Multi-band microwave radio frequency test method and system

Through the multi-band microwave RF testing method, frequency analysis and variational autoencoder optimization model are used to identify abnormal signals and analyze interference source characteristics, which solves the shortcomings of performance evaluation of multi-band systems in the existing technology, and improves the stability of the equipment and performance.

CN118890106AInactive Publication Date: 2024-11-01CHENGDU CHANGHESHUN ELECTRONIC TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410905330.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-08
Publication Date
2024-11-01
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing RF testing technologies are difficult to comprehensively evaluate the comprehensive performance of multi-band systems, especially in the problems of interference between signals and error accumulation, which may lead to unstable performance in actual applications of equipment and complex and time-consuming debugging and maintenance.

Method used

The multi-band microwave radio frequency testing method is used to analyze the segmented signals through frequency, train the variational autoencoder, optimize the encoding model, identify abnormal signals, analyze interference source characteristics, generate interference source characteristics sets, and improve fault detection sensitivity and accuracy.

Benefits of technology

In-depth analysis of multi-band signal characteristics is realized, reconstruction error is reduced, signal processing accuracy and efficiency are improved, and the operation stability and multi-band performance of the equipment are enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118890106A_ABST
    Figure CN118890106A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of radio frequency testing, in particular to a multi-band microwave radio frequency testing method and system, and the method comprises the following steps: receiving a multi-band microwave radio frequency signal, subdividing the signal into differentiated frequency bands through a frequency analysis technology, and carrying out the amplitude and frequency data collection of each frequency band. According to the method, deep analysis of signal characteristics is realized by subdividing multiband microwave radio-frequency signals and collecting detailed amplitude and frequency data, data are trained by adopting a variational auto-encoder, a model is optimized, reconstruction errors are reduced, and the accuracy and efficiency of signal processing are enhanced; by analyzing the difference between a reconstructed signal and an original signal, the standard of an abnormal signal is accurately set, the sensitivity and accuracy of fault detection are improved, an interference source is analyzed from the identified abnormal signal, detailed data support is provided for system maintenance and optimization, the adaptability of equipment to environment change is enhanced, and the fault detection efficiency is improved. And the operation stability and the multi-band performance of the wireless equipment are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of radio frequency testing technology, and in particular to a multi-band microwave radio frequency testing method and system. Background Art

[0002] RF testing is an important electronic test method used to measure and evaluate the performance of wireless devices and systems in the RF spectrum. This technical field includes various test instruments and technologies, such as spectrum analyzers, network analyzers, signal generators, and power meters, which are used to test the transmission and reception functions of the equipment to ensure that it operates effectively and safely within the specified frequency range. In addition, RF testing also focuses on the performance of the equipment in the actual operating environment, including signal stability, interference handling, and signal coverage. These tests are crucial to the development and maintenance of wireless communication systems, covering a wide range of applications such as mobile communications, satellite communications, radio and television, and military communications.

[0003] Among them, the multi-band microwave RF test method specifically refers to the technology and method of performing microwave RF testing in multiple different frequency bands. This test method is particularly suitable for complex wireless systems that need to operate in multiple frequency bands, such as multi-mode mobile phones, broadband wireless network equipment, and radar and satellite communication systems that work across frequency bands. Through multi-band testing, the performance of the device in each frequency band can be evaluated, including transmission power, frequency response, interference tolerance, and signal quality, so as to ensure that the device can meet the expected performance standards in all operating frequency bands. This test is critical to ensuring the reliability and compatibility of communication equipment and is an important step in the design and production process of wireless equipment.

[0004] Although existing RF test technologies provide a wide range of instruments and methods, they often lack comprehensive performance evaluation of multi-band and complex systems. In particular, in terms of the interaction and detail capture of multi-band systems, traditional methods have failed to fully address the problems of interference and error accumulation between signals, resulting in unstable performance of equipment in actual applications. For example, when dealing with multi-mode devices, traditional tests may not accurately reflect the interference effects between different frequency bands, which may cause signal distortion or breakage in critical communication applications, affecting the overall reliability of the system. In addition, these technologies also show limitations in quickly identifying and adjusting frequency band anomalies, making equipment debugging and maintenance more complex and time-consuming. This technical limitation is particularly prominent in the face of the growing demand for wireless communications, limiting the performance of equipment in high-efficiency and high-compatibility scenarios. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a multi-band microwave radio frequency testing method and system.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a multi-band microwave radio frequency testing method, comprising the following steps:

[0007] S1: Receive multi-band microwave radio frequency signals, apply frequency analysis technology to subdivide the signals into differentiated frequency bands, collect amplitude and frequency data for each frequency band, record the intensity and time series of multi-band signals, and generate frequency-stratified data;

[0008] S2: using the multiple frequency band data obtained from the frequency layered data to train a variational autoencoder, adjusting encoder parameters, matching multi-band characteristics, optimizing the model to reduce reconstruction errors, and obtaining an optimized encoding model;

[0009] S3: Processing the currently received microwave radio frequency signal through the optimized coding model, comparing the difference between the model reconstructed signal and the original signal, setting the abnormal signal discrimination standard according to the size of the difference, and obtaining the abnormal signal recognition result;

[0010] S4: extracting abnormal signal features of each frequency band from the abnormal signal identification results, analyzing and summarizing the estimated types and sources of interference sources, and summarizing and generating an interference source feature set.

