Moving and stationary target separation method applied to automobile FMCW radar system

Abstract

The invention relates to a moving and static target separation method applied to an automobile FMCW radar system, and the method comprises the steps: S1, obtaining a first beat frequency signal of a radar based on a radar signal received by an antenna through employing a frequency mixing technology, and carrying out the histogram detection and median filtering of the first beat frequency signal, and obtaining a second beat frequency signal; and S2, training a first RLS adaptive filter by using the first beat frequency signal and the second beat frequency signal, and reconstructing the first beat frequency signal based on the trained first RLS adaptive filter to obtain a third beat frequency signal. S3, averaging all the received pulses to obtain a reference received pulse, and subtracting the reference pulse from each received pulse to realize coarse separation of the moving target signal and the static target signal; and S4, respectively training two adaptive filters by using the main input signal subjected to anti-interference processing, and realizing separation of the moving target and the static target. Compared with the prior art, the method has the advantages of accurately identifying and separating moving and static targets in a complex environment and the like.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and particularly to a method for separating moving and stationary targets applied to an automotive FMCW radar system. Background Art

[0002] With the development of autonomous driving technology, higher requirements are put forward for the ability of radar systems to accurately identify and separate moving and stationary targets in complex environments. In complex environments, how to accurately distinguish moving targets and stationary backgrounds from radar echoes has become one of the keys to improving the target detection and tracking capabilities of radar systems.

[0003] For example, Chinese Patent CN105137423A discloses a method for real-time detection and separation of multiple moving targets by a through-wall radar. By using a clutter suppression algorithm based on an exponential smoothing filter, complex background signals are removed, and the information of moving human targets is retained more quickly and accurately; by using an envelope detection algorithm based on Hilbert transform, each moving target is placed at the peak of a different envelope; by using an algorithm that combines a fixed threshold and an adaptive threshold, the separation of each moving target in different regions is achieved more quickly.

[0004] However, such existing technologies have certain limitations in dealing with multiple targets and interference signals. Most methods only strengthen moving targets while ignoring the influence of the surrounding environment and stationary target signals, and there is no method for directly separating moving targets from stationary targets without considering the surrounding environment information.

[0005] Therefore, there is an urgent need for a separation method that can separate and independently display moving targets and stationary targets while retaining the information of radar beat signals, thereby improving the target detection ability of radar systems. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for separating moving and stationary targets applied to an automotive FMCW radar system.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for separating moving and stationary targets applied to an automotive FMCW radar system, comprising:

[0009] Step S1: Based on the radar signals received by the antenna, using mixing technology to obtain the first beat signal of the radar, and performing histogram detection and median filtering on the first beat signal to obtain the second beat signal;

[0010] Step S2: Train the first RLS adaptive filter using the first beat frequency signal and the second beat frequency signal, and reconstruct the first beat frequency signal based on the trained first RLS adaptive filter to obtain a third beat frequency signal.

[0011] Step S3: Utilize the principle that the distance between the stationary target and the radar antenna is constant and the phase is invariant. By averaging all received pulses, obtain a reference received pulse, and subtract the reference pulse from each received pulse to achieve a rough separation of the moving target signal and the stationary target signal.

[0012] Step S4: Train the second RLS adaptive filter and the third RLS adaptive filter respectively using the main input signal after anti-interference processing, and achieve the separation of moving and stationary targets.

[0013] The specific steps of step S1 include:

[0014] Step S1-1: Set at least one moving target and at least one stationary target, configure the positions and speeds of each target, and configure interference signals for at least some of the targets.

