Object recognition method for radar system
Through signal processing and error elimination of radar system, accurate identification of small aircraft and natural organisms is achieved, solving the problems of poor recognition capabilities and low accuracy of existing radar systems when identifying small aircraft, and improving the recognition accuracy and accuracy sex.
Patent Information
- Application Number
- CN202210327982.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-29
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-03-29
AI Technical Summary
The existing radar systems have problems with poor recognition capabilities, low accuracy and large errors when identifying small aircraft and natural organisms, and cannot effectively distinguish between birds and aircraft, resulting in omissions and false alarms.
By upconverting and mixing signal suppression of the initial signal of the radar system, the echo signal is obtained and Doppler information is extracted, the signal curve and frequency are judged to distinguish between aircraft and non-aircraft objects, and through Dechirp processing and two-dimensional Fourier Transforms to obtain the final signal, eliminate signal errors, calculate the blade length and position, and establish a tracking model to improve recognition accuracy.
It realizes accurate identification and distinction between small aircraft and natural organisms, improves the accuracy of the radar system, reduces errors, and can identify and output detailed characteristics of the aircraft, making up for the shortcomings of traditional radars in small aircraft recognition.
Smart Images

Figure CN114660565B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of radar systems, and in particular to an object recognition method for a radar system. Background Art
[0002] With the rapid development of science and technology in recent years, the unmanned aerial vehicle industry is undergoing an epoch-making change driven by the entire industry. The advantages of "low, small and slow" aircraft also provide great convenience for criminals to carry out illegal activities. As the most important means of detection, radar has reached an imminent critical juncture in the detection of drones.
[0003] The conventional radar detection method is to identify the overall distance and direction of the target. The resolution is low and the output image is a point value. When targeting small aircraft, it is unable to distinguish between birds and aircraft, resulting in omissions, false alarms and other problems.
[0004] The current radar system still has many shortcomings. For example, the current radar system mainly targets large aircraft, identifies and tracks the overall movement trend of the aircraft, and has weak detection and recognition capabilities for small aircraft such as drones, and cannot distinguish natural creatures such as birds; at the same time, most of the current radar systems have poor accuracy and large errors, and cannot target small spacecraft. Therefore, this application innovates and improves the radar system in view of the above problems. Summary of the invention
[0005] In view of the above defects in the prior art, the present invention provides an object recognition method for radar systems, which can distinguish whether the object to be detected is an aircraft object or a non-aircraft object. Compared with the traditional radar positioning method, it can not only identify large aircraft, but also detect and identify small aviation objects such as drones, and can distinguish small flying objects from natural creatures such as birds, making up for the shortcomings of radar in targeting small aircraft.
[0006] The present invention provides an object recognition method for a radar system, comprising: S1, up-converting an initial signal generated by the radar system to obtain a mixed signal; S2, suppressing out-of-band spurious signals of the mixed signal to obtain a transmission signal, and transmitting the transmission signal to an object to be detected; S3, acquiring an echo signal returned from the object to be detected, filtering the echo signal, extracting Doppler information from the filtered echo signal, and obtaining a final signal; S4, judging a signal curve and a frequency of the final signal, and determining whether the object to be detected is an aircraft object or a non-aircraft object.
[0007] Optionally, the method further includes: S5, when the object to be detected is an aircraft object, obtaining the blade length of the object to be detected according to the signal frequency of the final signal.
[0008] Optionally, the method further includes: S6, when the object to be detected is an aircraft object, obtaining a blade position of the object to be detected according to a signal curve of the final signal.
[0009] Optionally, filtering the echo signal and extracting Doppler information from the filtered echo signal to obtain a final signal includes: S31, modeling the echo signal, performing Dechirp processing on the modeled echo signal, and performing two-dimensional Fourier transform on the processed signal to obtain a final signal.
[0010] Optionally, after S3, the method further includes: S32, eliminating a signal error of the final signal.
[0011] Optionally, the blade length is L, then:
[0012] L=Ac 2 / (4πωf c -8πkR c ω)
[0013] Among them, A is the amplitude of the motion image of the instantaneous frequency of the flying arm of the object to be measured, c is the speed of light, ω is the angular velocity, and f c is the carrier frequency, k is the frequency modulation slope, R c is the distance between the target center and the radar.
[0014] Optionally, obtaining the blade position of the object to be detected based on the signal curve of the final signal includes: determining the position of at least one blade of the object to be detected based on at least one prominent high point of the signal curve of the final signal, wherein one prominent high point corresponds to one blade.
