Long-distance high-speed weak target detection method and device based on DP-RFT

Abstract

The invention provides a DP-RFT-based long-distance high-speed weak target detection method and device, and belongs to the technical field of radar detection. In a target detection process, a dynamic programming (DP) algorithm is firstly adopted to estimate target motion parameters, and a DP parameter estimation result is used as prior information of a Radon Fourier transform (RFT) algorithm to solve the problem of blind speed sidelobe accumulated for a long time by the RFT; and then taking the target motion parameter obtained by the DP algorithm as a center, performing coherent accumulation by using an RFT algorithm, and obtaining a target motion parameter estimation value for compensation. And then target high-resolution one-dimensional range profile synthesis is carried out on the target echo after motion parameter compensation. According to the method, the target motion parameters can be estimated and the target high-resolution one-dimensional range profile can be synthesized under the condition of range ambiguity, so that the blind speed sidelobe is avoided, the operand is reduced, and the stable detection of the long-distance high-speed weak target is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of radar detection, and in particular relates to a DP-RFT-based long-range high-speed weak target radar detection method and device. Background Art

[0002] Radar is an important means of air early warning, which can monitor all day and all weather. The characteristics of current air targets are becoming more and more complex, which is reflected in weak echoes, high-speed flight, and complex movement patterns. According to the characteristics of the target, the research on radar long-range detection algorithms for high-speed and weak targets is of great significance for realizing space monitoring and air threat early warning.

[0003] High-speed targets can fly at speeds of several Mach to twenty Mach, and need to be detected early from a distance to improve the warning time. However, the "plasma sheath" effect makes the aircraft "thermally invisible" to radar detection, and the radar cross-section (RCS) is only 0.02m 2 ~0.5m 2 , the target's electromagnetic scattering is weak and difficult to detect. The characteristics of high-speed and weak targets are deeply coupled with the need for long-range detection, resulting in the inability of traditional methods to effectively detect and a low detection rate. Therefore, it is urgent to develop a radar accumulation detection method that can achieve stable detection of long-range high-speed weak targets. Summary of the invention

[0004] In view of this, the present invention provides a long-range, high-speed, and weak target radar detection method based on dynamic programming and Radon Fourier transform (DP-RFT), which can achieve stable detection of long-range, high-speed, and weak targets.

[0005] In order to solve the above technical problems, the present invention is implemented as follows.

[0006] A long-range, high-speed, and faint target detection method based on DP-RFT comprises the following steps:

[0007] Step 1: Perform range gating filtering on the radar echo;

[0008] Step 2: Motion parameter estimation and compensation: The dynamic programming (DP) algorithm is used to estimate the target motion parameters. The DP algorithm parameter estimation results are used as prior information to solve the problem of long-term accumulation of blind velocity side lobes in the Radon Fourier transform (RFT) algorithm. The target motion parameters obtained by the DP algorithm are used as the center, and the RFT algorithm is used for coherent accumulation to obtain the target motion parameter estimation value and perform compensation.

[0009] Step 3: synthesize the target one-dimensional range image of the target echo after the target motion parameter compensation.

[0010] Preferably, the method further comprises: constructing a long-distance high-speed weak target echo model s(t,m) under range gating as:

[0011]

[0012] Where t is the fast time, m is the sub-pulse sequence number, A r is the target complex amplitude, rect() represents the rectangular function, R m is the instantaneous radial distance of the target center at the mth sub-pulse, T p is the sub-pulse width, exp() represents the exponential function, f 0 is the signal carrier frequency, Δf is the frequency hopping interval, d r (m) is the echo frequency sequence, c is the speed of light;

[0013] When the gate is matched, the filter gate frequency and the echo frequency correspond one to one, then d(m) = d r (m), d(m) filter selection frequency sequence, phase encoding decoding error-free, the error term ε(t,m) is 0, the corresponding target echo after down-conversion to baseband is:

[0014]

[0015] When the gating is not matched, there is a deviation between the filter gating frequency and the echo frequency, there is an error in the phase decoding, ε(t,m) is not 0, and the corresponding target echo after down-conversion to baseband is:

[0016]

[0017] Among them, the error term d′(m) is the offset echo frequency, d(m) is the gated frequency sequence, is the phase encoding sequence of the echo, is the phase-coded sequence of the transmitted signal;

[0018] The existence of error terms affects the estimation of target motion parameters and the synthesis of one-dimensional range images;

[0019] The phase encoding of the transmitted signal is optimized to maximize the error caused by phase decoding when the gating does not match, thereby minimizing the range gating amplitude peak when the gating does not match, and preventing misjudgment of the target detection.