[0015] Step S1-2: Based on the transmitted signal of the self-radar and all interference signals, combined with the known time delay and Doppler frequency shift, obtain the received radar signal, and convolve the transmitted signal of the self-radar and the received radar signal to obtain the first beat frequency signal:

[0016]

[0017] where: s IF (t s , t f ) is the first beat frequency signal, L is the total number of transmitted chirp signals, t s is the full time of the radar, T p is the pulse repetition time, P is the number of moving targets, α t,p is the echo amplitude of the p-th moving target, f r,m is the frequency value of the moving target, t f is the fast time of the radar, T c is the duration of the chirp signal, f D,m is the Doppler frequency of the moving target, θ r,m is the arrival angle of the p-th moving target, α t,q is the echo amplitude of the q-th stationary target, f r,s is the frequency value of the stationary target, θ r,s is the arrival angle of the q-th stationary target, α t,i is the amplitude of the interference, f i is the frequency offset caused by the interference signal.

[0018] Step S1-3: Select the first frame of the first beat frequency signal, extract its amplitude information, construct a histogram, calculate the cumulative distribution function of the amplitude through the constructed histogram, determine the noise detection threshold according to the cumulative distribution function of the amplitude, and use this noise detection threshold to process the first beat frequency signal, and regard the part with an amplitude greater than the noise detection threshold as the potential interference area;

[0019] Step S1-4: Perform median filtering to obtain the interference area, and regard the remaining part as the second beat frequency signal.

[0020] In the step S1-3, the noise detection threshold is the amplitude at which the cumulative distribution reaches the first set ratio.

[0021] The median filtering uses a median filter of [1, N m , where N m is selected by calculating the average interference duration through pre-detection.

[0022] The step S2 includes:

[0023] Step S2-1: Use the second beat frequency signal as the reference input signal and the first beat frequency signal as the signal to train the first RLS adaptive filter;

[0024] Step S2-2: Extract the filtering weights of the first RLS adaptive filter after training, convolve them with the first beat frequency signal, and reconstruct to obtain the third beat frequency signal, where the third beat frequency signal is the clean radar beat frequency signal after suppressing interference.

[0025] The step S3 includes:

[0026] Step S3-1: Perform Fourier transform on the third beat frequency signal to obtain the carrier-to-interference ratio of the radar signal as the first carrier-to-interference ratio, and calculate the average value of all received pulses to obtain the carrier-to-interference ratio of the reference received pulse as the second carrier-to-interference ratio;

[0027] Step S3-2: Subtract the second carrier-to-interference ratio from the first carrier-to-interference ratio to separately obtain the rough separation carrier-to-interference ratio of the moving target echo signal;

[0028] Step S3-3: Subtract the rough separation carrier-to-interference ratio of the moving target echo signal from the first carrier-to-interference ratio to obtain the rough separation carrier-to-interference ratio of the stationary target echo signal, and perform inverse Fourier transform on the rough separation carrier-to-interference ratio of the moving target echo signal and the rough separation carrier-to-interference ratio of the stationary target echo signal respectively to obtain the roughly separated moving target signal and stationary target signal.

[0029] The step S4 includes:

[0030] Step S4-1: Use the roughly separated moving target signal as the reference signal of the second RLS adaptive filter, and use the third beat frequency signal as the main input signal of the second RLS adaptive filter to complete the training of the second RLS adaptive filter.

[0031] Use the roughly separated stationary target signal as the reference signal of the third adaptive filter, and use the third beat frequency signal as the main input signal of the third RLS adaptive filter to complete the training of the third RLS adaptive filter.

[0032] Step S4-2: Use the trained second RLS adaptive filter to obtain the deeply separated moving target signal from the third beat frequency signal, and use the trained third RLS adaptive filter to obtain the deeply separated stationary target signal from the third beat frequency signal.

[0033] The self-radar is a vehicle-mounted radar.

[0034] A charging pile plug-and-charge function test device includes a memory, a processor, and a program stored in the memory. When the processor executes the program, the above-mentioned method is implemented.

[0035] A storage medium stores a program, and when the program is executed, the above-mentioned method is implemented.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. By adopting the recursive least squares method adaptive filter, moving and stationary targets are effectively separated, and at the same time, interference signals from various sources are suppressed. This method solves the problem of accurately identifying and separating moving and stationary targets in a complex environment, improves the accuracy of target detection, and at the same time maintains a low computational complexity.

