CN103257346B - Automotive anti-collision radar multi-target detecting method and system - Google Patents

Abstract

The invention provides an automotive anti-collision radar multi-target detecting method and system. The method comprises the steps of emitting two types of triangular waves with different modulation periods in an alternating mode and acquiring echo data, carrying out windowing processing, carrying out distance dimension FFT and speed dimension FFT, carrying out modulo processing to obtain frequency spectrum of two types of echo waves, carrying out target paring to obtain a spectral line of the same target of the two echo waves, calculating distance and speed of each target, and judging target distance and speed obtained by the two echo waves through a tolerance function to obtain a final target. A radio frequency emitting and receiving part of the system comprises a radar sensor and an intermediate frequency processing module, and a data processing part comprises a modulus, a modulus converting module and a central control processing module FPGA. The FPGA comprises a modulation signal producing sub-module, an echo wave signal acquisition sub-module, an algorithm sub-module and a control sub-module. The modulation waves and corresponding algorithms effectively remove false targets and improve accuracy that multiple moving targets are detected under strong noise. A hardware system is simplified in structure and easy to achieve.

Description

A kind of automobile collision avoidance radar multiple target detection method and system

Technical field

The present invention relates to automobile early warning technology field, be specially a kind of automobile collision avoidance radar multiple target detection method and system.

Background technology

Traffic hazard becomes " the first in the world evil ", how to guarantee traffic safety, and the incidence that how to reduce relevant traffic hazard has become the topic that the whole society pays close attention to.Under this background, the research and development of automobile early warning technology become focus, and wherein the development of anti-collision system for automobile has very important realistic meaning and using value.

Automobile collision avoidance radar can adopt ultrasound wave, laser, infrared and microwave technology to realize, but first three plants the impact that is easily subject to inclement weather and environmental factor, cannot guarantee detection accuracy, thereby expert concentrates notice to be placed on microwave radar both at home and abroad.

In intelligent transportation system, automobile collision avoidance radar adopts linear frequency modulation continuous wave system (LFMCW) conventionally, in the radar intermediate frequency signal producing at radar receiver, the IF-FRE that not only contains target, also contain radar noise signal and undesired signal, as protecting between the vehicle on contiguous track, track blocked, the trees in roadside and aerial and buildings at a distance etc., these all can form radar system and disturb, and cause radar to make false judgment.And lower, a large amount of omission target, the false targets of multi-target detection ability is its application of restriction and universal key issue.

According to the research of German Daimler-Benz company, show: driver, as long as can shift to an earlier date 0.5 second obtains " early warning ", just can avoid 50% rear-end collision accident, 30% central collision accident; If can shift to an earlier date 1 second, take measures, can avoid 90% traffic hazard generation.It is real-time that this just requires anti-collision radar system to have, the feature that processing speed is fast.The realization of collision avoidance radar signal processing circuit is based on general MCU chip, dsp chip or fpga chip at present.The mode of operation of DSP in essence with traditional C PU instruction fetch, decoding is identical with modes such as execution, is subject to the restriction of instruction cycle, a little less than sequential control, and complicated for computing, the algorithm that calculated amount is large is often difficult to meet real-time demand, is also unfavorable for the control of car radar.FPGA has the advantages such as the capable processing of powerful Bing Neng Li ﹑ fast operation, this Di of Cheng ﹑ reliability are high, flexible in programming, its inner integrated a large amount of distributed RAM is used for realizing logical design, block RAM is for data high-speed storage, PLL is for the management of clock, embedded multiplier is for the processing of digital signal, high-speed transceiver is for data communication etc., the design of streamline simultaneously can be dwindled processing time delay, improve the maximum operation frequency that FPGA can reach, have advantages of that the chips such as DSP are incomparable.

In the Chinese patent application that is 102353954A at publication number " linear frequency modulated continuous wave automobile anti-collision radar system and using method ", what system adopted is the bi-processor architecture of DSP and FPGA, the various algorithm reasonable distribution that adopt the complicated rate average algorithm of space-time to process signal are processed to DSP and FPGA, take full advantage of the advantage separately of DSP and FPGA.But complicated because of algorithm, realize very difficultly, and be difficult to meet the specific (special) requirements of anti-collision radar system real-time and processing speed, hardware configuration complexity too simultaneously, cost is very high, is unfavorable for business promotion.

Summary of the invention

The object of the invention is for the collision avoidance radar existing in prior art the detectability in complex environment poor, real-time is not strong, the problems such as poor stability and multi-target detection ability, design a kind of automobile collision avoidance radar multiple target detection method, the triangular wave in 2 kinds of different modulating cycles of radar system periodic transmission, receive the two-way echoed signal of every kind of echo, echoed signal is carried out to windowing, distance dimension FFT and speed dimension FFT and process, by target, match the information of target of obtaining afterwards; According to different cycles triangular wave echo data, obtain the information of target, in conjunction with error function judgement, obtain final goal information.

Another object of the present invention is the automobile collision avoidance radar multiple target detection system that design realizes automobile collision avoidance radar multiple target detection method of the present invention, it controls processing module centered by adopting field programmable gate array (FPGA), adopt the radar transceiver of radio frequency transmission reception combination as radar sensor, real-time is good, processing speed is fast, has improved multiple goal judgement.

Automobile collision avoidance radar multiple target detection method of the present invention comprises the following steps:

I, launch modulated triangular wave and gather echo data

Having 2 kinds the modulation period of described modulated triangular wave, is T ₁and T ₂.At modulated triangular wave T ₁with T ₂between leave interval time, do not produce modulation signal, cycle T ₁triangular wave produce after interval time be t ₁, cycle T ₂triangular wave after interval time be t ₂.Described T ₂< T ₁, T ₁> t ₁>=T ₁/ 2, T ₂> t ₂>=T ₂/ 2.

In cycle T ₁triangular wave produce after interval time t ₁, then produce cycle T ₂triangular wave, interval time t ₂afterwards, enter next circulation, produce again T ₁the triangular wave in cycle ... so move in circles.Leave system time enough for the interval time between the modulated triangular wave in 2 kinds of cycles and carry out signal modulate emission, be also convenient to follow-up signal and process, reduce the complexity of this method, alleviate the system burden of carrying out this method.

Modulated triangular wave signal of the present invention is different from traditional fixed cycle triangular modulation signal, is to have cycle T ₁and T ₂variable period triangular modulation signal.The cycle T of the distance of real goal and the calculating of speed and modulation signal is irrelevant, and the distance of false target is relevant with the cycle T of speed calculated value and modulation signal, according to different modulating cycle T ₁and T ₂it is different that the echoed signal of false target is calculated target range, the velocity amplitude of gained, can effectively get rid of false target, the impact that effectively solves background return in multiple target detection, interference and noise.

As transmitting T ₁the modulated triangular wave in cycle, the signal returning is echo 1, as transmitting T ₂the modulated triangular wave in cycle, the signal returning is echo 2.