[0015] Optionally, the signal error σ r Including thermal drying error σ r1 , receiving delay σ r2 , pulse jitter σ r3 and data quantization error σ r4 ; then:
[0016]
[0017]
[0018]
[0019] Where S / N is the signal-to-noise ratio, B is the signal bandwidth, c is the speed of light, Δt s is the maximum jitter time of the pulse, R is the distance, m is the number of pulses in a group of pulses, R mis the average distance within a group of pulses, or the distance measured per unit time; then:
[0020] Optionally, the final signal is S d (f m , t n ), then:
[0021] S d (f m , t n )=T P / [1+2lωsin(ωt n +θ) / c]sinc{T P / [1+2lωsin(ωt n +θ) / c][f m -4lωsin(ωt n +θ) / λ+2klcos(ωt n +θ) / λ]}exp[-j4πlcos(ωt n +θ) / λ+j8πkR C lcos(ωt n +θ) / c 2 ]
[0022] Among them, T P is the pulse period of the radar transmitting the signal, l is the distance from the target point to the rotation center, ω is the angular velocity, t n is the slow time, θ is the initial angle, c is the speed of light, f m is the frequency at fast time, k is the frequency modulation slope, j is an imaginary number; R C is the distance between the target center and the radar.
[0023] Optionally, the method further includes: S7, outputting a tracking model of the final signal, wherein the tracking model is:
[0024]
[0025]
[0026] in, is the state estimate based on N final signals, is the state estimate of the ith final signal, u k (i) is the probability of model i at time k, P tij is the transition probability from model i to model j in the algorithm,
[0027]
[0028] An object recognition method for a radar system disclosed in the present invention can distinguish whether the object to be detected is an aircraft object or a non-aircraft object. Compared with the traditional radar positioning method, it can not only identify large aircraft, but also detect and identify small aviation space such as drones. It can distinguish between small flying objects and natural creatures such as birds, making up for the shortcomings of radar in targeting small aircraft.
[0029] The object recognition method for a radar system disclosed in the present invention adopts front-end suppression and back-end calculation adjustment, calculates and eliminates hardware errors such as heating error of transmitting elements and atmospheric propagation error, and improves signal resolution through algorithm optimization, thereby reducing clutter of transmitted signals, compensating for errors in echo signals, and greatly improving the accuracy of the radar system.
[0030] An object recognition method for a radar system disclosed in the present invention establishes a new data model. By identifying and tracking the movement of the object to be detected, small flying objects can be quickly distinguished and located. At the same time, by tracking and calculating the movement of the blades, detailed flight characteristics such as the blade length and blade position of the aircraft can be identified and output. Compared with the traditional radar recognition method that can only recognize the whole and output it as a point value, it has higher accuracy and is more convenient to distinguish between aircraft and natural organisms. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 A flowchart of an object recognition method for a radar system according to an embodiment of the present disclosure is schematically shown;
[0032] Figure 2 The schematic diagram schematically shows an algorithm process for obtaining a tracking model according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0033] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0034] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0035] Figure 1The flowchart of the object recognition method for a radar system according to an embodiment of the present disclosure is schematically shown.
[0036] The embodiment of the present disclosure provides an object recognition method for a radar system, such as Figure 1 As shown, including:
[0037] S1, up-converting the initial signal generated by the radar system to obtain a mixed signal;
[0038] S2, suppressing out-of-band spurious signals of the mixing signal to obtain a transmission signal, and transmitting the transmission signal to the object to be detected;
[0039] S3, acquiring the echo signal returned from the object to be detected, filtering the echo signal, extracting Doppler information from the filtered echo signal, and obtaining a final signal;
[0040] S4, judging the signal curve and frequency of the final signal, and determining whether the object to be detected is an aircraft object or a non-aircraft object;
[0041] S5, when the object to be detected is an aircraft object, obtaining the blade length and blade position of the object to be detected according to the signal frequency of the final signal.
[0042] Then, the transmitted signal is s t (t m , t n ),
[0043] s t (t m , t n )=rect(t m / T P )exp[j2π(f c t+kt m 2 / 2)]
[0044] Among them, t m For fast time; T P is the pulse period of the radar transmitting the signal; j is an imaginary number; f c is the carrier frequency; k is the frequency modulation slope.