[0038] 2. Utilize the Doppler frequency difference to separate targets with different motion states, and iteratively optimize the filter weights until convergence, thereby improving the accuracy of target detection and dynamic / static separation. Using the weight extraction method allows the radar signal to be directly processed using the extracted weights, which not only reduces the computational complexity but also retains the inherent complexity characteristics of the signal, thus providing a more accurate representation of the target and the surrounding environment. This method shows good adaptability and stability in practical applications, can effectively reduce interference, and reduce the need for frequent weight updates. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the main step flow of the method of the present invention;

[0040] Figure 2Schematic diagram of the fast and slow time domains of the original radar beat frequency signal in a multi-interference scenario for an embodiment;

[0041] Figure 3 Schematic diagram of range-Doppler of the original radar beat frequency signal in a multi-interference scenario for an embodiment;

[0042] Figure 4 Schematic diagram of the detection result of the interference paragraph for an embodiment;

[0043] Figure 5 Comparison result of range-Doppler diagrams of the original radar beat frequency signal after interference suppression in a multi-interference scenario for an embodiment;

[0044] Figure 6 Schematic diagram of range-Doppler diagrams for roughly separating moving target signals and stationary target signals for an embodiment;

[0045] Figure 7 Schematic diagram of the training process of the RLS adaptive filter with the moving target signal as the reference input signal for an embodiment;

[0046] Figure 8 Separation results of moving targets and stationary targets for an embodiment. Detailed implementation manners

[0047] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.

[0048] A method for separating moving and stationary targets applied to an automotive FMCW radar system, which regards stationary targets, dynamic targets, and interference as three independent subspaces respectively, and uses the characteristic that the RLS adaptive filter can learn the signal characteristics of the reference input to select a suitable reference input signal, trains different RLS adaptive filters, enables them to distinguish stationary targets, dynamic targets, and interference signals, and outputs signals with the same characteristics as the reference input signal, thereby completing the separation of signals in different subspaces.

[0049] As Figure 1 shown, it includes:

[0050] Step S1: Based on the radar signal received by the antenna, use the mixing technology to obtain the first beat frequency signal of the radar, and perform histogram detection and median filtering on the first beat frequency signal to obtain the second beat frequency signal, specifically including:

[0051] Step S1-1: Set no less than one moving target and no less than one stationary target, configure the positions and speeds of each target, and configure interference signals for at least some of the targets;

[0052] This process is specifically implemented in MATLAB simulation. It is necessary to configure all radar signals, including the interference signals emitted by moving targets, the interference signals emitted by stationary targets, and the transmitted signals of the self-radar. Among them, for the transmitted signal of the self-radar, the waveform of the transmitted signal of the self-radar is a chirp, also known as a chirp signal. Assume that the radar transmits N chirp signals (slow time), and there are M sampling points (fast time) on each chirp signal. Specifically:

[0053]

[0054] Where: s(t s ,t f ) is the transmitted signal of the self-radar, and the discrete-time instance is the radar slow time t s = lT p for l ∈ [0, L-1], where L is the total number of transmitted chirp signals, and T p is called the "pulse repetition time (PRT)", which is defined as the time interval between the starts of two consecutive chirp signals. δ(·) represents the Dirac δ function, and t f ∈ [0, T chirp represents the discrete-time instance, and T chirp is the chirp duration, and Where f s is the sampling rate in the fast-time domain, n ∈ [0, N-1] is the sample index, and N = T chirp * f s . α t represents the constant amplitude of the chirp signal, f c is the carrier frequency, represents the frequency modulation (FM) slope of the chirp signal, where BW sw represents the total sweep bandwidth of the chirp signal.

