including: obtaining the output result error of each model according to the output results of each model and the true classification marked in the test set; determining the weight of each model through maximum ratio merging according to the output results of each model, the true classification marked in the test set, and the output result error of each model; constructing an enhanced UWB signal classification model according to the output results of each model and the weight of each model to determine the classification result of the UWB signal.

[0013] In one embodiment, the weight of each model is determined by maximum ratio merging based on the output results of each model, the true classification marked in the test set, and the output result error of each model, including: solving the weight of each model when the signal-to-noise ratio is maximum; wherein the signal-to-noise ratio is related to the output results of each model, the true classification marked in the test set, the output result error of each model, and the weight of each model.

[0014] An embodiment of the present application further provides an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the above-mentioned ultra-wideband signal classification method.

[0015] The solution provided in the above embodiments of the present application combines the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model and the RFA classification model, adopts the principal component analysis method at the input end and the maximum ratio merging method at the output end, ultimately achieving an improvement in classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solution of the embodiments of the present application, the drawings required for use in the embodiments of the present application are briefly introduced below.

[0017] Figure 1 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0018] Figure 2 It is a flowchart of the UWB signal classification method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0020] Similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0021] Figure 1 1 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 100 can be used to perform the UWB signal classification method provided in an embodiment of the present application. Figure 1 As shown, the electronic device 100 includes: one or more processors 102, and one or more memories 104 storing processor executable instructions. The processor 102 is configured to execute the UWB signal classification method provided in the following embodiment of the present application.

[0022] The processor 102 can be a gateway, a smart terminal, or a device including a central processing unit (CPU), a graphics processing unit (GPU), or other forms of processing units with data processing capabilities and / or instruction execution capabilities. It can process data of other components in the electronic device 100 and control other components in the electronic device 100 to perform desired functions.

[0023] The memory 104 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 102 may run the program instructions to implement the classification method of UWB signals described below. Various applications and various data, such as various data used and / or generated by the application, may also be stored in the computer-readable storage medium.

[0024] In one embodiment, Figure 1 The electronic device 100 shown may also include an input device 106, an output device 108, and a data acquisition device 110, which are interconnected via a bus system 112 and / or other forms of connection mechanisms (not shown). Figure 1 The components and structures of the electronic device 100 shown are merely exemplary and non-limiting. The electronic device 100 may also have other components and structures as required.

[0025] The input device 106 may be a device used by a user to input instructions, and may include one or more of a keyboard, a mouse, a microphone, a touch screen, etc. The output device 108 may output various information (e.g., images or sounds) to the outside (e.g., a user), and may include one or more of a display, a speaker, etc. The data acquisition device 110 may acquire the original UWB data set through the DW1000 chip, and store the acquired data in the memory 104 for use by other components. Exemplarily, the data acquisition device 110 may be a DWM1000 UWB radio module.

[0026] In one embodiment, the various components in the example electronic device 100 for implementing the UWB signal classification method of the embodiment of the present application can be integrated or dispersed, such as integrating the processor 102, memory 104, input device 106 and output device 108 into one, while separating the data acquisition device 110.

[0027] In one embodiment, the example electronic device 100 for implementing the UWB signal classification method of the embodiment of the present application can be implemented as an intelligent terminal such as a smart phone, a tablet computer, a desktop computer, a server, a vehicle-mounted device, etc.

[0028] Figure 2 FIG. 1 is a flow chart of a method for classifying UWB signals according to an embodiment of the present application. Figure 2 As shown, the method includes the following steps 210-230.

[0029] Step 210: Perform principal component analysis on the original UWB data set to obtain a new UWB data set.

[0030] Regarding the original UWB data set, it includes a large amount of unclassified and unarranged UWB data, which are collected by the DW1000 chip and obtained after being labeled (labeling the UWB signal as belonging to LOS or labeling the UWB signal as belonging to NLOS).

[0031] Regarding the principal component of the original UWB data set, it can be the feature of the UWB signal, or it can be obtained by a simple transformation based on the feature of the UWB signal, for example, the feature of the UWB signal is standardized to obtain the principal component, or the feature of the UWB signal is normalized to obtain the principal component.