[0020] Preferably, the optimization of phase encoding is achieved by using a genetic algorithm.

[0021] Preferably, the genetic algorithm uses the minimum autocorrelation sidelobe peak as a fitness function to optimize phase coding; the minimum autocorrelation sidelobe peak indicates that the error term is the largest when the gating is not matched;

[0022] in, represents the pth group of phase encoding designed by genetic algorithm The autocorrelation function value at the i-th side lobe in is:

[0023]

[0024] in,(·) * represents conjugation, n is the sequence number in phase encoding, and N is the phase encoding length;

[0025] Then the autocorrelation sidelobe peak is:

[0026]

[0027] Preferably, in the step 2, the coherent accumulation using the RFT algorithm is:

[0028]

[0029]

[0030] Among them, RFT(r,v) is the coherent accumulation result; (r,v) is the target motion parameter to be searched, r is the distance, v is the speed, and the search target motion parameter is centered on the distance and speed estimates obtained by the DP algorithm; p is the sub-pulse number to be searched, M is the total number of sub-pulses, and T r is the PRT of the sub-pulse, H(p,r,v) is the RFT filter bank; f p is the sub-pulse frequency; c is the speed of light; s() is the long-distance high-speed weak target echo model when the gate matching is shown in formula (II);

[0031] When the searched target motion parameters match the target's true motion parameters, the output value of RFT(r,v) reaches the maximum.

[0032] The present invention also provides a long-distance, high-speed, and faint target detection device based on DP-RFT, comprising a range gating module, a motion parameter estimation and compensation module, and a one-dimensional range image synthesis module connected in sequence;

[0033] The range gating module is used to perform range gating filtering on the radar echo;

[0034] The motion parameter estimation and compensation module is used to estimate the target motion parameters by using a dynamic programming (DP) algorithm, and the DP algorithm parameter estimation result is used as a priori information to solve the problem of long-term accumulation of blind speed side lobes by the Radon Fourier Transform (RFT) algorithm; with the target motion parameters obtained by the DP algorithm as the center, the RFT algorithm is used for coherent accumulation to obtain the target motion parameter estimation value and perform compensation;

[0035] The one-dimensional range image synthesis module is used to synthesize the target one-dimensional range image of the target echo after the target motion parameter is compensated.

[0036] Preferably, the device further includes a phase encoding optimization module, which is used to determine the target misjudgment factors in the case of gating mismatch based on the construction and analysis of the long-distance high-speed weak target echo model under range gating, and optimize the phase encoding of the transmitted signal to maximize the error caused by phase decoding when the gating mismatch occurs, thereby minimizing the range gating amplitude peak when the gating mismatch occurs, and preventing misjudgment of the detected target.

[0037] Preferably, the long-distance high-speed weak target echo model s(t,m) under range gating constructed by the phase encoding optimization module is:

[0038]

[0039] Where t is the fast time, m is the sub-pulse sequence number, A r is the target complex amplitude, rect() represents the rectangular function, R m is the instantaneous radial distance of the target center at the mth sub-pulse, T p is the sub-pulse width, exp() represents the exponential function, f 0 is the signal carrier frequency, Δf is the frequency hopping interval, d r (m) is the echo frequency sequence, c is the speed of light;

[0040] When the gate is matched, the filter gate frequency and the echo frequency correspond one to one, then d(m) = d r (m), d(m) filter selection frequency sequence, phase encoding decoding error-free, the error term ε(t,m) is 0, the corresponding target echo after down-conversion to baseband is:

[0041]

[0042] When the gating is not matched, there is a deviation between the filter gating frequency and the echo frequency, there is an error in the phase decoding, ε(t,m) is not 0, and the corresponding target echo after down-conversion to baseband is:

[0043]

[0044] Among them, the error term d′(m) is the offset echo frequency, d(m) is the gated frequency sequence, is the phase encoding sequence of the echo, is the phase-coded sequence of the transmitted signal;

[0045] The existence of error terms affects the estimation of target motion parameters and the synthesis of one-dimensional range images.

[0046] Preferably, the phase coding optimization module uses a genetic algorithm to optimize the phase coding of the transmission signal; the genetic algorithm uses the minimum autocorrelation sidelobe peak as the fitness function; the minimum autocorrelation sidelobe peak indicates that the error term is maximum when the selection is not matched.