[0055] Figure 2 shows the fast-time and slow-time domain diagrams of the original radar beat signal in a multi-interference scenario generated by MATLAB simulation;

[0056] Step S1-2: Based on the transmitted signal of the self-radar and all interference signals, combined with the known time delay and Doppler frequency shift, obtain the received radar signal, and convolve the transmitted signal of the self-radar and the received radar signal to obtain the first beat signal:

[0057]

[0058] Where: There are K radar antennas, P moving targets, and Q stationary targets in the sensor's field of view. Each target exhibits a single scattered echo. In addition, the receiving antennas are separated by half a wavelength, and sIF (t s ,t f 0 is the first beat frequency signal, L is the total number of transmitted chirp signals, t s is the radar full time, T p is the pulse repetition time, P is the number of moving targets, α t,p is the echo amplitude of the p-th moving target, f r,m is the frequency value of the moving target, t f is the radar fast time, T c is the duration of the chirp signal, f D,m is the Doppler frequency of the moving target, θ r,m is the arrival angle of the p-th moving target, α t,q is the echo amplitude of the q-th stationary target, f r,s is the frequency value of the stationary target, θ r,s is the arrival angle of the q-th stationary target, α t,i is the amplitude of the interference, f i is the frequency offset caused by the interference signal;

[0059] Figure 3 shows the range-Doppler map of the original radar beat frequency signal in a multi-interference scenario. The first beat frequency signal consists of three parts: stationary target signal, moving target signal, and interference signal. Here, the interference signal includes not only the noise in the environment but also the detection signals emitted by some target vehicles. These interference signals will affect the ability of the radar system to detect targets;

[0060] Step S1-3: Select the first frame of the first beat frequency signal. Based on the characteristics that the in-time domain distribution time of the co-frequency interference is limited and the amplitude is relatively high, and the overall amplitude distribution of the target signal is relatively stable, extract its amplitude information, construct a histogram, calculate the cumulative distribution function of the amplitude through the constructed histogram, determine the noise detection threshold according to the cumulative distribution function of the amplitude, and use this noise detection threshold to process the first beat frequency signal. The part with an amplitude greater than the noise detection threshold is used as the potential interference region. Specifically, because there is an absolute value difference in amplitude between the interference signal and the target signal on the histogram, the potential interference region can be identified;

[0061] Among them, the noise detection threshold is the amplitude at which the cumulative distribution reaches the first set ratio.

[0062] Step S1-4: After the histogram detection, in order to filter out the salt-and-pepper noise in the signal, median filtering is then used. Considering that the interference is generally distributed in a relatively short fast time, a median filter of [1, N m can be adopted, where N mThe value can be selected by calculating the average interference duration through pre-detection. Generally, the interference duration is greater than 7 fast time units, where N m takes 11. Figure 4 Shows the results of histogram detection and CDF calculation, as well as the detection results of the interference section after median filtering. The yellow part represents the section where the interference signal exists, and the blue part represents the clean signal.

[0063] Step S2: Train the first RLS adaptive filter using the first beat frequency signal and the second beat frequency signal, and reconstruct the first beat frequency signal based on the trained first RLS adaptive filter to obtain the third beat frequency signal. Since it is difficult to directly separate the clean signal from the beat frequency signal obtained by the radar, and the interference signal only affects the echo signal within a specific time, the interference signal section only exists within part of the time. In order to obtain a cleaner signal closer to the actual situation, we plan to perform signal reconstruction on the interference section, optimize the filter parameters to enhance the target signal and further suppress the interference, including:

[0064] Step S2-1: Use the second beat frequency signal as the reference input signal and the first beat frequency signal as the signal to train the first RLS adaptive filter. The purpose here is to retain the time continuity of the radar beat frequency signal, so that the RLS adaptive filter learns the time continuous variation of the signal, rather than just focusing on one time frame;

[0065] Among them, the training of the RLS filter is to find the optimal filter parameters by minimizing the following cost function, and its calculation formula is as follows:

[0066]

[0067] Among them, e(i) is the error of the i-th iteration, defined as:

[0068] e(i) = d(i) - w T (i)x(i)

[0069] Among them, d(i) is the reference input signal, w(i) is the filter coefficient vector, x(i) is the main input signal, and λ is the forgetting factor, satisfying 0 < λ ≤ 1.