The echo data gathering has I, Q two paths of signals, is respectively real part and the imaginary part of echo data.Below respectively the I of echo 1 and echo 2, Q two paths of signals are processed.Sample frequency is f _s, f _svalue meet the following conditions: meet nyquist sampling law, i.e. f _s>=2f _max, f wherein _mfor radar mean frequency frequency, f _maxfor radar mean frequency frequency maximal value, B is modulation signal bandwidth, the distance that R is target, and C is the light velocity, and T is the modulation signal cycle, and v is target velocity, f _ofor the carrier frequency centre frequency transmitting.

II, echo data is carried out to windowing process

What the windowing process of this method adopted is hamming window, and hamming window has outstanding side lobe attenuation and the main lobe width of relative narrower, with respect to rectangular window, quarter window and hanning window, reduces the impact of secondary lobe on object judgement, can effectively reduce spectral leakage.

When launching modulation period, be T ₁triangle signal time, the real part in the data of echo 1 is carried out to following processing:

The m collecting cycle echo 1 data form a matrix A ₁

$A_1=\left[\begin{array}{ccccc}{a}_{11}& \&CenterDot;& \&CenterDot;& \&CenterDot;& a_{{\ln }_1}\\ \&CenterDot;& \&CenterDot;& & & \&CenterDot;\\ \&CenterDot;& & \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& & & \&CenterDot;& \&CenterDot;\\ a_{m1}& \&CenterDot;& \&CenterDot;& \&CenterDot;& a_{{\mathrm{mn}}_1}\end{array}\right]=\left[\begin{array}{c}{a}_1\left(n_1\right)\\ a_2\left(n_1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ a_m\left(n_1\right)\end{array}\right],$

In formula, m represents the number of cycles of collected echo 1, the integer that m is 8～16; n ₁be illustrated in the echo 1 in one of them cycle with sample frequency f _sthe sampling number of sampling.A(n ₁) represent data that echo 1 is carried out after analog to digital conversion.

Identical with the windowing process of aforementioned real part to the windowing process of the imaginary part in the data of echo 1.

When launching modulation period, be T ₂triangle signal time, identical with the aforementioned windowing process to echo 1 to the windowing process of the real part in the data of echo 2 and imaginary part.The m collecting cycle echo 2 data form a matrix A ₂.

The FFT(Fast Fourier Transform (FFT) of III, distance dimension)

The FFT of described distance dimension refers to T ₁or T ₂the echo 1 in the m that the modulated triangular wave in cycle a collects cycle or echo 2 carry out after the windowing process of step II, carry out the corresponding target that obtains after the FFT for the first time frequency spectrum in distance dimension, the corresponding range unit of each root spectral line.

This method one-period equates with one-period sampling number apart from counting of FFT of dimension, for echo 1, apart from tieing up FFT, counts as T ₁* f _s, for echo 2, distance dimension FFT counts as T ₂* f _s.

The echo 1 windowing process data in first cycle are carried out to distance dimension FFT to be obtained:

X ₁(m ₁)=FFT[a ₁(n ₁)·w(n ₁)]，

W (n in formula ₁) be the window function of hamming window.

Successively echo 1 Data duplication in m cycle is carried out forming a matrix B after above-mentioned processing ₁,

$B_1=\left[\begin{array}{c}{X}_1\left(m_1\right)\\ X_2\left(m_1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_m\left(m_1\right)\end{array}\right].$

Processing to the distance dimension processing of FFT of the imaginary part in the data of echo 1 and the distance of aforementioned real part dimension FFT is identical.

The FFT of the distance dimension of echo 2 is identical with echo 1, obtains matrix B ₂.

IV, speed dimension FFT

Described speed dimension FFT refers to T ₁or T ₂the echo 1 in the m that the modulated triangular wave in cycle a collects cycle or echo 2 carry out after the processing of step III, the data on same range unit are carried out to the FFT for the second time of 16, i.e. speed dimension FFT.

The integer that in this method, m is 8～16.When m ≠ 16, m be except 16 8 to 16 between certain integer time, this step IV is first processed the data in m cycle, after the data of same range unit, mends 16-m zero, mends to 16, then carries out the speed dimension FFT of 16.

To step III gained matrix B ₁each columns carries out the FFT that FFT is speed dimension, $X\left(a\right)=\mathrm{FFT}\left(B_1^T\right).$

Processing to the speed dimension processing of FFT of the imaginary part in the data of echo 1 and the speed of aforementioned real part dimension FFT is identical.

Finally the echo 1 real part data after above-mentioned processing and imaginary part data are carried out to the frequency spectrum that delivery processing output obtains echo 1.

Speed dimension FFT processing to echo 2 is identical with echo 1 with delivery processing, obtains the frequency spectrum of echo 2.

Because the echo data of radar sampling is except comprising target echo, also have background return, interference and noise, so the frequency spectrum of step gained includes the frequency of target and clutter noise.Need the processing of following target pairing, could effectively extract target echo.

V, target pairing

Target spectral line has three important features, i.e. the fuzzy speed channels of spectral line range value ﹑ and range gate difference.Suppose to have N target, the spectral line peak value size of its modulation signal ascent stage, descending branch is respectively S _u1s _u2s _u3s _uN, S _d1s _d2s _d3s _dN.Fuzzy speed channels is respectively, δ _u1δ _u2δ _u3δ _uN, δ _d1δ _d2δ _d3δ _dN.

This method adopts linear frequency modulation continuous wave system (LFMCW), and the ascent stage of triangular modulation signal, descending branch can be described as frequency sweep section, lower frequency sweep section.

The absolute value judgement spectral line range value of the peak difference up and down that this method obtains according to the upper and lower frequency sweep section of modulation signal, judges fuzzy speed channels according to the fuzzy Doppler frequency of target.The absolute value that meets upper and lower peak difference when the spectral line of the fuzzy Doppler frequency of the echo 1 of step IV gained is less than or equal to 5, judges that their spectral line range value equates, is tentatively judged as the spectral line of same target, treats follow-up judgement.When the fuzzy Doppler frequency of while target spectral line equates the fuzzy speed channels that judges them, equate, these spectral lines are the difference frequency signal that same target produces in upper and lower frequency sweep section so.

The fuzzy Doppler frequency of target spectral line is counted M divided by the remainder of 16 gained for the computing of this spectral line, and remainder scope is 0～15, and calculating the equal target spectral line of gained remainder is that fuzzy Doppler frequency equates, namely fuzzy speed channels equates.

The frequency spectrum that the frequency spectrum that same target obtains in the upper and lower frequency sweep section of modulation signal and other target obtain, in amplitude with have in shape a maximum comparability.Under Shang ﹑, the comparison process of the spectrum peak of frequency sweep section is: by the spectral line peak value of upper certain target i of frequency sweep section and lower each target of frequency sweep section spectral line peak value contrast one by one, if the absolute value of the spectral line peak difference of the spectral line peak value of upper frequency sweep section target i and lower frequency sweep section target j is minimum in all contrasts, carry out again the comparison of fuzzy Doppler frequency, if the fuzzy Doppler frequency of target i and target j equates, thinks that target i and target j are same targets.