[0047] Preferably, the motion parameter estimation and compensation module uses the RFT algorithm to perform coherent accumulation as follows:

[0048]

[0049]

[0050] Among them, RFT(r,v) is the coherent accumulation result; (r,v) is the target motion parameter to be searched, r is the distance, v is the speed, and the search target motion parameter is centered on the distance and speed estimates obtained by the DP algorithm; p is the sub-pulse number to be searched, M is the total number of sub-pulses, and T r is the PRT of the sub-pulse; H(p,r,v) is the RFT filter bank; f p is the sub-pulse frequency; c is the speed of light; s() is the long-distance high-speed weak target echo model when the gate matching is shown in formula (II);

[0051] When the searched target motion parameters match the target's true motion parameters, the output value of RFT(r,v) reaches the maximum.

[0052] Beneficial effects:

[0053] (1) When estimating target motion parameters, the present invention adopts a Radon Fourier transform parameter estimation method based on DP optimization. Radon Fourier transform is a long-term accumulation algorithm with the largest target accumulation gain. However, the use of this algorithm alone cannot avoid the problem of blind speed side lobes and cannot meet the problem of long-distance, high-speed, weak and small targets. Therefore, the present invention adopts DP to solve the problem of RFT. DP provides a priori. The DP algorithm has the characteristics of low signal-to-noise ratio, fast trajectory acquisition, and is not easily affected by speed ambiguity. Therefore, the DP algorithm can avoid blind speed side lobes, while reducing the amount of calculation and solving the problem of speed ambiguity; based on DP optimization, RFT solves the problem of high-precision parameter estimation. Moreover, both DP and RFT are applicable to scenes with weak target echoes. Therefore, Radon Fourier transform parameter estimation based on DP optimization can estimate target motion parameters and synthesize high-resolution one-dimensional range images of targets in the case of distance ambiguity, avoid blind speed side lobes, reduce the amount of calculation, and thus achieve stable detection of long-distance, high-speed and weak targets. It has been verified that the speed estimation accuracy of the maneuvering target detection algorithm based on DP-RFT can be reduced to below 0.5m / s, the accuracy is improved by 2 times, and the distance error accuracy is reduced to 10 -4 m level, the accuracy is improved by nearly 2 times.

[0054] (2) The present invention also constructs a long-range high-speed weak target echo model under range gating, and analyzes and verifies the error of gating mismatch. According to the error factor, the present invention considers from the perspective of phase decoding, and designs to maximize the error caused by phase decoding when gating mismatches, so as to minimize the range gating amplitude peak when gating mismatches, reduce or eliminate the misjudgment of the detected target, and thus optimize the phase coding. By adopting an optimized phase-coded transmission signal, it is possible to reduce or avoid the misjudgment of the target still being detected in the case of gating mismatch, and further improve the stability of long-range high-speed weak target detection.

[0055] (3) In a preferred embodiment, a genetic algorithm is used to optimize phase coding, and autocorrelation is selected as an indicator to measure the performance of phase coding, so as to quickly determine the optimized phase coding. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of a long-distance, high-speed, and faint target detection method based on DP-RFT of the present invention;

[0057] Figure 2 This is a schematic diagram of distance gating;

[0058] Figure 3 The target motion parameter estimation error histograms: (a) is the distance error histogram of the DP algorithm; (b) is the speed error histogram of the DP algorithm; (c) is the distance error histogram of the DP-RFT algorithm; (d) is the speed error histogram of the DP-RFT algorithm;

[0059] Figure 4 The one-dimensional high-resolution range image comparison of the target of range gating of the present invention: (a) is the range gating statistical histogram; (b) is the single range gating comparison diagram;

[0060] Figure 5 It is a schematic diagram of a long-range, high-speed, faint target detection device based on DP-RFT of the present invention. DETAILED DESCRIPTION

[0061] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0062] The present invention provides a long-range, high-speed, and weak target detection method based on DP-RFT. The basic idea is to use DP-optimized Radon Fourier transform to perform coherent accumulation to obtain target motion parameters, and perform high-resolution one-dimensional range image synthesis on the target echo after motion parameter compensation. The scheme can estimate the target motion parameters and synthesize the target high-resolution one-dimensional range image under the condition of distance ambiguity, avoid blind speed side lobes, reduce the amount of calculation, and thus achieve stable detection of long-range, high-speed, and weak targets; the present invention also establishes a long-range, high-speed, and weak target echo model under range gating, which is used to verify the influence of errors on target detection under the two conditions of gating matching and gating mismatching, and optimizes the design of the phase encoding of the transmitted signal to maximize the error caused by phase decoding when the gating is mismatched, thereby minimizing the range gating amplitude peak when the gating is mismatched, reducing or eliminating the misjudgment of the detected target, and further improving the stability of long-range, high-speed, and weak target detection.

[0063] like Figure 1 As shown, the long-distance high-speed faint target detection method based on DP-RFT of the present invention comprises the following steps:

[0064] Step 1: Establish a long-range, high-speed, weak target echo model under range gating.