[0045] Thus, the echo signal is s ε (t m , t n ),
[0046] s ε (t m , t n )=rect[(t m -2R P (tm , t n ) / c) / T P ]exp{j2π[f c (t-2R P (t m , t n ) / c)+k(t m -2R P (t m , t n ) / c) 2 / 2]}
[0047] Among them, t m For fast time; R P (t m , t n ) is the distance from the target point to the radar transmitting the signal; c is the speed of light; T P is the pulse period of the radar transmitting the signal; j is an imaginary number; f c is the carrier frequency; k is the frequency modulation slope.
[0048] In some embodiments, filtering the echo signal, extracting Doppler information from the filtered echo signal, and obtaining the final signal includes: S31, modeling the echo signal, performing Dechirp processing on the modeled echo signal, and performing a two-dimensional Fourier transform on the processed signal to obtain the final signal. The modeling is to establish a calculable mathematical model based on the wave of the echo signal, so as to facilitate the subsequent Dechirp processing and Fourier transform.
[0049] From the above formula we can know that the signal after Dechirp processing is s d (t m , t n ),
[0050]
[0051] Among them, R P (t m , t n ) is the distance from the target point to the radar transmitting the signal, c is the speed of light; T p is the pulse period of the radar transmitting the signal; 2R ref / c=τ ref , τ ref is the reference time, i.e. the time when the radar obtains the target center point signal; R ref =R c , R c is the distance between the target center and the radar, λ is the wavelength, j is an imaginary number, k is the frequency modulation slope, t m For quick time.
[0052] Thus, in some embodiments, the final signal is S d (f m , t n ),
[0053] S d (f m , t n )=T P / [1+2lωsin(ωt n +θ) / c]sinc{T P / [1+2lωsin(ωt n +θ) / c][f m -4lωsin(ωt n +θ) / λ+2klcos(ωt n +θ) / λ]}exp[-j4πlcos(ωt n +θ) / λ+j8πkR C lcos(ωt n +θ) / c 2 ]
[0054] Among them, T P is the pulse period of the radar transmitting the signal, l is the distance from the target point to the rotation center, ω is the angular velocity, t n is the slow time, θ is the initial angle, c is the speed of light, f m is the frequency at fast time, k is the frequency modulation slope, j is an imaginary number; RC is the distance between the target center point and the radar.
[0055] Thus, the instantaneous frequency f can be obtained by taking the derivative of the final signal and dividing it by 2π m-D , then
[0056] f m-D =Asin(ωt n +θ)
[0057] Therefore, we know that the amplitude of the final signal curve is A = 4πlωf c / c-8πkR C lω / c 2 .
[0058] Furthermore, in some embodiments, when the object to be detected is an aircraft object, the length of the blade of the aircraft is L, then:
[0059] L=Ac 2 / (4πωf c -8πkR c ω)
[0060] Among them, A is the amplitude of the motion image of the instantaneous frequency of the flying arm of the object to be measured, c is the speed of light, ω is the angular velocity, and f c is the carrier frequency, k is the frequency modulation slope, R c is the distance between the target center and the radar.
[0061] In the present invention, by measuring the blade length L of the aircraft, the output of the image can be refined, which is convenient for operators to exclude natural organisms and distinguish the types of aircraft, so that operators can confirm the target and process it.
[0062] In some embodiments, after S3, the method further includes: S32, eliminating a signal error of the final signal.
[0063] In some embodiments, the signal error σ r Including thermal drying error σ r1 , receiving delay σ r2 , pulse jitter σ r3 and data quantization error σ r4 ; then:
[0064]
[0065]
[0066]
[0067] Where S / N is the signal-to-noise ratio, B is the signal bandwidth, c is the speed of light, Δt s is the maximum jitter time of the pulse, R is the distance, m is the number of pulses in a group of pulses, R m It is the average distance within a group of pulses, or it can be said to be the distance measured per unit time.
[0068] For example, when σ r1 =0.058m,σ r2 ≈0.3m,σ r3 =0.26m,σ r4 =0.053m, then:
[0069] The present invention can screen out errors caused by equipment heating, atmospheric propagation, signal-to-noise and other reasons by eliminating the total hardware error, thereby improving the accuracy of radar tracking.
[0070] In some embodiments, obtaining the blade position of the object to be detected based on the signal curve of the final signal includes: determining the position of at least one blade of the object to be detected based on at least one prominent high point of the signal curve of the final signal, wherein one prominent high point corresponds to one blade.
[0071] In some embodiments, the method further includes: S6, outputting a tracking model of the final signal, wherein the tracking model is:
[0072]
[0073]
[0074] in, is the state estimate based on N final signals, is the state estimate of the ith final signal, u k (i) is the probability of model i at time k, P tij is the transition probability from model i to model j in the algorithm,
[0075]
[0076] Figure 2 The schematic diagram schematically shows an algorithm process for obtaining a tracking model according to an embodiment of the present disclosure.