[0011] As a further solution of the present invention, the step of subdividing the differentiated frequency bands is specifically as follows:

[0012] S111: receiving a multi-band microwave radio frequency signal from a differentiated source, applying an improved frequency band separation algorithm to subdivide the signal into differentiated frequency bands, using the formula:

[0013]

[0014] Calculate the initial amplitude and frequency of the multi-band signal to generate a separated signal sequence Z i ;

[0015] Among them, x n represents the nth sample of the original signal, f i represents the center frequency of the jth frequency band, N represents the total number of samples, and j is an imaginary unit;

[0016] S112: Separating the signal sequence S i Perform amplitude analysis and use the composite amplitude calculation formula:

[0017]

[0018] Calculate the amplitude of multiple frequency bands and generate multi-frequency band amplitude data A i ;

[0019] in, and Represent the real and imaginary parts of the complex number respectively, and k is the weight factor used to adjust the influence of the logarithmic term;

[0020] S113: Based on the multi-band amplitude data A i Set the amplitude threshold for dynamic adjustment, using the formula:

[0021]

[0022] Matching the non-static characteristics of the signal to generate differentiated frequency bands P i ;

[0023] Among them, T a is the basic amplitude threshold, α and ω are the adjustment amplitude and frequency parameters of the amplitude threshold, respectively.

[0024] As a further solution of the present invention, the step of acquiring the frequency layered data is specifically as follows:

[0025] S121: Based on the differentiated frequency band P i , the integral transform method is used to analyze each frequency band and calculate the amplitude and frequency data using the formula:

[0026]

[0027] Calculate the frequency components and generate the frequency band data set F i ;

[0028] Where T is the sampling period, t is the time variable, f is the frequency variable, and j represents the imaginary unit;

[0029] S122: The frequency band data set F i Perform time series analysis and use sine wave superposition and time delay analysis to extract patterns in the time series using the formula:

[0030] R i (t)=∑ t τ=0 F i (τ)·sin(ω·τ+φ)

[0031] Thus, the time series model R is established i ;

[0032] Among them, F i is the frequency band data set, ω is the fundamental frequency, φ is the phase angle, and τ is the time delay variable used to represent the change of the frequency band over time;

[0033] S123: Using the time series model R i , by calculating the peak-to-valley difference, using the formula:

[0034] Si =max t∈T (R i (t))-min t∈T (R i (t))

[0035] Evaluate the stability and reliability of the signal and generate frequency-stratified data S i ;

[0036] Among them, R is the time series model, T represents the analysis period, and t is the time variable.

[0037] As a further solution of the present invention, the step of obtaining the optimized coding model is specifically as follows:

[0038] S211: Extract multi-band data from frequency-stratified data and synthesize feature vectors by weighted average method, using the formula:

[0039]

[0040] Apply weights to each frequency band data and add regularization terms to avoid overfitting to generate a weighted frequency band feature vector D i ;

[0041] Among them, S f (i) represents the signal strength of the f-th frequency band at the i-th sample point, w f is the weight of the frequency band, λ is the regularization parameter used to control the weight and;

[0042] S212: Adjust the parameters of the variational autoencoder to match the multi-band characteristics. The optimization algorithm uses the adjusted loss function using the formula:

[0043]

[0044] Add adjustment parameters to balance reconstruction loss and KL divergence and generate adjusted model parameters θ * ;

[0045] Among them, x i is the input data, z i is the latent variable, θ is the model parameter, and β is the loss adjustment factor, which is used to optimize the trade-off between the encoder and the decoder;

[0046] S213: Use the optimized model to reduce the reconstruction error and perform error evaluation. Apply the formula:

[0047]

[0048] Calculate the average reconstruction error of the model and refer to the continuity of the model output to generate the optimized coding model E;

[0049] Where N is the number of data points, x i is the actual data point, is the data point for model reconstruction, and γ is the continuity adjustment parameter used to evaluate the stability and continuity performance of the model.

[0050] As a further solution of the present invention, the step of obtaining the abnormal signal recognition result is specifically:

[0051] S311: Use the optimized coding model to process the currently received microwave radio frequency signal, perform signal preprocessing, and apply the improved formula:

[0052] R i =α·Model(S i )+β·log(1+|S i |)

[0053] Generate model to reconstruct signal R i ;

[0054] Among them, S i is the original microwave RF signal sample, Model represents the optimized coding model, α and β adjust the sensitivity and dynamic range of signal processing;

[0055] S312: Compare the model reconstructed signal R i With the original signal S i The difference is applied, and the improved difference measurement formula is used, using the formula:

[0056]

[0057] Calculate the square root of the sum of squared weighted errors of the overall signal to generate a difference metric D;

[0058] Among them, R is the reconstructed signal output by the model, S i is the original signal, N is the number of signal samples, σ i is the standard deviation of the i-th signal sample;

[0059] S313: Using the difference metric D, setting the abnormal signal discrimination standard according to the size of the difference, by setting the threshold method, using the formula:

[0060]

[0061] Determine whether the signal is abnormal and obtain abnormal signal identification result C;

[0062] Where D is the signal difference measure, θ is the basic threshold of the difference measure, and δ is the adjustment factor for environmental noise or changes in system conditions.