Target pairing to the frequency spectrum of echo 2 is identical with the processing procedure of above-mentioned echo 1.

Distance and the velocity information of VI, calculating target;

To T ₁or T ₂the echo 1 that the modulated triangular wave in cycle collects or echo 2 data are calculated respectively each target range and target velocity separately after step V is processed, and wherein according to the data of echo 1, calculate the target range R of certain target i _1iwith target velocity v _1i, formula as follows:

$R_{1i}=\frac{{\mathrm{CT}}_1}{4B}(f_{b1i}^{+}+f_{b1i}^{-})$

$v_{1i}=\frac{C}{{4f}_o}(f_{b1i}^{-}-f_{b1i}^{+})$

In formula, C is the light velocity, for T ₁certain target i spectrum peak place frequency of upper frequency sweep section, for T ₁certain target i spectrum peak place frequency of lower frequency sweep section, f _ofor the carrier frequency centre frequency transmitting, B is modulation signal bandwidth.

According to the data of echo 2, calculate the R of certain target i _2iand v _2iaccount form and above-mentioned R _1iand v _1iaccount form identical.

Side circuit after the FFT for the second time that carries out m cycle data is speed dimension FFT, the computing that the obtains echoed signal spectrum peak place M that counts, the data that side circuit calculates echo 1 are calculated the target range R of certain target i _1iwith target velocity v _1iformula as follows:

$R_{1i}=\frac{M_{1i}^{+}+M_{1i}^{-}}{2}\&times;\frac{C}{32B}$

$v_{1i}=(M_{1i}^{+}-M_{1i}^{+})\frac{{\mathrm{Cf}}_s}{{32f}_0{N}_{\mathrm{FFT}1}}$

Wherein for modulation signal T ₁the computing at certain target i spectrum peak place of upper frequency sweep section is counted, for modulation signal T ₁the computing at certain target i spectrum peak place of lower frequency sweep section is counted, N _fFT1, N _fFT2be respectively echo 1 and echo 2 data are carried out to counting of distance dimension FFT, f _sfor sample frequency.

The data of side circuit calculating echo 2 are calculated the target range R of certain target i _2iwith target velocity v _2imode identical with the account form of above-mentioned echo 1.

The information of VII, error function judgement final goal

If the distance velocity composition of each target calculating respectively after processing according to the signal of echo 1 and echo 2 meets certain error range and thinks that this target range velocity composition equates, is defined as final goal information.

If R _1i, R _2ibe respectively the wherein distance value of certain target i that echo 1 and echo 2 calculate, v _1i, v _2ibe respectively echo 1 and echo 2 calculate certain target i wherein velocity amplitude.

Error function: $|R_{1i}-R_{2i}|\&le;\frac{C}{4B}(\frac{n_1}{N_{\mathrm{FFT}1}}+\frac{n_2}{N_{\mathrm{FFT}2}}),$

$|v_{1i}-v_{2i}|\&le;\frac{{\mathrm{Cf}}_S}{4f_0}(\frac{1}{N_{\mathrm{FFT}1}}+\frac{1}{N_{\mathrm{FFT}2}}),$

Wherein, B is modulation signal bandwidth, n ₁, n ₂be respectively the number of samples of echo 1 and echo 2, N _fFT1, N _fFT2be respectively echo 1 and echo 2 data are carried out to counting of distance dimension FFT, f _ofor the centre frequency transmitting, f _sfor sample frequency.

If R _1i, R _2i, v _1i, v _2imeet error function, the distance value of this final goal is (R _1i+ R _2i)/2, relative velocity is (v _1i+ v _2i)/2; If do not meet error function, think false target.

According to the automobile collision avoidance radar multiple target detection system of above-mentioned automobile collision avoidance radar multiple target detection method design, comprise radio-frequency transmissions receiving unit, data processing section and power unit, power unit is connected, provides corresponding voltage with various parts; Radio-frequency transmissions receiving unit comprises radar sensor and intermediate-freuqncy signal processing module; Data processing section comprises modulus, D/A converter module and center control processing module.It is on-site programmable gate array FPGA (Field Programmable Gate Array) that processing module is controlled at center, FPGA comprises that modulation signal produces submodule, echoed signal gathers submodule, algorithm submodule and control submodule, control submodule and connect control modulation signal generation submodule, echoed signal gathers submodule and algorithm submodule, modulation signal produces the modulation signal access modulus that submodule produces, the digital signal input end of the DA conversion chip of D/A converter module, digital signal is converted to simulating signal, the analog signal output of DA conversion chip connects the input end that transmits of radar sensor, through radar sensor, launch.Radar sensor receives real part I and the imaginary part Q two-way output terminal access intermediate-freuqncy signal processing module of signal, the echoed signal that radar sensor receives is carried out pre-service by intermediate-freuqncy signal processing module, I, Q two paths of signals access respectively AD conversion I passage and the AD conversion Q passage of the AD conversion chip of modulus, D/A converter module afterwards, the echoed signal that is converted to digital I signal and Q signal access FPGA gathers submodule, through echoed signal, gather submodule and send into algorithm submodule, obtain the information of final goal.Algorithm submodule is furnished with data output interface, in order to be connected with host computer.The information of final goal is sent into connected host computer through data output interface.

Described radar sensor is a kind of narrow beam radar sensor, is K-wave band band VCO(voltage controlled oscillator) radar transceiver, be furnished with and independently receive sendaisle, to obtain maximum gain.Native system radar sensor used has that riches all the way send passage, there are I, Q two-way receiving cable, I road receiving cable receives the real part of echoed signal, Q road receiving cable receives the imaginary part of echoed signal, double reception channel architecture has reduced sampling rate, has also reduced the requirement to the AD conversion chip of modulus, D/A converter module.Simultaneously because retained phase information in the echoed signal receiving, so can judge according to this positional information of target.This radar sensor is furnished with plane microstrip antenna, and profile is small and exquisite.Not only energy saving at work, and be highly susceptible to being integrated in various circuit, is also easy to build in the installation environment in automobile module protection facility.

Described intermediate-freuqncy signal processing module is amplified the power of the echoed signal receiving, and reduces to send into the dynamic range of signals of data processing section.When radar sensor is surveyed remote target, gained echo power is less, and gained echo power while surveying close-in target is larger.When detection range is 150m, the power bracket of the intermediate-freuqncy signal of I/Q frequency mixer output is-94dBm～-14dBm that dynamic range is 90dB.The power of sending into the input signal of data processing section after the pre-service of intermediate-freuqncy signal processing module is at least greater than an A/D quantization level, and signal amplitude will be unified as far as possible.

Control submodule in processing module FPGA is controlled according to the PLL(phaselocked loop in FPGA in described center) the Clock management modulation signal of coordinating its connection produces the logical sequence that submodule, echoed signal gather submodule and algorithm processing module, guarantees each submodule co-ordination.