[0065] The radar transmits a frequency-stepped signal composed of M sub-pulses, and the multi-channel receiver simultaneously performs down-conversion reception processing with the carrier frequencies of the M sub-pulses. For a multi-channel receiver, different channels can only match and parse the target echo of the corresponding carrier frequency. The filter will only have a corresponding output when there is a target within the assumed distance range; if there is no target within the assumed distance range, the filter will have no output.

[0066] The expression of the stepped frequency simple pulse transmission signal s(t,m) is as follows:

[0067]

[0068] Among them, rect() represents the rectangular function, and the sub-pulse width is T p , t is the fast time, exp() represents the exponential function, f 0 is the signal carrier frequency, m is the sub-pulse sequence, M is the total number of sub-pulses, Δf is the frequency hopping interval, d s (m) represents the frequency sequence of the transmitted signal, more precisely, the frequency of the mth sub-pulse, represents the phase coding sequence of the transmitted signal, more precisely, the phase coding corresponding to the mth sub-pulse.

[0069] Consider a single target scene, assuming that the motion parameters of the target center are α M =[α 0 ,α 1 ,α 2 ,…αK-1 ] indicates that α 0 ,α 1 ,α 2 Respectively represent the initial radial distance of the target R 0 , initial velocity v 0 , initial acceleration a 0 , and so on. K represents the highest order of target motion. At the mth sub-pulse, the instantaneous radial distance R of the target center m It can be expressed as:

[0070]

[0071] Among them, T r It is the PRT (Pulse Repetition Time) of the sub-pulse.

[0072] Substituting equation (2) into equation (1), taking into account the delay of the transmitted signal when the target echoes, the target echo expression can be obtained:

[0073]

[0074] In the above formula, c is the speed of light.

[0075] Based on formula (3), distance gating filtering is performed later. Distance gating means that the filter at the corresponding frequency hopping point is used at the receiving end to filter the received wave. The filter will have a corresponding output only when there is a target within the gating matching distance segment; if there is no target within the gating matching distance segment, the filter will have no output.

[0076] The echo expression after range gating is as follows:

[0077]

[0078] Among them, A r is the target complex amplitude, d r (m) is the echo frequency sequence, and ε(t,m) is the error term.

[0079] When the gate is matched, the filter gate frequency and the echo frequency correspond one to one, that is, d r (m)d=m(), d(m) filter selects the frequency sequence; phase encoding decoding is error-free, the error term ε(t,m) is 0, and the corresponding target echo after down-conversion to baseband is:

[0080]

[0081] When the gating does not match, there is a deviation between the filter gating frequency and the echo frequency, there is an error in the phase decoding, ε(t,m) is not 0, and the echo expression is as follows:

[0082]

[0083] in, is the error term, d′(m) is the offset echo frequency, d(m) is the gated frequency sequence, is the phase encoding sequence of the echo; It is the phase coding sequence of the transmitted signal and also the phase coding sequence of the pass filter.

[0084] In order to maximize the error caused by phase decoding when the gate does not match, the peak value of the range gate amplitude is minimized when the gate does not match (such as Figure 4 The peak value of the red signal in (b) is the smallest), and no misjudgment of the target is detected. Phase encoding needs to be optimized. For a set of phase encoding sets, its autocorrelation and cross-correlation are important indicators to measure the performance. The calculation method is as follows:

[0085]

[0086]

[0087] in, Indicates the pth group code The autocorrelation function at the i-th side lobe in , i = 0 represents the autocorrelation main lobe, i ≠ 0 represents the autocorrelation side lobe; represents the cross-correlation function between the pth group code and the qth group code. n is the sequence number in the phase coding, and N is the phase coding length.

[0088] ASP represents the autocorrelation sidelobe peak, and CP represents the cross-correlation peak, that is:

[0089]

[0090] Among them, max means taking the maximum value.

[0091] The present invention adopts an inter-pulse orthogonal phase coding design algorithm based on a genetic algorithm, which minimizes the autocorrelation sidelobe peak value I s As a fitness function, the minimum autocorrelation sidelobe peak indicates that the error term is the largest when the gating does not match, and its expression is as follows:

[0092]

[0093] in, represents the phase encoding designed by the algorithm, Represents the autocorrelation function of the group of codes.

[0094] The optimized phase encoding is used to adjust the transmission signal, so that after the echo is range-gated, in the case of gating mismatch, an excessively large range gating amplitude peak is not generated, thus avoiding false target detection.

[0095] Step 2: Perform range-select filtering on the radar echo.

[0096] Step 3: Target motion parameter estimation and compensation are performed based on Radon Fourier transform (RFT) optimized by dynamic programming (DP).