[0063] As a further solution of the present invention, the step of acquiring the interference source feature set is specifically:

[0064] S411: Extracting abnormal signal features of each frequency band from the abnormal signal identification results, using a feature extraction formula:

[0065]

[0066] Calculate the weighted abnormal feature F of each frequency band ei ;

[0067] Where M is the number of samples in each frequency band, X im is the signal strength of the i-th frequency band at the m-th sample, σ im is the weight of the abnormal signal, λ is the nonlinear adjustment parameter of the signal strength, which enhances the sensitivity of the model to the signal characteristics;

[0068] S412: Based on the weighted abnormal feature F ei , analyze and summarize the estimated types and sources of interference sources, and apply the analysis formula:

[0069]

[0070] Generate a summary score G for each interference source type i ;

[0071] Where E is the total number of interference source types, κ e is the classification weight of the interference source, μ is the feature balance adjustment coefficient;

[0072] S413: G is the summary score of each interference source type. i , by summarizing the formula:

[0073]

[0074] Summarize into interference source feature set H;

[0075] Among them, I is the number of reference interference source types, and δ is the score enhancement factor, which is used to emphasize the target interference source.

[0076] A multi-band microwave radio frequency testing system, the multi-band microwave radio frequency testing system is used to perform the multi-band microwave radio frequency testing method, the system comprises:

[0077] After receiving the multi-band microwave RF signal, the signal receiving module applies frequency analysis to the signal, subdividing it into multiple differentiated frequency bands, collecting the amplitude and frequency data of each frequency band, and recording the signal strength and time series to generate frequency-layered data;

[0078] The data training module selects multi-band data to train the variational autoencoder based on the frequency layered data, adjusts the encoder parameters to match the multi-band characteristics, performs model optimization to reduce the reconstruction error, and obtains an optimized encoding model;

[0079] The abnormality recognition module uses the optimized coding model to process the currently received microwave radio frequency signal, calculates the difference between the reconstructed signal and the original signal, sets the judgment standard of the abnormal signal, recognizes and calibrates the abnormality, and obtains the abnormal signal recognition result;

[0080] The feature analysis module extracts the abnormal signal features of each frequency band from the abnormal signal identification results, analyzes multiple features to estimate the type and source of the interference source, and forms preliminary interference source features;

[0081] The interference source summary module analyzes the preliminary interference source characteristics, summarizes and improves the interference source information, enhances the integration and practicability of the data, and generates an interference source feature set.

[0082] Compared with the prior art, the advantages and positive effects of the present invention are:

[0083] In the present invention, by subdividing multi-band microwave radio frequency signals and collecting detailed amplitude and frequency data, in-depth analysis of signal characteristics is achieved. Variational autoencoder training data and optimizing models are used to reduce reconstruction errors and enhance the accuracy and efficiency of signal processing. The standard of abnormal signals is accurately set by analyzing the difference between the reconstructed signal and the original signal, thereby improving the sensitivity and accuracy of fault detection. The interference source is analyzed from the identified abnormal signal, providing detailed data support for system maintenance and optimization, strengthening the adaptability of the equipment in the face of environmental changes, and increasing the operational stability and multi-band performance of wireless equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0085] Figure 2 A flow chart of the subdivision steps of differentiated frequency bands of the present invention;

[0086] Figure 3 A flow chart of the steps for obtaining frequency layered data of the present invention;

[0087] Figure 4 A flowchart of the steps for obtaining the optimized coding model of the present invention;

[0088] Figure 5 A flowchart of the steps for obtaining abnormal signal recognition results of the present invention;

[0089] Figure 6 The figure is a flow chart of the steps for obtaining the interference source feature set of the present invention. DETAILED DESCRIPTION

[0090] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0091] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0092] Embodiment 1

[0093] See also Figure 1 The present invention provides a technical solution: a multi-band microwave radio frequency testing method, comprising the following steps:

[0094] S1: Receive multi-band microwave radio frequency signals, apply frequency analysis technology to subdivide the signals into differentiated frequency bands, collect amplitude and frequency data for each frequency band, record the intensity and time series of multi-band signals, and generate frequency-stratified data;

[0095] S2: Use multiple frequency band data obtained from the frequency layered data to train the variational autoencoder, adjust the encoder parameters, match the multi-band characteristics, optimize the model to reduce the reconstruction error, and obtain the optimized coding model;

[0096] S3: Process the currently received microwave radio frequency signal by optimizing the coding model, compare the difference between the model reconstructed signal and the original signal, set the abnormal signal discrimination standard according to the size of the difference, and obtain the abnormal signal recognition result;

[0097] S4: Extract the abnormal signal features of each frequency band from the abnormal signal identification results, analyze and summarize the estimated types and sources of interference sources, and generate an interference source feature set.

[0098] Frequency stratification data specifically includes frequency band division, amplitude acquisition, and frequency acquisition. Optimization of the coding model specifically refers to parameter adjustment, feature matching, and error optimization. Abnormal signal identification results specifically include judgment criterion setting and abnormal signal comparison. Interference source feature sets specifically include abnormal feature extraction, type estimation, and source analysis.