Described modulation signal produces submodule and alternately produces T by the instruction of controlling submodule ₁and T ₂two kinds of different cycles triangular modulation signals.At modulated triangular wave T ₁with T ₂between leave interval time, do not produce modulation signal, cycle T ₁triangular wave produce after interval time be t ₁, cycle T ₂triangular wave after interval time be t ₂.Described T ₂< T ₁, T ₁> t ₁>=T ₁/ 2, T ₂> t ₂>=T ₂/ 2.

Described echoed signal gather submodule by the instruction of controlling submodule with certain sampling rate digitized I, the Q two-way echoed signal after to analog to digital conversion carry out quantised samples.FPGA, when each triangle finishes modulation period, drives analog to digital conversion submodule to gather echoed signal and also deposits its quantification in storer in FPGA.

Described algoritic module is the most crucial part of native system, and this module is carried out corresponding algorithm process to echoed signal, and automobile collision avoidance radar multiple target detection method according to the present invention is to T ₁and T ₂two kinds of different cycles triangular modulation signals carry out distance dimension FFT, speed dimension FFT, target pairing and target information and calculate, and finally by error function judgement, obtain final goal information.

The data output interface of described algorithm submodule is USB interface, it meets the user demand of vehicle anti-collision radar system, and its interface is simple, portable, support the advantage of plug and play to simplify to a great extent again the hardware device of system, improved the portability of this radar system.

Described host computer is the microcomputer that reference driver is used, host computer can also connect vehicle alarming device and/or emergency braking apparatus, the final goal distance that host computer storage arranges and the early warning value of speed, final goal distance and the velocity amplitude when this radar system, sent to reach early warning value, and host computer starts warning device and/or emergency braking apparatus.

Compared with prior art, the advantage of automobile collision avoidance radar multiple target detection method of the present invention is: 1, for multi-target detection, from transmitted waveform improvement and Radar Signal Processing two aspects, address, design is easy to generation, strong anti-interference performance, has the radar emission waveform of good multi-target detection performance, be equipped with corresponding multiple goal decision algorithm, realize multiobject accurate detection; 2, adopt the modulated triangular wave of modified variable period and corresponding algorithm, the distance speed of the target obtaining is through the judgement of error function, effectively retain real target range and velocity amplitude, get rid of false target, be lifted at the ability that detects a plurality of moving targets under strong background clutter noise, improve and detect multiobject accuracy; 3, adopt distance dimension FFT and speed dimension FFT, feature in conjunction with linear frequency modulation continuous wave system (LFMCW) radar three angle-modulated signals symmetries, be that the caused ranging offset that is coupled of the distance speed of moving target has symmetry, utilize this feature to carry out the pairing of target, be lifted at the ability that detects a plurality of moving targets under strong background clutter noise; 4, simplified echo signal processing algorithm, made to implement the corresponding simplification of hardware system structure of this law, be easy to realize.

Compared with prior art, the advantage of automobile collision avoidance radar multiple target detection system of the present invention is: 1, adopt FPGA as the uniprocessor structure of core control processor, the inner integrated a large amount of distributed RAM of FPGA are used for realizing logical design, block RAM is for data high-speed storage, PLL is for the management of clock, embedded multiplier is for the processing of digital signal, high-speed transceiver is for data communication etc., therefore native system has the capable processing of powerful Bing Neng Li ﹑ fast operation, this Di of Cheng ﹑ reliability is high, the advantages such as flexible in programming, when being processed, radar signal uses the inner integrated RAM of FPGA, make native system real-time good, processing speed is fast, 2, adopt the radar transceiver of K-wave band band VCO as radar sensor, radio frequency sends to receive and combines, and saves hardware resource and cost, saves space, and energy consumption also decreases, 3, radar sensor has and independently receives sendaisle, can obtain maximum gain, wherein have that riches all the way send passage, receiving cable is divided into I, Q two-way, and the direction that double reception channel architecture can the detection of a target improves multiple goal judgement, 4, native system is connected with the host computer of automobile, gained objective result directly supplies reference driver, this host computer also can with vehicle alarming device and emergency braking apparatus, automatically start and report to the police and clamping device.

Accompanying drawing explanation

Fig. 1 is this automobile collision avoidance radar multiple target detection system embodiment general frame block diagram;

Fig. 2 is FPGA Central Control Module structured flowchart in Fig. 2;

Fig. 3 is this automobile collision avoidance radar multiple target detection embodiment of the method signal processing flow figure.

Embodiment

Below in conjunction with drawings and Examples, be described in further detail.

Automobile collision avoidance radar multiple target detection system embodiment

This automobile collision avoidance radar multiple target detection system embodiment overall architecture as shown in Figure 1, comprises radio-frequency transmissions receiving unit, data processing section and power unit, and power unit is connected, provides corresponding voltage with various parts; Radio-frequency transmissions receiving unit comprises radar sensor and intermediate-freuqncy signal processing module; Data processing section comprises modulus, D/A converter module and center control processing module.It is on-site programmable gate array FPGA that processing module is controlled at center, as shown in Figure 2, FPGA comprises that modulation signal produces submodule, echoed signal gathers submodule, algorithm submodule and control submodule, and the connection of control submodule is controlled modulation signal and produced submodule, echoed signal collection submodule and algorithm submodule.Modulation signal produces the modulation signal access modulus of submodule generation, the digital signal input end of the DA conversion chip of D/A converter module, and the analog signal output of DA conversion chip connects the input end that transmits of radar sensor.Radar sensor receives real part I and the imaginary part Q two-way output terminal access intermediate-freuqncy signal processing module of signal, intermediate-freuqncy signal processing module is carried out AD conversion I passage and the AD conversion Q passage that pretreated I, Q two paths of signals access respectively the AD conversion chip of modulus, D/A converter module, the echoed signal that is converted to digital I signal and Q signal access FPGA gathers submodule, through echoed signal, gather submodule and send into algorithm submodule, obtain the information of final goal.Algorithm submodule is furnished with data output interface, and this example is USB interface, through USB interface, connects host computer.

This routine radar sensor is the radar transceiver of K-wave band band VCO, is furnished with that riches all the way send passage, I, Q two-way receiving cable, and I road receiving cable receives the real part of echoed signal, and Q road receiving cable receives the imaginary part of echoed signal.This routine radar sensor is furnished with plane microstrip antenna.The working frequency range of the carrier frequency of the radar sensor emitting electromagnetic wave in this example is 24GHz.

This routine intermediate-freuqncy signal processing module comprises frequency domain dynamic compression circuit and intermediate frequency amplifier circuit, and the simulation of employing bandpass filter, U.S. MAXIM company is integrated with source filter MAX275 and realizes frequency domain dynamic compression curve.This example is used the AD820 operational amplifier chip of AD company, realizes low-power consumption, FET inputs operation amplifier.

The CycloneIIEP2C70F672C8 chip of the ALTERA company of processing module FPGA employing is controlled at Ben Li center, inner integrated a large amount of distributed RAM for realize logical design, block RAM for data high-speed storage, PLL for the management of clock, embedded multiplier for the processing of digital signal, high-speed transceiver for data communication etc., there is the advantages such as the capable processing of powerful Bing Neng Li ﹑ fast operation, this Di of Cheng ﹑ reliability are high, flexible in programming.