[0032] The above-mentioned principal component analysis is performed on the original UWB data set to obtain a new UWB data set, which specifically includes: obtaining p principal components according to the characteristics of the UWB signal; dividing the original UWB data set into p UWB data subsets according to the p principal components; using each UWB data subset as a column of a matrix to form an original UWB data set matrix; solving the correlation coefficient matrix of the original UWB data set matrix, and solving the eigenvalues ​​corresponding to the p eigenroots of the correlation coefficient matrix; and obtaining the new UWB data set according to the eigenvalues ​​corresponding to the p eigenroots.

[0033] The characteristics of the above-mentioned UWB signal include: signal flight time, total channel impulse response power, the first path element detected in the channel impulse response accumulator, the maximum noise value, the received receive (RX) preamble number, the frame length, the preamble length, the bit rate, the pulse repetition frequency rate, and the absolute value of the channel impulse response.

[0034] The above-mentioned method of using each UWB data subset as a column of a matrix to form the original UWB data set matrix specifically includes: standardizing each UWB data subset, and using each standardized UWB data subset as a column of a matrix to form the original UWB data set matrix.

[0035] The above-mentioned method obtains a new UWB data set based on the p characteristic roots, specifically including: determining the contribution rates of the p principal components respectively according to the eigenvalues ​​corresponding to the p characteristic roots; arranging the contribution rates of the p principal components in descending order, and selecting m UWB data subsets corresponding to the first m principal components; and merging the m UWB data subsets to form the new UWB data set.

[0036] The cumulative contribution rate of the first m principal components is greater than a preset contribution rate, and the preset contribution rate can be preset according to an empirical value.

[0037] The m UWB data subsets mentioned above can be arranged in descending order according to the principal component contribution rate. Therefore, the data in the new UWB data set composed of them is also arranged in descending order according to the principal component contribution rate.

[0038] Specifically, the UWB data set after data centering is a matrix X, which has a total of p principal components. i represents the UWB data corresponding to the i-th principal component. There are n sets of data in total. X i is one of the p UWB data subsets. Specifically expressed as:

[0039] X=(X1,X2,…,X i ,…,Xp)

[0040] X i =(X i1 ,Xi2 ,…,X ij ,…,X in ) T

[0041] The matrix X is standardized to obtain the matrix Z (that is, the original UWB data set matrix), and the correlation coefficient matrix R of the matrix Z is solved, which is specifically expressed as:

[0042]

[0043] Solve the eigenvalues ​​corresponding to the p eigenroots of the correlation coefficient matrix R, which is specifically expressed as:

[0044] (λ1,λ2,…,λ p )

[0045] The contribution rate of each principal component is obtained through the eigenvalues ​​corresponding to the p eigenroots. Taking the i-th principal component as an example, its contribution rate G i Specifically expressed as:

[0046]

[0047] Arrange the contribution rates of the p principal components in descending order, select the top m principal components (m≤p) among the p principal components, and obtain the cumulative contribution rate C of the first m principal components according to the contribution rate of each principal component, which can be expressed as:

[0048]

[0049] The cumulative contribution rate C needs to be greater than the preset contribution rate, such as 70%. Therefore, the value of m can be inferred from the above cumulative contribution rate formula. For example, m can be 12, and the first 12 principal components are selected as model input to achieve input data dimensionality reduction preprocessing, reduce the complexity of model training, and the cumulative contribution rate of these 12 principal components is greater than 70%, which can ensure that a qualified model is trained.

[0050] Step 220: Input the new UWB data set into the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model, and the RFA classification model respectively to obtain the output results of each model.

[0051] The prediction and training of the above four classification models are relatively similar. The BP neural network classification model is used as an example below, and the other classification models can be used as reference.

[0052] Inputting the new UWB data set into the BP neural network classification model to obtain the output result of the model can specifically include: dividing the new UWB data set into a training set and a test set; inputting the training set into the BP neural network classification model for model training to obtain the trained BP neural network classification model; inputting the test set into the trained BP neural network classification model for model testing to obtain the output result of the BP neural network classification model.

[0053] The output results of the GA-BP neural network classification model, the SVM classification model, and the RFA classification model can also be obtained using the above method, but different models can be set with different parameters during model training. For example, different models can be set with different adjustment learning rates, different training rounds, etc.