[0099] See also Figure 2, the subdivision steps of differentiated frequency bands are as follows:

[0100] S111: receiving a multi-band microwave radio frequency signal from a differentiated source, applying an improved frequency band separation algorithm to subdivide the signal into differentiated frequency bands, using the formula:

[0101]

[0102] Calculate the initial amplitude and frequency of the multi-band signal to generate a separated signal sequence S i ;

[0103] Among them, x n represents the nth sample of the original signal, f i represents the center frequency of the jth frequency band, N represents the total number of samples, and j is an imaginary unit;

[0104] S112: Separating the signal sequence S i Perform amplitude analysis and use the composite amplitude calculation formula:

[0105]

[0106] Calculate the amplitude of multiple frequency bands and generate multi-frequency band amplitude data A i ;

[0107] in, and Represent the real and imaginary parts of the complex number respectively, and k is the weight factor used to adjust the influence of the logarithmic term;

[0108] S113: Based on the multi-band amplitude data A i Set the amplitude threshold for dynamic adjustment, using the formula:

[0109]

[0110] Matching the non-static characteristics of the signal to generate differentiated frequency bands P i ;

[0111] Among them, T a is the basic amplitude threshold, α and ω are the adjustment amplitude and frequency parameters of the amplitude threshold, respectively.

[0112] formula:

[0113]

[0114] This is a discrete Fourier transform (DFT) expression, which is used to convert a time domain signal into a frequency domain signal.

[0115] Parameter explanation:

[0116] x n: The nth sample of the original signal.

[0117] N: The total number of samples, i.e. the length of the signal.

[0118] f i : Center frequency of the second frequency band.

[0119] j: imaginary unit.

[0120] Derivation process:

[0121] Assume there is a signal sequence x = [x0, x1, x2, ..., x N-1 ], where each x n Represents the amplitude of the signal at time point n.

[0122] To convert the signal to the frequency domain, we apply a discrete Fourier transform to it.

[0123] For the target frequency band i, calculate its frequency component S i This is done by taking the signal samples x at all time points n Multiply And sum to achieve.

[0124] Calculation example:

[0125] Assume N=4 (signal length) and the signal samples are [1,2,3,4].

[0126] Select i=1 (first frequency band), f1=0.25 (frequency).

[0127] Calculate Z i :

[0128] Z i =1·e -j·2π·0.25·0 / 4 +2·e -j·2π·0.25·1 / 4 +3·e -j·2π·0.25·2 / 4 +4·e -j·2π·0.25·3 / 4

[0129] =1·1+2·e -j·π / 2 +3·e -j·π +4·e -j·3π / 2

[0130] =1+2(-j)-3-4j

[0131] =-2-6j

[0132] The results show that in the frequency band of 0.25 cycle / sample, the complex frequency component of the signal is -2-6j.

[0133] formula:

[0134]

[0135] The formula calculates the amplitude, and a logarithmic term is added to enhance the recognition accuracy of small amplitude signals.

[0136] Parameter explanation:

[0137] Z i : The obtained band-separated signal.

[0138] and Z i The real and imaginary parts of .

[0139] k: weight factor, used to adjust the influence of the logarithmic term.

[0140] Derivation process:

[0141] Using the Z i =-2-6j.

[0142] Calculate A1:

[0143]

[0144] Assume k = 0.5, then:

[0145]

[0146] The amplitude value A1=6.4 indicates the signal amplitude in frequency band 1, and the weighted logarithmic term improves the recognition ability of weak signals.

[0147] Reasonable interval parameter setting

[0148] For frequency f i , which is below half the signal sampling rate (the Nyquist frequency). In a practical setting, f i is a value from 0 to half the sampling rate. For example, if the sampling rate is 8000Hz, then f i It can be from 0 to 4000 Hz. The common range of weight factor k can be 0.1 to 1.0. The numerical value ensures the practical feasibility and scientific rationality of the parameter.

[0149] See also Figure 3 , the specific steps for obtaining frequency layered data are:

[0150] S121: Based on differentiated frequency band D i , the integral transform method is used to analyze each frequency band and calculate the amplitude and frequency data using the formula:

[0151]

[0152] Calculate the frequency components and generate the frequency band data set F i ;

[0153] Where T is the sampling period, t is the time variable, f is the frequency variable, and j represents the imaginary unit;

[0154] S122: Frequency band data set F i Perform time series analysis and use sine wave superposition and time delay analysis to extract patterns in the time series using the formula:

[0155]

[0156] Thus, the time series model R is established i ;

[0157] Among them, F i is the frequency band data set, ω is the fundamental frequency, φ is the phase angle, and τ is the time delay variable used to represent the change of the frequency band over time;

[0158] S123: Using Time Series Models in R i , by calculating the peak-to-valley difference, using the formula:

[0159] S i =max t∈T (R i (t))-min t∈T (R i (t))

[0160] Evaluate the stability and reliability of the signal and generate frequency-stratified data S i ;

[0161] Among them, R is the time series model, T represents the analysis period, and t is the time variable.

[0162] formula:

[0163]

[0164] Parameter explanation:

[0165] P i (t): The intensity of the differentiated frequency band signal at time t.

[0166] T: Sampling period, set to 1 second, indicating a complete sampling period.

[0167] f: frequency variable, set to 1 Hz, indicating that the reference is a signal that changes once per second.

[0168] j: Imaginary unit, used to represent complex numbers.

[0169] t: time variable, from 0 to T.

[0170] Calculation process:

[0171] Assume P i (t) is a sine wave, for example, P i (t) = sin(2πt), which represents the change within one cycle.

[0172] P i (t) and the exponential term are substituted into the integral.

[0173] Compute the integral:

[0174] The result of the integration is a complex number representing the amplitude and phase of the frequency component.