Control submodule in described FPGA produces the logical sequence of submodule, echoed signal collection submodule and algorithm processing module according to the modulation signal of its connection of pll clock management coordination in FPGA, guarantee each submodule co-ordination.In system, need the work clock of different frequency to drive corresponding operating circuit, the phase-locked loop pll of FPGA is processed accordingly to outside input clock.PLL provides advanced Clock management ability, for example frequency synthesis, phase shift able to programme, external clock output, programmable duty cycle, lock-in detection, programmable bandwidth, input clock spread spectrum and support high-speed-differential input and output clock signal.

The parameter of the variable period modulated triangular wave of described needs is first with matlab software, to produce with mif file and preserve, and then after Quartus II software is opened, saves as the type of hex file, and hex file now imports in the ROM of FPGA generation again.Control submodule and by address, read the parameter in ROM with certain clock period, send into modulation signal and produce submodule, modulation signal produces submodule, by this, alternately produces T ₁and T ₂two kinds of different cycles triangular modulation signals.

Described echoed signal gather submodule by the instruction of controlling submodule with 625KHz sample frequency digitized I, the Q two-way echoed signal after to analog to digital conversion carry out quantised samples.FPGA, when each triangle finishes modulation period, drives the AD conversion chip of modulus, D/A converter module to gather echoed signal and also deposits its digitizing in storer in FPGA.

This routine host computer is the microcomputer being installed in automobile, also connect vehicle alarming device and emergency braking apparatus, the final goal distance that host computer storage arranges and the early warning value of speed, final goal distance and the velocity amplitude when this radar system, sent to reach early warning value, and host computer starts warning device and emergency braking apparatus.

Described algoritic module carries out corresponding algorithm process to echoed signal, and automobile collision avoidance radar multiple target detection method according to the present invention is to T ₁and T ₂two kinds of different cycles triangular modulation signals carry out distance dimension FFT, speed dimension FFT, target pairing and target information and calculate, and finally by error function judgement, obtain final goal information, through USB interface, are sent to host computer.

Automobile collision avoidance radar multiple target detection embodiment of the method

The environment of train experiment is: the experimental system using above-mentioned automobile collision avoidance radar multiple target detection system embodiment as this method is fixed on static experiment porch, tests with two, the place ahead target carriage.Target vehicle 1 is apart from the experimental system 15m on experiment porch, stationary state, and target vehicle 2 is apart from the experimental system 60m on experiment porch, stationary state.

This routine automobile collision avoidance radar multiple target detection method comprises the following steps, and its signal processing flow block diagram as shown in Figure 3.

I, launch modulated triangular wave and gather echo data

This routine modulated triangular wave cycle T ₁for 1.6384ms, T ₂for 0.8192ms, interval time t ₁for 1ms, t ₂for 0.5ms, the carrier frequency frequency f of transmitting modulated triangular wave _ofor 24GHz, modulation signal bandwidth B is 92MHz, and radar detection target range scope is 1.6m～150m, and the velocity range of target is 0～150Km/h.

When experimental system brings into operation, the T of the continuous alternate emission system modulation of transmitting terminal of radar sensor ₁and T ₂electromagnetic wave, receiving end receives I, the Q two paths of signals of echo 1 and echo 2, sample frequency f _s=625KHz.

II, echo data is carried out to windowing process

What the windowing process of this method adopted is hamming window, when launching modulation period, is T ₁triangle signal time, the real part in the data of echo 1 is carried out to following processing:

The m collecting cycle echo 1 data form a matrix A ₁

$A_1=\left[\begin{array}{ccccc}{a}_{11}& \&CenterDot;& \&CenterDot;& \&CenterDot;& a_{{\ln }_1}\\ \&CenterDot;& \&CenterDot;& & & \&CenterDot;\\ \&CenterDot;& & \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& & & \&CenterDot;& \&CenterDot;\\ a_{m1}& \&CenterDot;& \&CenterDot;& \&CenterDot;& a_{{\mathrm{mn}}_1}\end{array}\right]=\left[\begin{array}{c}{a}_1\left(n_1\right)\\ a_2\left(n_1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ a_m\left(n_1\right)\end{array}\right],$

In formula, m represents the number of cycles of collected echo 1, and this routine m=16 gets the echo 1 in 16 cycles that collect; n ₁be illustrated in the echo 1 in one of them cycle is counted with the 625KHz sample frequency post-sampling of sampling, n in this example ₁value be 1024.A (n) represents the data that echo 1 is carried out after analog to digital conversion.

Identical with the windowing process of aforementioned real part to the windowing process of the imaginary part in the data of echo 1.

When launching modulation period, be T ₂triangle signal time, identical with the aforementioned windowing process to echo 1 to the windowing process of the real part in the data of echo 2 and imaginary part, n ₂value be 512.

The FFT of III, distance dimension

The FFT of described distance dimension refers to T ₁or T ₂the echo 1 in 16 cycles that the modulated triangular wave in cycle collects or echo 2 carry out after primary FFT, obtaining the frequency spectrum of corresponding target in distance dimension, the corresponding range unit of each root spectral line.

To modulation signal T ₁the distance dimension FFT of echo 1 data in first cycle, upper and lower frequency sweep section is respectively carried out the FFT of 512, is the FFT of 1024 altogether, only retains each 256 points of head and the tail, the data of totally 512.

X ₁(m ₁)=FFT[a ₁(n ₁)·w(n ₁)]，

W (n in formula ₁) be the window function of hamming window.

Successively echo 1 Data duplication in m=16 cycle is carried out forming a matrix B after above-mentioned processing ₁

$B_1=\left[\begin{array}{c}{X}_1\left(m_1\right)\\ X_2\left(m_1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_m\left(m_1\right)\end{array}\right].$

Processing to the distance dimension processing of FFT of the imaginary part in the data of echo 1 and the distance of aforementioned real part dimension FFT is identical.

Through FFT for the first time the data of distance dimension FFT store in the middle of the RAM of FPGA.

To T ₂the distance dimension FFT of echo 2 data of modulation signal, upper and lower frequency sweep section is respectively carried out the FFT of 256, is the FFT of 512 altogether, only retains each 128 of the head and the tail data of totally 256.Through the data of FFT for the first time, store in the middle of RAM,

In this example, the data of real part data and imaginary part minute are stored in two RAM, because real part and imaginary part are separately to carry out FFT processing.4 RAM in this example, have been shared.

IV, speed dimension FFT

Matrix B to step III gained echo 1 data through distance dimension FFT ₁each columns again carry out FFT, i.e. the FFT of speed dimension, it is 16 that speed dimension FFT counts,

The process that imaginary part in the data of echo 1 is processed is identical with real part.

These routine echo 1 real part data and imaginary part data are carried out to delivery processing, and last delivery output obtains the frequency spectrum of echo 1 data, is the 16*512 frequency spectrum data of totally 8192 points.