[0070] The core idea of the RLS algorithm is to recursively update the filter coefficient vector w(n) and the error covariance matrix P(n), thereby effectively reducing the computational complexity and realizing adaptive filtering at the same time. The specific steps are as follows:

[0071] Step 1: Initialization. At the beginning of the algorithm, it is necessary to initialize the filter coefficients and the error covariance matrix, and initialize the filter coefficients w(0) and the inverse correlation matrix P(0):

[0072] w(0) = 0

[0073] P(0) = δ -1 I

[0074] Where δ is a small positive number, usually used to avoid division by zero errors, and I is the identity matrix.

[0075] Step 2: Gain vector calculation. In each iteration, first calculate the gain vector k(n), which determines the sensitivity of the newly estimated filter coefficients to the current error:

[0076]

[0077] Step 3: Filter coefficient update, that is, update the filter coefficient vector w(n):

[0078] w(n) = w(n - 1) + k(n)e(n)

[0079] Where e(n) = d(n) - w T (n - 1)x(n).

[0080] Step 4: Inverse correlation matrix update, that is, update the inverse correlation matrix P(n):

[0081] P(n) = λ -1 (P(n - 1) - k(n)x T (n)P(n - 1))

[0082] By iterating these steps, the RLS algorithm recursively updates the filter coefficients w(n) and the inverse correlation matrix P(n), reducing the computational complexity while maintaining accurate adaptation to the input signal.

[0083] Step S2 - 2: Extract the filtering weights of the first RLS adaptive filter after training is completed, convolve them with the first beat frequency signal, and reconstruct to obtain the third beat frequency signal. Among them, the third beat frequency signal is the clean radar beat frequency signal after interference suppression. Figure 5 Shows the comparison results of the range - Doppler maps of the original radar beat frequency signal under multi - interference scenarios after interference suppression. It can be seen that after interference suppression and reconstruction with filtering weight extraction, the interference signal is significantly suppressed and the target is more prominently highlighted;

[0084] Step S3: Utilize the principle that the distance between the stationary target and the radar antenna is constant and the phase is invariant. By averaging all received pulses, obtain the reference received pulse, and subtract the reference pulse from each received pulse to achieve a rough separation of the moving target signal and the stationary target signal, including:

[0085] Step S3-1: Perform Fourier transform on the third beat frequency signal to obtain the carrier-to-interference ratio of the radar signal as the first carrier-to-interference ratio, and calculate the average value of all received pulses to obtain the carrier-to-interference ratio of the reference received pulse as the second carrier-to-interference ratio:

[0086]

[0087] where: m is the sampling point in the fast time dimension (range dimension), i is the sampling point in the slow time dimension (velocity dimension), and N is the total number of sampling points in the fast time dimension. Figure 6 The range-Doppler map of the roughly separated moving target signal and stationary target signal is shown in

[0088] Step S3-2: Subtract the second carrier-to-interference ratio from the first carrier-to-interference ratio to separately obtain the roughly separated carrier-to-interference ratio of the moving target echo signal.

[0089] Step S3-3: Subtract the roughly separated carrier-to-interference ratio of the moving target echo signal from the first carrier-to-interference ratio to obtain the roughly separated carrier-to-interference ratio of the stationary target echo signal. Perform inverse Fourier transform on the roughly separated carrier-to-interference ratio of the moving target echo signal and the roughly separated carrier-to-interference ratio of the stationary target echo signal respectively to obtain the roughly separated moving target signal and stationary target signal.