[0175] Calculation example:

[0176] Assume t is from 0 to 1 second, frequency f = 1 Hz, period T = 1 second. Calculate the integral:

[0177]

[0178] The result of the integration will give the complex spectral components of the signal at a frequency of 1 Hz. The modulus of the complex number represents the amplitude of the frequency component, and the angle represents the phase.

[0179] formula:

[0180]

[0181] Parameter explanation:

[0182] F i (τ): Calculated frequency component.

[0183] ω: base frequency, set to 1Hz.

[0184] φ: Phase angle, set to 0 (assuming no phase shift).

[0185] τ: time delay variable, from 0 to t.

[0186] Calculation process:

[0187] Using F i (τ).

[0188] Accumulate τ at each time point, considering the superposition of sine waves.

[0189] The result of the summation expression represents the cumulative effect of the signal at time t.

[0190] Calculation example:

[0191] Assume t = 1 second, F at each time point i (τ) is obtained from the above calculation. It is necessary to set all F i(τ) are superimposed according to the rules of sine functions.

[0192] formula:

[0193]

[0194] Parameter explanation:

[0195] R i (t): The calculated time series model.

[0196] T: Analysis period, which is still set to 1 second here.

[0197] Calculation process:

[0198] Capture R i (t) Maximum and minimum values ​​within period T.

[0199] The difference between these two values ​​is calculated, representing the dynamic range of the signal.

[0200] Calculation example:

[0201] According to the time series model, the maximum and minimum values ​​are captured and then the difference is calculated. The difference reflects the magnitude of the signal change during the observation period.

[0202] See also Figure 4 , the specific steps for obtaining the optimized coding model are:

[0203] S211: Extract multi-band data from frequency-stratified data and synthesize feature vectors by weighted average method, using the formula:

[0204]

[0205] Apply weights to each frequency band data and add regularization terms to avoid overfitting to generate a weighted frequency band feature vector D i ;

[0206] Among them, S f (i) represents the signal strength of the f-th frequency band at the i-th sample point, w f is the weight of the frequency band, λ is the regularization parameter used to control the weight and;

[0207] S212: Adjust the parameters of the variational autoencoder to match the multi-band characteristics. The optimization algorithm uses the adjusted loss function using the formula:

[0208]

[0209] Add adjustment parameters to balance reconstruction loss and KL divergence and generate adjusted model parameters θ * ;

[0210] Among them, xi is the input data, z i is the latent variable, θ is the model parameter, and β is the loss adjustment factor, which is used to optimize the trade-off between the encoder and the decoder;

[0211] S213: Use the optimized model to reduce the reconstruction error and perform error evaluation. Apply the formula:

[0212]

[0213] Calculate the average reconstruction error of the model and refer to the continuity of the model output to generate the optimized coding model E;

[0214] Where N is the number of data points, x i is the actual data point, is the data point for model reconstruction, and γ is the continuity adjustment parameter used to evaluate the stability and continuity performance of the model.

[0215] formula:

[0216]

[0217] Parameter explanation:

[0218] S f (i): Signal strength of the f-th frequency band at the i-th sample point.

[0219] w f : Weight of the f-th frequency band.

[0220] λ: Regularization parameter used to control the weights and to avoid overfitting.

[0221] F: Total number of frequency bands.

[0222] i: sample index.

[0223] Calculation example:

[0224] Assume there are 3 frequency bands (F=3), the signal strength of each frequency band is S1(i)=2, S2(i)=3, S3(i)=5. The weights are w1=0.2, w2=0.3, w3=0.5 respectively. The regularization parameter λ=0.1.

[0225] Calculation process:

[0226] Calculate the weighted signal strength sum:

[0227]

[0228] Calculate the weighted sum squared:

[0229]

[0230] Apply the regularization parameter:

[0231] λ·1=0.1·1=0.1

[0232] D i value:

[0233] D i =3.8+0.1=3.9

[0234] Result analysis:

[0235] D i =3.9 represents the weighted frequency band feature vector value, reflecting the total influence of the weighted frequency band data, referring to the influence of overfitting.

[0236] formula:

[0237]

[0238] Parameter explanation:

[0239] x i : Input data.

[0240] z i : Latent variable.

[0241] θ: model parameters.

[0242] β: loss adjustment factor used to optimize the trade-off between encoder and decoder.

[0243] N: number of data points.

[0244] Calculation example:

[0245] Assume N = 100 data points, and x of each data point is i and z i All are randomly generated from a standard normal distribution. β = 0.5.

[0246] Calculation process:

[0247] Assume that the initial θ is 0.5.

[0248] Randomly generate x i and z i value.

[0249] Calculate logp θ (x i ∣z i ) and KL(q θ (z i ∣x i )∥p(z)).

[0250] Calculate the loss and adjust θ to minimize the loss function.

[0251] Result analysis:

[0252] The calculated θ * represents the optimized model parameters, aiming to minimize the combined loss of reconstruction error and KL divergence.

[0253] formula:

[0254]

[0255] Parameter explanation:

[0256] Data reconstructed by the model.

[0257] x prev,i : The output of the previous model.

[0258] γ: Smoothing coefficient, used to evaluate the continuity and improvement effect of the model.

[0259] N: number of data points.

[0260] Calculation example:

[0261] Assume N = 100, each x i and Randomly generated from a normal distribution, x prev,i This is the data generated by the previous model, γ = 0.1.