The data of these routine echo 2 real parts and imaginary part are carried out the speed dimension FFT of 16, then carry out delivery, obtain the frequency spectrum of echo 2 data, are the 16*256 frequency spectrum data of totally 4096 points.

In view of the susceptibility of collision avoidance radar to the time, in conjunction with the powerful concurrent operation performance of FPGA, in the processing of step III and IV, adopt a kind of frame mode of slip to improve counting yield.

Every pair of one-period data carry out once, after distance dimension FFT, result being saved as to a line, this routine m=16, and the data in 16 cycles are stored altogether 16 row after tieing up FFT, and the distance of every behavior one-period data is tieed up FFT frequency spectrum data, is equivalent to the FIFO of the degree of depth 16.The capable frequency spectrum data of m=16 is carried out to speed dimension FFT.Carry out again after distance dimension FFT, the row distance dimension FFT frequency spectrum data depositing in comes last column, and the distance dimension FFT frequency spectrum data that comes the first row formerly depositing in is deleted, former come the second row and later the data of each row move forward successively, often carry out again after distance dimension FFT, still only store m=16 row distance dimension FFT frequency spectrum data, and then the capable frequency spectrum data of current m=16 is carried out to speed dimension FFT.Each echo cycle is exported valid data, has improved data throughput.

V, target pairing

Frequency sweep Duan Yousan bar spectral line on the frequency spectrum of echo 1 data, peak value is respectively 63,48,80, and there are altogether two spectral lines in lower frequency sweep section, and peak value is respectively 47,82.Because the spectral line peak difference that the spectral line that upper frequency sweep section peak value is 63 and lower frequency sweep section peak value are 47,82 is greater than 5, the preliminary target of spectral line that upper frequency sweep section peak value is 63 is matched unsuccessfully.

On the frequency spectrum of echo 2 data, frequency sweep section has three spectral lines, and peak value is respectively 70,50,60, and three spectral lines appear in lower frequency sweep section altogether, and peak value is respectively 73,49,60.

On the frequency spectrum of echo 1 data, three frequency spectrums corresponding to target peak of frequency sweep section are counted and are respectively 100,150,600, two target spectrums of lower frequency sweep section are counted and are respectively 150,600, the spectral line that upper frequency sweep section peak value is 63 does not have the corresponding lower frequency sweep section frequency spectrum identical with it spectral line of counting, can determine that this spectral line target matches unsuccessfully, can ignore it.The upper and lower frequency sweep section frequency spectrum identical spectral line of counting is same target, therefore corresponding 2 targets of the upper and lower frequency sweep section of echo 1 data gained spectral line.

On the frequency spectrum of echo 2 data, three frequency spectrums corresponding to target peak of frequency sweep section are counted and are respectively 150,200,600, and three target spectrums of lower frequency sweep section are counted and are respectively 150,200,600.The upper and lower frequency sweep section frequency spectrum identical spectral line of counting is same target, therefore corresponding 3 targets of the upper and lower frequency sweep section of echo 2 data gained spectral line.

Distance and the velocity information of VI, calculating target;

To T ₁echo 1 data that the modulated triangular wave in cycle collects are calculated respectively each target range and target velocity separately after step V is processed, and wherein according to the data of echo 1, calculate the target range R of certain target i _1iwith target velocity v _1i, formula as follows:

$R_{1i}=\frac{M_{1i}^{+}{M}_{1i}^{-}}{2}\&times;\frac{C}{32B}$

$v_{1i}=(M_{1i}^{-}-M_{1i}^{+})\frac{{\mathrm{Cf}}_s}{{32f}_0{N}_{\mathrm{FFT}1}}$

Wherein for T ₁certain target i spectrum peak place's computing of upper frequency sweep section is counted, for T ₁certain target i spectrum peak place's computing of lower frequency sweep section is counted, N _fFT1, N _fFT2be respectively echo 1 and echo 2 data are carried out to counting of distance dimension FFT, f _sfor sample frequency.

The upper and lower frequency sweep section frequency spectrum identical spectral line of counting is same target, bring into above-mentioned apart from speed computing formula,

$\frac{C}{32B}=\frac{3*{10}^8}{32*92*{10}^6}\&ap;0.1,$

$\frac{{\mathrm{Cf}}_s}{{32f}_o{N}_{\mathrm{FFT}1}}=\frac{32*{10}^8*625*{10}^3}{32*24*{10}^9*1024}\&ap;0.23,$

$\frac{{\mathrm{Cf}}_s}{{32f}_o{N}_{\mathrm{FFT}2}}=\frac{3*{10}^8*625*{10}^3}{32*24*{10}^9*512}\&ap;0.46,$

$R_{11}=\frac{150+150}{2}*0.1=15m$

v ₁₁=（150-150）*0.23=0m/s

$R_{12}=\frac{600+600}{2}*0.1=60m$

v ₁₂=（600-600）*0.23=0m/s

$R_{21}=\frac{150+150}{2}*0.1=15m$

v ₂₁=（150-150）*0.46=0m/s

$R_{22}=\frac{200+200}{2}*0.1=20m$

v ₂₂=（200-200）*0.46=0m/s

$R_{23}=\frac{600+600}{2}*00.1=60m$

v ₂₃=（600-600）*0.46=0m/s

The information of VII, error function judgement final goal

Step VI gained target information substitution error function:

$|R_{1i}-R_{2i}|\&le;\frac{C}{4B}(\frac{n_1}{N_{\mathrm{FFT}1}}+\frac{n_2}{n_{\mathrm{FFT}2}}),$

$|v_{1i}-v_{2i}|\&le;\frac{{\mathrm{Cf}}_S}{{4f}_0}(\frac{1}{N_{\mathrm{FFT}1}}+\frac{1}{N_{\mathrm{FFT}2}})$

Wherein, modulation signal bandwidth B=92MHz,

The sampling number n of echo 1 ₁=1024

The sampling number n of echo 2 ₂=512

Echo 1 data are carried out the points N of distance dimension FFT _fFT1=1024

Echo 2 data are carried out the points N of distance dimension FFT _fFT2=512

The centre frequency f transmitting ₀=24GHz

Sample frequency f _s=625KHz.

R ₁₁, R ₂₁, v ₁₁, v ₂₁meet error function, the distance R of final goal I and this radar system _i=(R ₁₁+ R ₂₁)/2=(15+15)/2=15m, the relative velocity v of final goal I and this radar system _i=(v ₁₁+ v ₂₁)/2=0m/s.

R ₁₂, R ₂₃, v ₁₂, v ₂₃meet error function, the distance R of final goal II and this radar system _iI=(60+60)/2=60m, the relative velocity v of final goal II and this radar system _iI=0m/s.

The R that echo 2 obtains ₂₂, v ₂₂not meeting error function, is false target.

This routine experimental result is: find altogether 2 targets, distance is 15m, 60m, and speed is 0m/s.Conform to actual.