[0054] The new UWB data set is divided into a training set and a test set. For example, the first 500 data of the new UWB data set can be selected as the test set, and the remaining 5500 data can be selected as the training set for model training.

[0055] The above output results are used to characterize that the UWB signal belongs to line-of-sight LOS or the UWB signal belongs to non-line-of-sight NLOS.

[0056] For the above four classification models, after obtaining the output results of each model, the classification accuracy of each classification model can also be counted. The specific method can be: compare the output results with the real classification marked in the test set (marking UWB signals as LOS or marking UWB signals as NLOS). If the output results are consistent with the markings, it means that the output results are correct. If they are inconsistent, it means that the output results are wrong. Compare the number of correct output results with the total number of output results to obtain the classification accuracy of each classification model. Since each classification model is trained based on a different initial model, each initial model has its own advantages and limitations. Therefore, the classification accuracy of a single classification model is not high.

[0057] Step 230: Based on the output results of each model, an enhanced UWB signal classification model is constructed through maximum ratio combining to determine the UWB signal classification result.

[0058] The above-mentioned step 230 may specifically include: obtaining the output result error of each model according to the output results of each model and the true classification marked in the test set; constructing an enhanced UWB signal classification model through maximum ratio merging according to the output results of each model, the true classification marked in the test set, and the output result error of each model, and determining the weight of each model; determining the classification result of the UWB signal according to the output results of each model and the weight of each model.

[0059] Regarding maximum ratio combining, it is a signal recovery model in communication, which aims to "maximize the received signal-to-noise ratio (SNR)". Here, it draws on the communication method and is applied to the classification of UWB signals.

[0060] Regarding the enhanced UWB signal classification model, its main function is to classify the UWB signal into LOS and NLOS according to the input UWB signal.

[0061] The above SNR is related to the output results of each model, the true classification marked in the test set, the output result error of each model, and the weight of each model. The output results of each model have been obtained in step 220. The true classification marked in the test set is the data contained in the new UWB data set itself. The output result error of each model can be obtained through the output results of each model and the true classification marked in the test set, which is also known. Therefore, the weight of each model can be solved when the SNR is maximum, and then the final classification result of the UWB signal can be obtained based on the weight and the output results of each model.

[0062] More specifically, suppose X est1 , X est2 , X est3 , X est4 They represent the output results of the four classification model test sets, and they are all column vectors. ACC represents the true classification marked in the test set, and N1, N2, N3, and N4 represent the error between the true classification and the output result. The relationship is:

[0063]

[0064] Y represents the matrix composed of the output results of different models, expressed as:

[0065]

[0066] Let W = (w1, w2, w3, w4) represent the weights of different classification models, we can get:

[0067]

[0068] Then the signal-to-noise ratio SNR can be expressed as:

[0069]

[0070] In order to maximize the SNR, is the true classification annotated in the test set and should not be affected by the weight assignment. Therefore, the above maximization of SNR is converted into an optimization problem of minimizing the above denominator, which is expressed as:

[0071]

[0072] Because w i (i=1,2,3,4) represents the weight, so w1+w2+w3+w4=1.

[0073] According to the Lagrange multiplier method, the optimal solution to the above optimization problem can be solved to satisfy the following equation:

[0074]

[0075] Therefore, the result of the original result after the maximum ratio merging process is:

[0076]

[0077] After the above process, the enhanced UWB signal classification model can be constructed through maximum ratio combining, and the above result is used as the final classification result of the UWB signal.

[0078] After obtaining the classification results of the UWB signal, the classification accuracy can also be counted. The specific method can be: compare the classification results with the real classifications marked in the test set (marking the UWB signal as LOS or marking the UWB signal as NLOS). If the classification results are consistent with the markings, it means that the output result is correct. If they are inconsistent, it means that the output result is wrong. Compare the number of correct classification results with the total number of classification results to obtain the classification accuracy of the classification results. Actual statistics show that the classification accuracy of the classification results is much higher than the classification accuracy of the output results of the above four models.

[0079] The solution provided in the above embodiments of the present application combines the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model and the RFA classification model, adopts the principal component analysis method at the input end and the maximum ratio merging method at the output end, ultimately achieving an improvement in classification accuracy.