[0262] Calculation process:

[0263] Compute the reconstruction error for each data point.

[0264] Apply a smoothing factor γ to adjust the continuity.

[0265] Calculate the overall model error E.

[0266] Result analysis:

[0267] The calculated E value represents the overall reconstruction error of the model, including the continuity reference of the new model with the output of the previous model.

[0268] See also Figure 5 , the specific steps for obtaining the abnormal signal recognition results are:

[0269] S311: Use the optimized coding model to process the currently received microwave radio frequency signal, perform signal preprocessing, and apply the improved formula:

[0270] R i =α·Model(S i )+β·log(1+|S i |)

[0271] Generate model to reconstruct signal R i ;

[0272] Among them, S i is the original microwave RF signal sample, Model represents the optimized coding model, α and β adjust the sensitivity and dynamic range of signal processing;

[0273] S312: Comparison model reconstructed signal R i With the original signal S i The difference is applied, and the improved difference measurement formula is used, using the formula:

[0274]

[0275] Calculate the square root of the sum of squared weighted errors of the overall signal to generate a difference metric D;

[0276] Among them, R is the reconstructed signal output by the model, S i is the original signal, N is the number of signal samples, σ i is the standard deviation of the i-th signal sample;

[0277] S313: Using the difference metric D, the judgment standard of abnormal signals is set according to the size of the difference, and the threshold method is set, using the formula:

[0278]

[0279] Determine whether the signal is abnormal and obtain abnormal signal identification result C;

[0280] Where D is the signal difference measure, θ is the basic threshold of the difference measure, and δ is the adjustment factor for environmental noise or changes in system conditions.

[0281] formula:

[0282] R i =α·Model(S i )+β·log(1+|S i |)

[0283] Parameter details:

[0284] α and β are adjustment coefficients in the range of [0.1, 10], and their specific values ​​depend on the sensitivity of the model to the signal amplitude.

[0285] S i are the received signal samples.

[0286] R i is the output signal after model processing.

[0287] Derivation process:

[0288] Assumptions:

[0289] S i =5 is the value of the currently received signal sample.

[0290] α=2, β=1.

[0291] Calculation process:

[0292] Calculate log(1+|S i |)=log(1+5)=log(6)≈1.79.

[0293] Assume that the model output Model(S i )=3.

[0294] Then R i =2·3+1·1.79=6+1.79=7.79.

[0295] Analysis of calculation results:

[0296] R i =7.79 represents the signal output value after model processing and logarithmic adjustment.

[0297] formula:

[0298]

[0299] Parameter details:

[0300] R i and S i are the reconstructed signal and original signal output by the model respectively.

[0301] N is the number of signal samples.

[0302] Derivation process:

[0303] Assumptions:

[0304] R=[7.79,7.79], S=[5,5].

[0305] N=2.

[0306] Calculation process:

[0307] Calculate the difference (R i -S i ) 2 They are (7.79-5) 2 =7.8481 and (7.79-5) 2 =7.8481.

[0308] Then

[0309] Analysis of calculation results:

[0310] D = 3.96 represents a measure of the overall error between the model output and the original signal.

[0311] formula:

[0312]

[0313] Parameter details:

[0314] θ is the base threshold of the difference measure, assumed to be 4.

[0315] δ is the adjustment factor, assumed to be 0.1.

[0316] C indicates the signal status.

[0317] Derivation process:

[0318] The threshold value θ·(1+δ)=4·(1.1)=4.4 is calculated.

[0319] Compare D=3.96 and 4.4. Since 3.96<4.4, C=Normal.

[0320] Analysis of calculation results:

[0321] C=Normal means that the current signal is classified as normal according to the set threshold.

[0322] See also Figure 6 , the steps for obtaining the interference source feature set are as follows:

[0323] S411: Extract the abnormal signal features of each frequency band from the abnormal signal identification results, using the feature extraction formula:

[0324]

[0325] Calculate the weighted abnormal feature F of each frequency band ei ;

[0326] Where M is the number of samples in each frequency band, X im is the signal strength of the i-th frequency band at the m-th sample, σ im is the weight of the abnormal signal, λ is the nonlinear adjustment parameter of the signal strength, which enhances the sensitivity of the model to the signal characteristics;

[0327] S412: Based on weighted abnormal feature F ei , analyze and summarize the estimated types and sources of interference sources, and apply the analysis formula:

[0328]

[0329] Generate a summary score G for each interference source typei ;

[0330] Where E is the total number of interference source types, κ e is the classification weight of the interference source, μ is the feature balance adjustment coefficient;

[0331] S413: Summarize the score G for each interference source type i , by summarizing the formula:

[0332]

[0333] Summarize into interference source feature set H;

[0334] Among them, I is the number of reference interference source types, and δ is the score enhancement factor, which is used to emphasize the target interference source.

[0335] formula:

[0336]

[0337] Parameter explanation:

[0338] M: number of samples in each frequency band.

[0339] X im : Signal strength of the i-th frequency band at the m-th sample.

[0340] σ im : Abnormal signal strength factor, used to measure the abnormality of the signal.

[0341] λ: Nonlinear adjustment parameter of signal strength, used to adjust the influence of the square term of signal strength.

[0342] Derivation and examples:

[0343] Assume that a frequency has 3 samples (M=3), the signal strength X i is [2,3,5], the abnormal signal strength factor σ i is [0.5, 0.3, 0.9], and the nonlinear adjustment parameter λ = 0.1.