[0090] Step S4: Use the main input signal after anti-interference processing to train the second RLS adaptive filter and the third RLS adaptive filter respectively, and achieve the separation of moving and stationary targets, including:

[0091] Step S4-1: Use the roughly separated moving target signal as the reference signal of the second RLS adaptive filter, and use the third beat frequency signal as the main input signal of the second RLS adaptive filter to complete the training of the second RLS adaptive filter.

[0092] Use the roughly separated stationary target signal as the reference signal of the third adaptive filter, and use the third beat frequency signal as the main input signal of the third RLS adaptive filter to complete the training of the third RLS adaptive filter.

[0093] Step S4-2: Use the trained second RLS adaptive filter to obtain the deeply separated moving target signal from the third beat frequency signal, and use the trained third RLS adaptive filter to obtain the deeply separated stationary target signal from the third beat frequency signal.

[0094] In addition, in this embodiment, the self-radar is a vehicle-mounted radar.

[0095] When all the estimation processes are completed, the algorithm outputs the separation results of the input radar signal into interference signal, moving target signal, and stationary target signal, as well as the filtering weights of the corresponding RLS adaptive filter, so as to achieve the purpose of separating moving and stationary targets with anti-interference in continuous time. The separation results of the moving targets and stationary targets obtained by the algorithm are as Figure 8 shown.

[0096] Analysis of interference suppression performance and separation accuracy of moving and stationary targets. For the measurement of interference suppression performance and separation accuracy of moving and stationary targets, we evaluated by selecting the powers of several landmark points in different states on the range-Doppler spectrogram, and focused on observing the power changes of the marked points under four different operating nodes. The results are as Figure 1 shown. It can be seen from the result comparison that after the anti-interference operation, the power of the interference marked points has been significantly suppressed. It can be observed from the image that the overall power of the interference signal segment has been significantly suppressed. Taking the case of separating moving targets after the moving and stationary separation operation as an example, it can be found that the marked points of stationary targets have been significantly suppressed. Even without the anti-interference operation and only using the filtering weights of moving targets to operate on the original radar beat frequency signal, effective separation of moving and stationary targets can still be achieved.

[0097] Table 1

[0098] Sample point (a) (b) (c) (d) A (moving target) 84.0143 83.7593 82.1234 82.3778 B (moving target) 83.7221 83.3101 81.0318 81.4363 C (moving target) 83.5735 83.1768 81.4867 81.8761 D (static target) 83.4333 83.2219 50.7967 50.8483 E (interference) 62.9046 60.1367 55.6754 58.7915 F (interference) 59.5044 44.9263 44.1103 53.1017

[0099] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs and other various media that can store program codes.

Claims

1. A method for separating moving and stationary targets in an automotive FMCW radar system, characterized in that: include: Step S1: Based on the radar signal received by the antenna, a first beat frequency signal of the radar is obtained by using a frequency mixing technique, and the first beat frequency signal is subjected to histogram detection and median filtering to obtain a second beat frequency signal; Step S2: training the first RLS adaptive filter by using the first beat frequency signal and the second beat frequency signal, and reconstructing the first beat frequency signal based on the trained first RLS adaptive filter to obtain a third beat frequency signal; Step S3: utilizing the principle that the distance between the stationary target and the radar antenna is constant and the phase is unchanged, by averaging all received pulses to obtain a reference received pulse, and subtracting the reference pulse from each received pulse, thereby achieving a rough separation of the moving target signal and the stationary target signal; Step S4: using the main input signal after anti-interference processing to train the second RLS adaptive filter and the third RLS adaptive filter respectively, and realize the separation of moving and stationary targets.

2. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 1, characterized in that: The step S1 specifically includes: Step S1-1: setting at least one moving target and at least one stationary target and configuring the position and speed of each target, and configuring interference signals for at least some of the targets; Step S1-2: Based on the transmission signal of the self radar and all interference signals, combined with the known time delay and Doppler frequency shift, the received radar signal is obtained, and the transmission signal of the self radar and the received radar signal are convolved to obtain the first beat frequency signal: Where: s IF (t s ,t f ) is the first beat frequency signal, L is the total number of transmitted chirp signals, t s is the radar full time, T p is the pulse repetition time, P is the number of moving targets, α t,p is the echo amplitude of the pth moving target, f r,m is the frequency value of the moving target, t f is the radar fast time, T c is the duration of the chirp signal, f D,m is the Doppler frequency of the moving target, θ r,m is the arrival angle of the pth moving target, α t,q is the echo amplitude of the qth stationary target, f r,s is the frequency value of the stationary target, θ r,s is the arrival angle of the qth stationary target, α t,i is the amplitude of the interference, f i The frequency deviation caused by the interference signal; Step S1-3: Select the first frame of the first beat frequency signal, extract its amplitude information, and construct a histogram, calculate the cumulative distribution function of the amplitude through the constructed histogram, determine the noise detection threshold according to the cumulative distribution function of the amplitude, and use the noise detection threshold to process the first beat frequency signal, and take the part with amplitude greater than the noise detection threshold as a potential interference area; Step S1-4: Perform median filtering to obtain the interference area, and use the remaining part as the second beat frequency signal.

3. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 2, characterized in that: In step S1-3, the noise detection threshold is the amplitude at which the cumulative distribution reaches a first set ratio.

4. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 2, characterized in that: The median filter adopts [1,N m ] median filter, where N m The value is selected by calculating the average duration of interference through pre-detection.

5. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 2, characterized in that: The step S2 comprises: Step S2-1: using the second beat frequency signal as a reference input signal and the first beat frequency signal as a signal to train the first RLS adaptive filter; Step S2-2: extract the filter weight of the first RLS adaptive filter after training, and convolve it with the first beat frequency signal to reconstruct a third beat frequency signal, wherein the third beat frequency signal is a clean radar beat frequency signal after interference suppression.

6. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 1, characterized in that: The step S3 comprises: Step S3-1: Performing Fourier transform on the third beat frequency signal to obtain the carrier-to-noise ratio of the radar signal as the first carrier-to-noise ratio, and calculating the average value of all received pulses to obtain the carrier-to-noise ratio of the reference received pulse as the second carrier-to-noise ratio; Step S3-2: subtracting the second carrier-to-noise ratio from the first carrier-to-noise ratio to obtain a coarsely separated carrier-to-noise ratio of the moving target echo signal; Step S3-3: subtract the coarse separation carrier-to-noise ratio of the moving target echo signal from the first carrier-to-noise ratio to obtain the coarse separation carrier-to-noise ratio of the stationary target echo signal, and perform inverse Fourier transform on the coarse separation carrier-to-noise ratio of the moving target echo signal and the coarse separation carrier-to-noise ratio of the stationary target echo signal to obtain the coarsely separated moving target signal and stationary target signal respectively.

7. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 6, characterized in that: The step S4 comprises: Step S4-1: using the roughly separated moving target signal as the reference signal of the second RLS adaptive filter, and using the third beat frequency signal as the main input signal of the second RLS adaptive filter, to complete the training of the second RLS adaptive filter. The roughly separated stationary target signal is used as a reference signal of the third adaptive filter, and the third beat frequency signal is used as a main input signal of the third RLS adaptive filter to complete the training of the third RLS adaptive filter; Step S4-2: using the trained second RLS adaptive filter to obtain a moving target signal after depth separation from the third beat frequency signal, and using the trained third RLS adaptive filter to obtain a stationary target signal after depth separation from the third beat frequency signal.

8. The method for separating moving and stationary targets in an automotive FMCW radar system according to claim 2, characterized in that: The self radar is a vehicle-mounted radar.

9. A moving and stationary target separation device applied to an automobile FMCW radar system, comprising a memory, a processor, and a program stored in the memory, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

10. A storage medium having a program stored thereon, characterized in that: When the program is executed, the method according to any one of claims 1 to 8 is implemented.

Images