[0097] High-speed targets move fast, and the distance movement between step-frequency pulses cannot be ignored, so motion parameter estimation and compensation are required. However, at the same time, the target echo signal-to-noise ratio is low in the long-range detection background, and long-term pulse accumulation is required to improve the processing gain. In addition, the target moves fast, which is prone to velocity ambiguity.

[0098] There are many traditional long-term target accumulation algorithms, among which the Radon Fourier Transform (RFT) algorithm matches the target motion trajectory by searching the distance and speed in two dimensions, thereby achieving long-term coherent accumulation of the target and estimation of motion parameters, and its theoretical accumulation gain is optimal. However, for target characteristics with weak motion of target echo and high-speed motion, using RFT to estimate target motion parameters has problems such as blind speed sidelobes and large amount of calculation.

[0099] Therefore, the present invention improves RFT to form a Radon Fourier Transform (RFT) parameter estimation method based on dynamic programming (DP) optimization, which is called DP-RFT algorithm. The DP algorithm preferably adopts DP-TBD. First, the DP-TBD algorithm is used to realize the rapid acquisition of low signal-to-noise ratio target trajectory, thereby avoiding the influence of speed ambiguity problem on the accumulation algorithm; then, the target motion parameters are roughly estimated based on the target motion trajectory obtained by DP-TBD, and the rough estimate is used as the prior information of the RFT algorithm. By reasonably setting the parameter search range, the RFT calculation amount is reduced, and the RFT blind speed sidelobe problem is solved; finally, the high accumulation gain and high search accuracy of the RFT algorithm are used to complete the accurate estimation of the target motion parameters.

[0100] The present invention adopts the pre-placed DP algorithm, the purpose of which is not simply to increase a search, but to use the DP algorithm to help the RFT algorithm avoid blind speed side lobes, while reducing the amount of calculation and solving the speed ambiguity problem.

[0101] Specifically, assuming that the nth distance sampling unit of the kth PRT is The corresponding value function is The observed value is The state transition range in the k-1th PRT corresponding to the distance sampling unit is The state transfer matrix is State transfer range The state information is used to predict the possible state of the target. The state transition matrix records the state correspondence between adjacent PRTs. The DP algorithm process is as follows:

[0102] Initialize the value function and state transfer matrix as follows:

[0103]

[0104]

[0105] in, is the value function of the nth distance sampling unit of the first pulse; is the observation value corresponding to the nth distance sampling unit of the first pulse;

[0106] After initialization, the recursive accumulation function and the update state transfer matrix are started as follows:

[0107]

[0108]

[0109] After completing the accumulation of K PRTs, the accumulated and the detection threshold V T Compare and find all states greater than the threshold, recorded as

[0110]

[0111] By backtracking the state transfer matrix, the complete track can be detected. Dynamic programming can quickly obtain the status of each stage of potential targets, and By doing the least squares estimation, we can get a rough estimate of the initial radial distance and velocity of the target.

[0112] Target motion parameters obtained by DP algorithm With the RFT-based parameter estimation algorithm as the center, the parameter estimation accuracy is further improved. According to the phase form of (5), RFT is used to achieve coherent accumulation. The specific form is as follows:

[0113]

[0114] Where (r, v) is the search motion parameter, r is the distance, v is the speed, and p is the sub-pulse number to be searched; H(p, r, v) is the RFT filter bank, f p =(f 0 +d(p)Δf) is the sub-pulse frequency. By judging the accumulated peak value of the two-dimensional function RFT(r,v), if the search parameters match the target's true motion parameters, the output value of RFT reaches the maximum, and a peak is generated on the two-dimensional plane, such as Figure 4 The peak shown in Figure (b) can realize the parameter estimation of the target echo signal.

[0115] Step 4: synthesize the target high-resolution one-dimensional range image of the target echo after motion parameter compensation.

[0116] By determining in which assumed distance segment the target can obtain a high-resolution range profile (HRRP), the target's true range ambiguity number can be determined, thereby enabling the detection of long-distance and high-speed targets. After the target motion parameters are compensated, its echo expression is as follows:

[0117]

[0118] Based on equation (17), a high-resolution one-dimensional range image of the target can be synthesized. Only when the gating is matched and the motion parameters are estimated accurately, can a high-resolution one-dimensional range image of the target be obtained.