[0344] Calculation process:

[0345]

[0346] F ei =3.4

[0347] The value (3.4) represents the average abnormal signal characteristic strength of the frequency band.

[0348] formula:

[0349]

[0350] Parameter explanation:

[0351] E: Total number of interference source types.

[0352] F ei : Abnormal signal characteristics, obtained from step 1.

[0353] κ e : Classification weight of interference source.

[0354] μ: characteristic balance adjustment coefficient.

[0355] Derivation and examples:

[0356] Assume that there are two interference sources (E=2), the classification weight κ is [0.2, 0.5], and the adjustment coefficient μ=0.05.

[0357] Calculation process:

[0358] G i =(0.2·3.4 2 +0.05·3.4)+(0.5·3.4 2 +0.05 3.4

[0359] G i =(0.2·11.56+0.17)+(0.5·11.56+0.17)

[0360] G i =(2.312+0.17)+(5.78+0.17)

[0361] G i =2.482+5.95

[0362] G i =8.432

[0363] The value (8.432) represents the score of the combined impact of the two interference sources on the frequency band.

[0364] formula:

[0365]

[0366] Parameter explanation:

[0367] G i : The summary score of each interference source obtained from step 2.

[0368] δ: score enhancement factor.

[0369] I: The number of reference interference source types.

[0370] Derivation and examples:

[0371] Assume δ=0.1.

[0372] Calculation process:

[0373] H=max{8.432+0.1·8.432 2}

[0374] H=8.432+0.1·71.099

[0375] H=8.432+7.11

[0376] H=15.542

[0377] The value (15.542) represents the score of the interferer feature set.

[0378] A multi-band microwave radio frequency test system, the multi-band microwave radio frequency test system is used to perform the multi-band microwave radio frequency test method, the system comprises:

[0379] After receiving the multi-band microwave RF signal, the signal receiving module applies frequency analysis to the signal, subdividing it into multiple differentiated frequency bands, collecting the amplitude and frequency data of each frequency band, and recording the signal strength and time series to generate frequency-layered data;

[0380] The data training module uses frequency-stratified data as the basis, selects multi-band data for variational autoencoder training, adjusts encoder parameters to match multi-band characteristics, performs model optimization to reduce reconstruction errors, and obtains an optimized encoding model;

[0381] The abnormality recognition module uses the optimized coding model to process the currently received microwave radio frequency signal, calculates the difference between the reconstructed signal and the original signal, sets the judgment standard of the abnormal signal, identifies and calibrates the abnormality, and obtains the abnormal signal recognition result;

[0382] The feature analysis module extracts the abnormal signal features of each frequency band from the abnormal signal identification results, analyzes multiple features to estimate the type and source of the interference source, and forms preliminary interference source features;

[0383] The interference source summary module analyzes the preliminary interference source characteristics, summarizes and improves the interference source information, enhances the integration and practicality of the data, and generates an interference source feature set.

[0384] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A multi-band microwave radio frequency testing method, characterized in that: The following steps are involved: Receive multi-band microwave radio frequency signals, apply frequency analysis technology to subdivide the signals into differentiated frequency bands, collect amplitude and frequency data for each frequency band, record the intensity and time series of multi-band signals, and generate frequency-stratified data; Using the multiple frequency band data obtained from the frequency layered data to train a variational autoencoder, adjusting encoder parameters, matching multi-frequency band characteristics, optimizing the model to reduce reconstruction errors, and obtaining an optimized encoding model; The currently received microwave radio frequency signal is processed by the optimized coding model, the difference between the model reconstructed signal and the original signal is compared, and a judgment standard for abnormal signals is set according to the size of the difference to obtain an abnormal signal recognition result; The abnormal signal features of each frequency band are extracted from the abnormal signal identification results, the estimated types and sources of interference sources are analyzed and summarized, and an interference source feature set is generated in summary.

2. The multi-band microwave radio frequency testing method according to claim 1, characterized in that: The subdivision steps of the differentiated frequency bands are specifically as follows: Receive a multi-band microwave RF signal from a differentiated source and apply an improved frequency band separation algorithm to subdivide the signal into differentiated frequency bands using the formula: Calculate the initial amplitude and frequency of the multi-band signal to generate a separated signal sequence Z i ; Among them, x n represents the nth sample of the original signal, f i represents the center frequency of the jth frequency band, N represents the total number of samples, and j is an imaginary unit; The isolated signal sequence S i Perform amplitude analysis and use the composite amplitude calculation formula: Calculate the amplitude of multiple frequency bands and generate multi-frequency band amplitude data A i ; in, and Represent the real and imaginary parts of the complex number respectively, and k is the weight factor used to adjust the influence of the logarithmic term; The dynamically adjusted amplitude threshold is set according to the multi-band amplitude data Ai, using the formula: Matching the non-static characteristics of the signal to generate differentiated frequency bands P i ; Among them, T a is the basic amplitude threshold, α and ω are the adjustment amplitude and frequency parameters of the amplitude threshold, respectively.