[0080] In several embodiments provided in the present application, the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and a part of a module, a program segment or a code contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0081] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0082] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

Claims

1. A classification method for ultra-wideband signals, characterized in that: The following steps are involved: Perform principal component analysis on the original ultra-wideband UWB dataset to obtain a new UWB dataset; The new UWB data set is respectively input into the back propagation BP neural network classification model, the genetic algorithm and back propagation GA-BP neural network classification model, the support vector machine SVM classification model, and the random forest algorithm RFA classification model to obtain the output results of each model; According to the output results of each model, an enhanced UWB signal classification model is constructed through maximum ratio combining to determine the classification result of the UWB signal.

2. The method for classifying ultra-wideband signals according to claim 1, characterized in that: The principal component analysis is performed on the original UWB data set to obtain a new UWB data set, including: According to the characteristics of the UWB signal, p principal components are obtained; According to the p principal components, the original UWB data set is divided into p UWB data subsets; Each UWB data subset is used as a column of the matrix to form the original UWB data set matrix; Solving the correlation coefficient matrix of the original UWB data set matrix, and solving the eigenvalues ​​corresponding to p eigenroots of the correlation coefficient matrix; A new UWB data set is obtained according to the eigenvalues ​​corresponding to the p eigenroots.

3. The method for classifying ultra-wideband signals according to claim 2, characterized in that: The characteristics of the UWB signal include: signal flight time, total channel impulse response power, the first path element detected in the channel impulse response accumulator, the maximum noise value, the received receive preamble number, the frame length, the preamble length, the bit rate, the pulse repetition frequency rate, and the absolute value of the channel impulse response.

4. The method for classifying ultra-wideband signals according to claim 2, characterized in that: The method uses each UWB data subset as a column of a matrix to form an original UWB data set matrix, including: Each UWB data subset is standardized, and each standardized UWB data subset is used as a column of a matrix to form the original UWB data set matrix.

5. The method for classifying ultra-wideband signals according to claim 2, characterized in that: The obtaining of a new UWB data set according to the eigenvalues ​​corresponding to the p eigenroots includes: Determining contribution rates of the p principal components respectively according to eigenvalues ​​corresponding to the p eigenvalues; Arrange the contribution rates of the p principal components in descending order, and select m UWB data subsets corresponding to the first m principal components; The m UWB data subsets are combined to form the new UWB data set.

6. The method for constructing a classification model for ultra-wideband signals according to claim 5, characterized in that: The cumulative contribution rate of the first m principal components is greater than the preset contribution rate.

7. The method for classifying ultra-wideband signals according to claim 1, characterized in that: The new UWB data set is input into the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model, and the RFA classification model respectively to obtain the output results of each model, including: Dividing the new UWB data set into a training set and a test set; The training sets are respectively input into the BP neural network classification model, the GA-BP neural network classification model, the SVM classification model, and the RFA classification model for model training to obtain the trained models; Inputting the test set into each of the trained models to perform model testing to obtain output results of each model; The output result is used to indicate whether the UWB signal belongs to line-of-sight LOS or the UWB signal belongs to non-line-of-sight NLOS.

8. The method for classifying ultra-wideband signals according to claim 7, characterized in that: The method of constructing an enhanced UWB signal classification model based on the output results of each model by maximum ratio merging to determine the classification result of the UWB signal includes: According to the output results of each model and the true classification marked in the test set, the output result error of each model is obtained; Determine the weight of each model by maximum ratio merging according to the output results of each model, the true classification marked in the test set, and the output result error of each model; According to the output results of each model and the weight of each model, an enhanced UWB signal classification model is constructed to determine the classification result of the UWB signal.

9. The ultra-wideband signal classification method according to claim 8, characterized in that: The step of determining the weight of each model by maximum ratio merging according to the output results of each model, the true classifications marked in the test set, and the output result errors of each model includes: The weights of the models are obtained when the signal-to-noise ratio is the largest; The signal-to-noise ratio is related to the output results of each model, the true classification marked in the test set, the output result error of each model, and the weight of each model.

10. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing processor-executable instructions; The processor is configured to execute the ultra-wideband signal classification method according to any one of claims 1 to 9.

Images