[0077] in, is the state estimate based on N final signals, is the state estimate of model j, P j (k|k) is the corresponding state covariance matrix, u k (j) is the probability of model j at time k, and i, j = 1, 2, ..., N, then Figure 2 As shown, if, for example, at time k, the filter output at time k-1 is Through the interactive multi-model algorithm, it interacts as the filter input at time k, and is input into the filter together with the measurement Z(k) obtained by the radar signal feedback at time k, and the output
[0078]
[0079]
[0080]
[0081] At the same time, the output According to the Gaussian distribution, the probability of model j is updated. The filter residual of model j is The corresponding covariance is Obeying Gaussian distribution, the probability of model j is:
[0082]
[0083]
[0084] The state vector and its variance P j (k|k) together with the observation value Z(k) is used as the input value of the jth model at time k to obtain the corresponding filter output P j (k+1|k+1):
[0085]
[0086]
[0087] In summary, we know that the tracking model is:
[0088]
[0089]
[0090] in, is the state estimate based on N final signals, is the state estimate of the ith final signal, u k (i) is the probability of model i at time k, P tij is the transition probability from model i to model j in the algorithm,
[0091]
[0092] Since the present invention establishes a new data model, by identifying and tracking the motion of the detected object itself, and taking into account the characteristics of small spacecraft using propeller motion, since the radar will continuously transmit, receive and calculate, it can track the blade motion to achieve rapid identification of the target, and then achieve precise positioning through echo angle and echo time, which makes up for the defects of existing radar in identifying small spacecraft.
[0093] In addition, in some embodiments, the suppression of out-of-band spurious signals of the mixing signal in S2 is performed by using a filter, wherein the filter includes a cavity filter and a microstrip filter.
[0094] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. An object recognition method for a radar system, characterized in that: include: S1, up-converting the initial signal generated by the radar system to obtain a mixed signal; S2, suppressing out-of-band spurious signals of the mixing signal to obtain a transmission signal, and transmitting the transmission signal to the object to be detected; S3, acquiring the echo signal returned from the object to be detected, filtering the echo signal, extracting Doppler information from the filtered echo signal, and obtaining a final signal; S4, judging the signal curve and frequency of the final signal, and determining whether the object to be detected is an aircraft object or a non-aircraft object; S5, when the object to be detected is an aircraft object, the blade length of the object to be detected is obtained according to the signal frequency of the final signal; the blade length is L, then: in, is the amplitude of the motion image of the instantaneous frequency of the flying arm of the object to be measured, is the speed of light, is the angular velocity, is the carrier frequency, is the frequency modulation slope, is the distance between the target center point and the radar; S6, when the object to be detected is an aircraft object, obtaining the blade position of the object to be detected according to the signal curve of the final signal, including: determining the position of at least one blade of the object to be detected according to at least one protruding high point of the signal curve of the final signal, wherein one protruding high point corresponds to one blade.
2. The object recognition method for a radar system according to claim 1, characterized in that: The filtering of the echo signal and extracting Doppler information from the filtered echo signal to obtain a final signal comprises: S31, modeling the echo signal, performing Dechirp processing on the modeled echo signal, and performing two-dimensional Fourier transform on the processed signal to obtain a final signal.
3. The object recognition method for a radar system according to claim 1, characterized in that: After S3, it also includes: S32, eliminating the signal error of the final signal.
4. The object recognition method for a radar system according to claim 3, characterized in that: The signal error Including thermal error , receiving delay , Pulse jitter and data quantization error ; then: in, is the signal-to-noise ratio, is the signal bandwidth, is the speed of light, is the maximum jitter time of the pulse, For distance, is the number of pulses in a group of pulses, is the average distance within a group of pulses, or the distance measured per unit time; then: .
5. The object recognition method for a radar system according to claim 1, characterized in that: The final signal is , then: in, is the pulse period of the radar transmitting the signal, is the distance from the target point to the rotation center, is the angular velocity, For slow time, is the initial angle, is the speed of light, is the frequency in fast time, is the frequency modulation slope, is an imaginary number; is the distance between the target center and the radar.
6. The object recognition method for a radar system according to claim 1, characterized in that: Also includes: S7, outputting a tracking model of the final signal, wherein the tracking model is: in, is the state estimate based on N final signals, It is The state estimate of the final signal, for Moment Model The probability of From the model in the algorithm Transfer to Model The transition probability is: 。