3. The multi-band microwave radio frequency testing method according to claim 2, characterized in that: The steps for obtaining the frequency layered data are specifically as follows: Based on the differentiated frequency band P i , the integral transform method is used to analyze each frequency band and calculate the amplitude and frequency data using the formula: Calculate the frequency components and generate the frequency band data set F i ; Where T is the sampling period, t is the time variable, f is the frequency variable, and j represents the imaginary unit; For the frequency band dataset F i Perform time series analysis and use sine wave superposition and time delay analysis to extract patterns in the time series using the formula: Thus, the time series model Ri is established; Among them, F i is the frequency band data set, ω is the fundamental frequency, φ is the phase angle, and T is the time delay variable to represent the change of the frequency band over time; Using the time series model Ri, by calculating the peak-to-valley difference, the formula is: S i =max t∈T (R i (t))-min t∈T (R i (t)) Evaluate the stability and reliability of the signal and generate frequency-stratified data S i ; Among them, R is the time series model, T represents the analysis period, and t is the time variable.

4. The multi-band microwave radio frequency testing method according to claim 3, characterized in that: The steps for obtaining the optimized coding model are specifically as follows: Extract multi-band data from frequency-stratified data and synthesize feature vectors by weighted average method using the formula: Apply weights to each frequency band data and add regularization terms to avoid overfitting to generate a weighted frequency band feature vector D i ; Among them, S f (i) represents the signal strength of the f-th frequency band at the i-th sample point, w f is the weight of the frequency band, λ is the regularization parameter used to control the weight and; The parameters of the variational autoencoder are adjusted to match the multi-band characteristics, and the optimization algorithm uses the adjusted loss function using the formula: Add adjustment parameters to balance the reconstruction loss and KL divergence and generate the adjusted model parameters θ*; Among them, x i is the input data, z i is the latent variable, θ is the model parameter, and β is the loss adjustment factor, which is used to optimize the trade-off between the encoder and the decoder; Use the optimized model to reduce the reconstruction error, perform error evaluation, and apply the formula: Calculate the average reconstruction error of the model and refer to the continuity of the model output to generate the optimized coding model E; Where N is the number of data points, x i is the actual data point, is the data point for model reconstruction, and γ is the continuity adjustment parameter used to evaluate the stability and continuity performance of the model.

5. The multi-band microwave radio frequency testing method according to claim 4, characterized in that: The steps for obtaining the abnormal signal recognition result are specifically as follows: Use the optimized coding model to process the currently received microwave radio frequency signal, perform signal preprocessing, and apply the improved formula: R i =α·Model(S i )+β·log(1+|S i |) Generate model to reconstruct signal R i ; Among them, S i is the original microwave RF signal sample, Model represents the optimized coding model, α and β adjust the sensitivity and dynamic range of signal processing; Compared with the model reconstructed signal R i With the original signal S i The difference is applied, and the improved difference measurement formula is used, using the formula: Calculate the square root of the sum of squared weighted errors of the overall signal to generate a difference metric D; Among them, R is the reconstructed signal output by the model, S i is the original signal, N is the number of signal samples, σ i is the standard deviation of the i-th signal sample; The difference metric D is used to set the abnormal signal discrimination standard according to the size of the difference, and the threshold method is set using the formula: Determine whether the signal is abnormal and obtain abnormal signal identification result C; Where D is the signal difference measure, θ is the basic threshold of the difference measure, and δ is the adjustment factor for environmental noise or changes in system conditions.

6. The multi-band microwave radio frequency testing method according to claim 5, characterized in that: The steps of obtaining the interference source feature set are specifically as follows: The abnormal signal features of each frequency band are extracted from the abnormal signal recognition results using the feature extraction formula: Calculate the weighted abnormal feature F of each frequency band ei ; Where M is the number of samples in each frequency band, X im is the signal strength of the i-th frequency band at the m-th sample, σ im is the weight of the abnormal signal, λ is the nonlinear adjustment parameter of the signal strength, which enhances the sensitivity of the model to the signal characteristics; Based on the weighted abnormal feature F ei , analyze and summarize the estimated types and sources of interference sources, and apply the analysis formula: Generate a summary score G for each interference source type i ; Where E is the total number of interference source types, κ e is the classification weight of the interference source, μ is the feature balance adjustment coefficient; The summary score G of each interference source type is i , by summarizing the formula: Summarize into interference source feature set H; Among them, I is the number of reference interference source types, and 6 is the score enhancement factor used to emphasize the target interference source.

7. A multi-band microwave radio frequency test system, characterized in that: According to the multi-band microwave radio frequency testing method according to any one of claims 1 to 6, the system comprises: After receiving the multi-band microwave RF signal, the signal receiving module applies frequency analysis to the signal, subdividing it into multiple differentiated frequency bands, collecting the amplitude and frequency data of each frequency band, and recording the signal strength and time series to generate frequency-layered data; The data training module selects multi-band data to train the variational autoencoder based on the frequency layered data, adjusts the encoder parameters to match the multi-band characteristics, performs model optimization to reduce the reconstruction error, and obtains an optimized encoding model; The abnormality recognition module uses the optimized coding model to process the currently received microwave radio frequency signal, calculates the difference between the reconstructed signal and the original signal, sets the judgment standard of the abnormal signal, recognizes and calibrates the abnormality, and obtains the abnormal signal recognition result; The feature analysis module extracts the abnormal signal features of each frequency band from the abnormal signal identification results, analyzes multiple features to estimate the type and source of the interference source, and forms preliminary interference source features; The interference source summary module analyzes the preliminary interference source characteristics, summarizes and improves the interference source information, enhances the integration and practicability of the data, and generates an interference source feature set.