[0119] In order to verify the target parameter estimation accuracy and range gating performance of the long-range, high-speed and weak target detection method based on DP-RFT described above, a simulation experiment was conducted to verify the radar system parameters and target motion parameters as follows:

[0120] Table 1 Radar system parameters and target motion parameters

[0121] Parameter properties Parameter Value Carrier center frequency 10GHz Frequency hopping points 128 Frequency hopping interval 20MHz Pulse Repetition Time (PRT) 50us Sub-pulse width 50ns Distance sampling unit width 0.9375m Target initial slope range 460km speed 1500m / s

[0122] According to the system parameters, the single pulse echo signal-to-noise ratio can be calculated by the radar equation. The radar equation parameters are as follows:

[0123] Table 2 Radar equation parameters

[0124] Parameter properties Parameter Value <![CDATA[Peak power P t > 1GW RCSσ <![CDATA[0.1m 2 ]]> <![CDATA[Transmit antenna gain G t > 48.19dB <![CDATA[Received antenna gain G r > 48.19dB <![CDATA[System noise temperature T s > 627.06K Sub-pulse width B 50ns Wavelength 3cm <![CDATA[System loss L s > 10dB Boltzmann constant k 1.38*10^(-23)

[0125] The radar equation is:

[0126]

[0127] Among them, R max is the maximum detection distance.

[0128] From the radar equation, we know that the SNR of a single pulse corresponding to a target at 460 km is 4.05 dB, so the echo signal-to-noise ratio is set to 4 dB.

[0129] On this basis, the performance of the proposed maneuvering target detection algorithm based on DP-RFT is verified from the aspects of accuracy and precision of parameter estimation, and compared with the traditional DP-TBD.

[0130] The error calculation method is as follows:

[0131]

[0132] Among them, ΔR and Δv represent the distance error and speed error respectively, R and v represent the estimated distance and speed respectively, the superscript i represents the result of the i-th experiment, R real With v real Represents the true value of distance and the true value of speed.

[0133] Assume N is the number of Monte Carlo simulations, and calculate the statistical mean and statistical error according to the following formula:

[0134]

[0135] Among them, x mean represents the statistical mean, x std represents the sample mean square error of the statistic. x represents the data sample, and x represents the mean of the data sample. The statistical results are as follows:

[0136] Table 3. Statistical results of target motion parameter estimation errors

[0137] Parameter properties DP-TBD algorithm DP-RFT algorithm of the present invention Speed ​​error mean 16.7574m / s 0.4776m / s Speed ​​error mean square error 243.7526m / s 100.7277m / s Mean distance error 0.5363m <![CDATA[-6*10 -4 m]]> Distance error mean square error 0.8094m 0.4273m

[0138] like Figure 3 As shown in the figure, after 1000 Monte Carlo simulations, the statistical histograms of the distance error and speed error of the two methods can be obtained. It can be seen that compared with the traditional DP-based detection-before-tracking algorithm, the speed estimation accuracy of the maneuvering target detection algorithm based on DP-RFT of the present invention can be reduced to less than 0.5m / s, the accuracy is improved by 2 times, and the distance error accuracy is reduced to 10 -4 m level, the accuracy is improved by nearly 2 times.

[0139] Subsequently, the range gating capability was verified, and the verification standard was the ratio of the peak power of the next-dimensional range image between the gated match and the gated non-match. Among them, "gated match" means that the target exists in the gated range segment, and "gated non-match" means that the target does not exist in the gated range segment.

[0140] Assume that the peak power of the one-dimensional range image of the gated matching target is P m , the peak power of the one-dimensional range image of the non-matching target is P n, and the ratio of the two is used as a measure of the range gating capability, as shown below:

[0141]

[0142] Among them, Q mf Q is a measure of the range gating capability, which indicates the magnitude of the peak power drop of the gated mismatched target compared to the gated matched target. mf The smaller it is, the stronger the range gating capability is.

[0143] like Figure 4 As shown in Figure 2, after 1000 Monte Carlo simulations, the range gating capability Q can be obtained. mf The statistical histogram of . The average value of the statistical results is taken as the final result, as shown below:

[0144]

[0145] Among them, mean means finding the mean.

[0146] The result of formula (22) is -12.9017dB. Figure 4 It can be seen that compared with the gated matched target, the peak power of the one-dimensional range image of the gated non-matching target decreases by an average of 13 dB, which can distinguish the gated matched target from the gated non-matching target. Therefore, this method has the ability of range gating and can achieve stable detection of long-distance, high-speed and weak targets.

[0147] Based on the above method, the present invention also provides a long-distance high-speed weak target detection device based on DP-RFT, such as Figure 5 As shown, it includes a distance gating module, a motion parameter estimation and compensation module and a one-dimensional range image synthesis module which are connected in sequence.