Comparative example

Adopt identical experiment condition with above-mentioned automobile collision avoidance radar multiple target detection embodiment of the method, by traditional monocycle modulated triangular wave automobile collision avoidance radar detection system as a comparison system be fixed on static experiment porch, test with two, the place ahead target carriage.Target vehicle 1 is apart from the comparison system 15m on experiment porch, stationary state, and target vehicle 2 is apart from the comparison system 60m on experiment porch, stationary state.This comparative example experimental result is: find altogether 3 targets, distance is 15m, 20m, and 60m, speed is 0m/s, apart from the false target of 20m, fails to remove.

This automobile collision avoidance radar multiple target detection embodiment of the method is tested under actual environment, by the present embodiment and comparative example, can find out that this method gets rid of false target effectively than traditional use monocycle modulated triangular wave automobile collision avoidance radar detection method, retain real goal, resulting result is more accurate.

Above-described embodiment, is only the specific case that object of the present invention, technical scheme and beneficial effect are further described, and the present invention is not defined in this.All any modifications of making, be equal to replacement, improvement etc., within being all included in protection scope of the present invention within scope of disclosure of the present invention.

Claims (10)

1. an automobile collision avoidance radar multiple target detection method, is characterized in that comprising the following steps:

I, launch modulated triangular wave and gather echo data

Having 2 kinds the modulation period of described modulated triangular wave, is T ₁and T ₂; At modulated triangular wave T ₁with T ₂between leave interval time, cycle T ₁triangular wave produce after interval time be t ₁, cycle T ₂triangular wave after interval time be t ₂;

In cycle T ₁triangular wave produce after interval time t ₁, then produce cycle T ₂triangular wave, interval time t ₂afterwards, enter next circulation, produce again T ₁the triangular wave in cycle ... so move in circles;

As transmitting T ₁the modulated triangular wave in cycle, the signal returning is echo 1, as transmitting T ₂the modulated triangular wave in cycle, the signal returning is echo 2;

The echo data gathering has I, Q two paths of signals, is respectively real part and the imaginary part of echo data; Below respectively the I of echo 1 and echo 2, Q two paths of signals are processed, sample frequency is f _s;

II, echo data is carried out to windowing process

When launching modulation period, be T ₁triangle signal time, the real part in the data of echo 1 is carried out to following processing:

The m collecting cycle echo 1 data form a matrix A ₁

In formula, m represents the number of cycles of collected echo 1, n ₁be illustrated in the echo 1 in one of them cycle with sample frequency f _ssampling gained sampling number, n ₁=f _s* T ₁, a (n ₁) represent data that echo 1 is carried out after analog to digital conversion,

Identical with the windowing process of aforementioned real part to the windowing process of the imaginary part in the data of echo 1;

When launching modulation period, be T ₂triangle signal time, identical with the aforementioned windowing process to echo 1 to the windowing process of the real part in the data of echo 2 and imaginary part, n ₂be illustrated in the echo 1 in one of them cycle with sample frequency f _ssampling gained sampling number, n ₂=f _s* T ₂;

The FFT of III, distance dimension

The FFT of described distance dimension refers to T ₁or T ₂the echo 1 in the m that the modulated triangular wave in cycle a collects cycle or echo 2 carry out after the windowing process of step II, carry out the corresponding target that obtains after the FFT for the first time frequency spectrum in distance dimension, the corresponding range unit of each root spectral line;

Counting of FFT of distance dimension equates with one-period sampling number, for echo 1, apart from dimension, counts as T ₁* f _s, for echo 2, distance dimension is counted as T ₂* f _s;

The echo 1 windowing process data in first cycle are carried out to distance dimension FFT to be obtained:

X ₁(m ₁)＝FFT[a ₁(n ₁)·w(n ₁)]，

W (n in formula ₁) be the window function of hamming window;

Successively echo 1 Data duplication in m cycle is carried out forming a matrix B after above-mentioned processing ₁,

$B_1=\left[\begin{array}{c}{X}_1\left(m_1\right)\\ X_2\left(m_1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_m\left(m_1\right)\end{array}\right];$

Processing to the distance dimension processing of FFT of the imaginary part in the data of echo 1 and the distance of real part dimension FFT is identical;

The FFT of the real part in echo 2 data and the distance of imaginary part dimension is identical with echo 1, obtains matrix B ₂;

IV, speed dimension FFT

Described speed dimension FFT refers to T ₁or T ₂the echo 1 in the m that the modulated triangular wave in cycle a collects cycle or echo 2 carry out after the processing of step III, the data on same range unit are carried out to the FFT for the second time of 16;

To step III gained matrix B ₁each columns carries out the FFT that FFT is speed dimension, $X\left(a\right)=\mathrm{FFT}\left(B_1^T\right),$

Processing to the speed dimension processing of FFT of the imaginary part in the data of echo 1 and the speed of real part dimension FFT is identical,

Finally the real part data after above-mentioned processing and imaginary part data are carried out to the frequency spectrum that delivery processing output obtains echo 1;

Speed dimension FFT processing to the real part in echo 2 data and imaginary part is identical with echo 1 with delivery processing, obtains the frequency spectrum of echo 2;

V, target pairing

The absolute value that meets upper and lower peak difference when the spectral line of the frequency spectrum of the echo 1 of step IV gained is less than or equal to 5, judges that their spectral line range value equates, is tentatively judged as the spectral line of same target, treats follow-up judgement; When the fuzzy Doppler frequency of while target spectral line equates the fuzzy speed channels that judges them, equate, these spectral lines are the difference frequency signal that same target produces in upper and lower frequency sweep section so;

Target pairing to the frequency spectrum of echo 2 is identical with the processing procedure of echo 1;

Distance and the velocity information of VI, calculating target;

To T ₁or T ₂the echo 1 that the modulated triangular wave in cycle collects or echo 2 data are calculated respectively each target range and target velocity separately after step V is processed, and wherein according to the data of echo 1, calculate the target range R of certain target i _1iwith target velocity v _1i, formula as follows:

$R_{1i}=\frac{{\mathrm{CT}}_1}{4B}(f_{b1i}^{+}+f_{b1i}^{-})$

$v_{1i}=\frac{C}{4f_0}(f_{b1i}^{-}-f_{b1i}^{+})$

In formula, C is the light velocity, for T ₁certain target i spectrum peak place frequency of upper frequency sweep section, for T ₁certain target i spectrum peak place frequency of lower frequency sweep section, f _ofor the carrier frequency centre frequency transmitting, B is modulation signal bandwidth;

According to the data of echo 2, calculate the R of certain target i _2iand v _2iaccount form and above-mentioned R _1iand v _1iaccount form identical;

The information of VII, error function judgement final goal

If the distance velocity composition of each target calculating respectively after processing according to the signal of echo 1 and echo 2 meets certain error range and thinks that this target range velocity composition equates, is defined as final goal information;

If R _1i, R _2ibe respectively the wherein distance value of certain target i that echo 1 and echo 2 calculate, v _1i, v _2ibe respectively echo 1 and echo 2 calculate certain target i wherein velocity amplitude;

Error function: $|R_{1i}-R_{2i}|\&le;\frac{C}{4B}(\frac{n_1}{N_{\mathrm{FFT}1}}+\frac{n_2}{N_{\mathrm{FFT}2}}),$

$|v_{1i}-v_{2i}|\&le;\frac{C{f}_S}{4f_0}(\frac{1}{N_{\mathrm{FFT}1}}+\frac{1}{N_{\mathrm{FFT}2}}),$

Wherein, N _fFT1, N _fFT2be respectively echo 1 and echo 2 data are carried out to counting of distance dimension FFT, f _sfor sample frequency; B and f ₀with identical in step VI;

If R _1i, R _2i, v _1i, v _2imeet error function, the distance value of this final goal is (R _1i+ R _2i)/2, relative velocity is (v _1i+ v _2i)/2; If do not meet error function, think false target.

2. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

T in described step I ₂< T ₁, T ₁> t ₁>=T ₁/ 2, T ₂> t ₂>=T ₂/ 2.

3. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

Sample frequency f in described step I _svalue meet the following conditions:

F _s>=2f _max, f wherein _mfor radar mean frequency frequency, f _maxfor radar mean frequency frequency maximal value, B is modulation signal bandwidth, the distance that R is target, and C is the light velocity, and T is the modulation signal cycle, and v is target velocity, f _ofor the carrier frequency centre frequency transmitting.

4. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

The integer that in described step II, m is 8～16; When m ≠ 16, in step IV, first the data in m cycle are processed, after the data of same range unit, mend 16-m individual zero, then carry out the FFT of the speed dimension of 16.

5. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

In processing in described step III and IV, adopt the frame mode sliding, every pair of one-period data are carried out once after distance dimension FFT, result is saved as to a line, after m distance dimension FFT, store altogether m capable, the distance dimension FFT frequency spectrum data of every behavior one-period data, carries out speed dimension FFT to the capable frequency spectrum data of m; Carry out again after distance dimension FFT, the row distance dimension FFT frequency spectrum data depositing in comes last column, and the distance dimension FFT frequency spectrum data that comes the first row formerly depositing in is deleted, former come the second row and later the data of each row move forward successively, and then the capable frequency spectrum data of current m is carried out to speed dimension FFT.

6. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

Under described step V Zhong Shang ﹑, the comparison process of the spectrum peak of frequency sweep section is: by the spectral line peak value of upper certain target i of frequency sweep section and lower each target of frequency sweep section spectral line peak value contrast one by one, if the absolute value of the spectral line peak difference of the spectral line peak value of upper frequency sweep section target i and lower frequency sweep section target j is minimum in all contrasts, carry out again the comparison of fuzzy Doppler frequency, if the fuzzy Doppler frequency of target i and target j equates, thinks that target i and target j are same targets.

7. automobile collision avoidance radar multiple target detection method according to claim 1, is characterized in that:

In described step VI, the data of side circuit calculating echo 1 are calculated the target range R of certain target i _1iwith target velocity v _1iformula as follows:

$R_{1i}=\frac{M_{1i}^{+}+M_{1i}^{-}}{2}\&times;\frac{C}{32B}$

$v_{1i}=(M_{1i}^{-}-M_{1i}^{+})\frac{{\mathrm{Cf}}_s}{32f_0{N}_{\mathrm{FFT}1}}$

Wherein for T ₁the computing at certain target i spectrum peak place of upper frequency sweep section is counted, for T ₁the computing at certain target i spectrum peak place of lower frequency sweep section is counted, N _fFT1, N _fFT2be respectively echo 1 and echo 2 data are carried out to counting of distance dimension FFT, f _sfor sample frequency;

The data of side circuit calculating echo 2 are calculated the target range R of certain target i _2iwith target velocity v _2imode identical with the account form of above-mentioned echo 1.

8. according to the automobile collision avoidance radar multiple target detection system of the automobile collision avoidance radar multiple target detection method design described in any one in claim 1 to 7, comprise radio-frequency transmissions receiving unit, data processing section and power unit, power unit is connected, provides corresponding voltage with various parts; It is characterized in that:

Described radio-frequency transmissions receiving unit comprises radar sensor and intermediate-freuqncy signal processing module, described data processing section comprises modulus, D/A converter module and center control processing module, it is FPGA that processing module is controlled at center, FPGA comprises that modulation signal produces submodule, echoed signal gathers submodule, algorithm submodule and control submodule, control submodule and connect control modulation signal generation submodule, echoed signal gathers submodule and algorithm submodule, modulation signal produces the modulation signal access modulus that submodule produces, the digital signal input end of the DA conversion chip of D/A converter module, digital signal is converted to simulating signal, the analog signal output of DA conversion chip connects the input end that transmits of radar sensor, through radar sensor, launch, radar sensor receives real part I and the imaginary part Q two-way output terminal access intermediate-freuqncy signal processing module of signal, the echoed signal that radar sensor receives is carried out pre-service by intermediate-freuqncy signal processing module, I, Q two paths of signals access respectively AD conversion I passage and the AD conversion Q passage of the AD conversion chip of modulus, D/A converter module afterwards, the echoed signal that is converted to digital I signal and Q signal access FPGA gathers submodule, through echoed signal, gather submodule and send into algorithm submodule, obtain the information of final goal, algorithm submodule is furnished with data output interface, is connected with host computer.

9. automobile collision avoidance radar multiple target detection system according to claim 8, is characterized in that:

Described radar sensor is the radar transceiver of K-wave band band voltage controlled oscillator, is furnished with that riches all the way and send passage and I, Q two-way receiving cable.

10. automobile collision avoidance radar multiple target detection system according to claim 8, is characterized in that:

The control submodule that described center is controlled in processing module FPGA produces the logical sequence of submodule, echoed signal collection submodule and algorithm processing module according to the modulation signal of its connection of phase-locked loop clock management coordination in FPGA, guarantee each submodule co-ordination;

Described modulation signal produces submodule and alternately produces T by the instruction of controlling submodule ₁and T ₂two kinds of different cycles triangular modulation signals, at modulated triangular wave T ₁with T ₂between leave interval time, cycle T ₁triangular wave produce after interval time be t ₁, cycle T ₂triangular wave after interval time be t ₂, described T ₂< T ₁, T ₁> t ₁>=T ₁/ 2, T ₂> t ₂>=T ₂/ 2;

Described echoed signal gather submodule by the instruction of controlling submodule with certain sampling rate digitized I, the Q two-way echoed signal after to analog to digital conversion carry out quantised samples; FPGA, when each triangle finishes modulation period, drives analog to digital conversion submodule to gather echoed signal and also deposits its quantification in storer in FPGA;

Described algoritic module is that native system carries out corresponding algorithm process to echoed signal, to T ₁and T ₂two kinds of different cycles triangular modulation signals carry out distance dimension FFT, speed dimension FFT, target pairing and target information and calculate, and finally by error function judgement, obtain final goal information.