[0148] Range gating module, used to perform range gating filtering on radar echoes;

[0149] The motion parameter estimation and compensation module is used to estimate the target motion parameters using a dynamic programming (DP) algorithm. The DP parameter estimation result is used as the prior information of the Radon Fourier transform (RFT) algorithm to solve the blind speed sidelobe problem accumulated over a long period of time by the RFT algorithm. The target motion parameters obtained by the DP algorithm are taken as the center, and the RFT algorithm is used for coherent accumulation to obtain the target motion parameter estimation value for compensation. The RFT algorithm is shown in the above formula (16).

[0150] The one-dimensional range image synthesis module is used to synthesize the target high-resolution one-dimensional range image of the target echo after motion parameter compensation.

[0151] The device also includes a phase coding optimization module, which determines the target misjudgment factor in the case of gating mismatch based on the construction and analysis of the long-distance high-speed weak target echo model under range gating, optimizes the phase coding of the transmitted signal, maximizes the error caused by phase decoding when the gating mismatch occurs, and minimizes the range gating amplitude peak when the gating mismatch occurs, so as to avoid misjudgment of the detected target. The long-distance high-speed weak target echo model s(t,m) under range gating constructed by the phase coding optimization module is shown in the above formulas (4)-(6).

[0152] Among them, the phase decoding optimization can adopt a genetic algorithm; the genetic algorithm takes the minimum autocorrelation sidelobe peak as the fitness function; the minimum autocorrelation sidelobe peak means that the error term is the largest when the selection is not matched.

[0153] The above specific embodiments only describe the design principle of the present invention. The shapes and names of the components in the description may be different and are not limited. Therefore, those skilled in the art in the field of the present invention may modify or replace the technical solutions recorded in the above embodiments; and these modifications and replacements do not deviate from the creative purpose and technical solutions of the present invention and should all fall within the protection scope of the present invention.

Claims

1. A long-distance, high-speed, and faint target detection method based on DP-RFT, characterized in that: The following steps are involved: Step 1: Perform range gating filtering on the radar echo; Step 2: Motion parameter estimation and compensation: The dynamic programming (DP) algorithm is used to estimate the target motion parameters. The DP algorithm parameter estimation results are used as prior information to solve the problem of long-term accumulation of blind velocity side lobes in the Radon Fourier transform (RFT) algorithm. The target motion parameters obtained by the DP algorithm are used as the center, and the RFT algorithm is used for coherent accumulation to obtain the target motion parameter estimation value and perform compensation. Step 3: synthesize the target one-dimensional range image of the target echo after the target motion parameter compensation.

2. The long-distance, high-speed, faint target detection method based on DP-RFT as claimed in claim 1, characterized in that: The method further includes: constructing a long-distance high-speed weak target echo model s(t,m) under range gating as: Where t is the fast time, m is the sub-pulse sequence number, A r is the target complex amplitude, rect() represents the rectangular function, R m is the instantaneous radial distance of the target center at the mth sub-pulse, T p is the sub-pulse width, exp() represents the exponential function, f0 is the signal carrier frequency, Δf is the frequency hopping interval, d r (m) is the echo frequency sequence, c is the speed of light; When the gate is matched, the filter gate frequency and the echo frequency correspond one to one, then d(m) = d r (m), d(m) filter selection frequency sequence, phase encoding decoding error-free, the error term ε(t,m) is 0, the corresponding target echo after down-conversion to baseband is: When the gating is not matched, there is a deviation between the filter gating frequency and the echo frequency, there is an error in the phase decoding, ε(t,m) is not 0, and the corresponding target echo after down-conversion to baseband is: Among them, the error term d′(m) is the offset echo frequency, d(m) is the gated frequency sequence, is the phase encoding sequence of the echo, is the phase-coded sequence of the transmitted signal; The existence of error terms affects the estimation of target motion parameters and the synthesis of one-dimensional range images; The phase encoding of the transmitted signal is optimized to maximize the error caused by phase decoding when the gating does not match, thereby minimizing the range gating amplitude peak when the gating does not match, and preventing misjudgment of the target detection.

3. The long-distance, high-speed, faint target detection method based on DP-RFT as claimed in claim 2, characterized in that: The optimization of phase encoding is achieved using a genetic algorithm.

4. The long-distance, high-speed, faint target detection method based on DP-RFT as claimed in claim 3, characterized in that: The genetic algorithm uses the minimum autocorrelation sidelobe peak as a fitness function to optimize phase coding; the minimum autocorrelation sidelobe peak indicates that the error term is the largest when the gating is not matched; in, represents the pth group of phase encoding designed by genetic algorithm The autocorrelation function value at the i-th side lobe in is: in,(·) * represents conjugation, n is the sequence number in phase encoding, and N is the phase encoding length; Then the autocorrelation sidelobe peak is:

5. The long-distance, high-speed, faint target detection method based on DP-RFT as claimed in claim 2, characterized in that: In the step 2, the coherent accumulation using the RFT algorithm is as follows: Among them, RFT(r,v) is the coherent accumulation result; (r,v) is the target motion parameter to be searched, r is the distance, v is the speed, and the search target motion parameter is centered on the distance and speed estimates obtained by the DP algorithm; p is the sub-pulse number to be searched, M is the total number of sub-pulses, and T r is the PRT of the sub-pulse, H(p,r,v) is the RFT filter bank; f p is the sub-pulse frequency; c is the speed of light; s() is the long-distance high-speed weak target echo model when the gate matching is shown in formula (II); When the searched target motion parameters match the target's true motion parameters, the output value of RFT(r,v) reaches the maximum.

6. A long-distance, high-speed, faint target detection device based on DP-RFT, characterized in that: It includes a range gating module, a motion parameter estimation and compensation module, and a one-dimensional range image synthesis module which are connected in sequence; The range gating module is used to perform range gating filtering on the radar echo; The motion parameter estimation and compensation module is used to estimate the target motion parameters by using a dynamic programming (DP) algorithm, and the DP algorithm parameter estimation result is used as a priori information to solve the problem of long-term accumulation of blind speed side lobes by the Radon Fourier Transform (RFT) algorithm; with the target motion parameters obtained by the DP algorithm as the center, the RFT algorithm is used for coherent accumulation to obtain the target motion parameter estimation value and perform compensation; The one-dimensional range image synthesis module is used to synthesize the target one-dimensional range image of the target echo after the target motion parameter is compensated.

7. The long-distance, high-speed, faint target detection device based on DP-RFT as claimed in claim 6, characterized in that: The device further includes a phase coding optimization module, which is used to determine the target misjudgment factors in the case of gating mismatch based on the construction and analysis of the long-distance high-speed weak target echo model under range gating, and optimize the phase coding of the transmitted signal to maximize the error caused by phase decoding when the gating mismatch occurs, thereby minimizing the range gating amplitude peak when the gating mismatch occurs, and preventing misjudgment of the detected target.

8. The DP-RFT-based long-range, high-speed, faint target detection device as claimed in claim 7, characterized in that: The long-range high-speed weak target echo model s(t,m) under range gating constructed by the phase encoding optimization module is: Where t is the fast time, m is the sub-pulse sequence number, A r is the target complex amplitude, rect() represents the rectangular function, R m is the instantaneous radial distance of the target center at the mth sub-pulse, T p is the sub-pulse width, exp() represents the exponential function, f0 is the signal carrier frequency, Δf is the frequency hopping interval, d r (m) is the echo frequency sequence, c is the speed of light; When the gate is matched, the filter gate frequency and the echo frequency correspond one to one, then d(m) = d r (m), d(m) filter selection frequency sequence, phase encoding decoding error-free, the error term ε(t,m) is 0, the corresponding target echo after down-conversion to baseband is: When the gating is not matched, there is a deviation between the filter gating frequency and the echo frequency, there is an error in the phase decoding, ε(t,m) is not 0, and the corresponding target echo after down-conversion to baseband is: Among them, the error term d′(m) is the offset echo frequency, d(m) is the gated frequency sequence, is the phase encoding sequence of the echo, is the phase-coded sequence of the transmitted signal; The existence of error terms affects the estimation of target motion parameters and the synthesis of one-dimensional range images.

9. The long-distance, high-speed, faint target detection device based on DP-RFT as claimed in claim 6, characterized in that: The phase coding optimization module uses a genetic algorithm to optimize the phase coding of the transmission signal; the genetic algorithm uses the minimum autocorrelation sidelobe peak as the fitness function; the minimum autocorrelation sidelobe peak indicates that the error term is maximum when the selection is not matched.

10. The DP-RFT-based long-range high-speed faint target detection device according to claim 6, characterized in that: The motion parameter estimation and compensation module uses the RFT algorithm to perform coherent accumulation as follows: Among them, RFT(r,v) is the coherent accumulation result; (r,v) is the target motion parameter to be searched, r is the distance, v is the speed, and the search target motion parameter is centered on the distance and speed estimates obtained by the DP algorithm; p is the sub-pulse number to be searched, M is the total number of sub-pulses, and T r is the PRT of the sub-pulse; H(p,r,v) is the RFT filter bank; f p is the sub-pulse frequency; c is the speed of light; s() is the long-distance high-speed weak target echo model when the gate matching is shown in formula (II); When the searched target motion parameters match the target's true motion parameters, the output value of RFT(r,v) reaches the maximum.

Images