CN107783117A - Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method - Google Patents
Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method Download PDFInfo
- Publication number
- CN107783117A CN107783117A CN201610724241.XA CN201610724241A CN107783117A CN 107783117 A CN107783117 A CN 107783117A CN 201610724241 A CN201610724241 A CN 201610724241A CN 107783117 A CN107783117 A CN 107783117A
- Authority
- CN
- China
- Prior art keywords
- data
- passage
- point
- frequency sweep
- thresholding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/93—Radar or analogous systems specially adapted for specific applications for anti-collision purposes
- G01S13/931—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of land vehicles
Abstract
A kind of pilotless automobile anticollision MMW RADAR SIGNAL USING processing method, belongs to field of signal processing, and the collision between barrier easily occurs during in order to solve the problems, such as pilotless automobile traveling, and technical essential is:S1.AD data acquisitions;S2. direct current is removed;S3. window function is handled;S4.FFT is converted;S5. Threshold detection;S6. binary detection;S7. one kind in computing speed, distance or angle or combination.
Description
Technical field
The invention belongs to field of signal processing, is related to a kind of pilotless automobile anticollision MMW RADAR SIGNAL USING processing side
Method.
Background technology
In recent years, as expanding economy, transport need increasingly increase, urban traffic blocking, traffic accident take place frequently etc. into
The common issue faced for our times various countries.Analysis to road traffic accident is shown, in three driver, automobile, road rings
In section, driver is the most weak link of reliability, therefore in recent years, the pilotless automobile for substituting driver driving breed and
Raw, autonomous driving vehicle is also known as pilotless automobile, computer driving is that one kind realizes unpiloted by computer system
Intelligent automobile.
To improve the security of autonomous driving vehicle traveling, autonomous driving vehicle calculates by artificial intelligence, vision, thunder
Reach, supervising device and global positioning system cooperative cooperating, allow computer can be under the operation of nobody class active, automatic peace
Motor vehicles are operated entirely.Therefore autonomous driving vehicle is needed to judge running car situation, and the security of vehicle is predicted,
The automatic generation for taking measures to try to forestall traffic accidents, reduce the system of contingency occurrence probability, such as deviation system, forward direction vehicle
Collision-warning system, forward direction avoidance accessory system, driver attention's monitoring etc..Wherein, automobile collision avoidance radar is automatic Pilot
One of most important sensor of automobile.It is a kind of active safety equipment mainly due to automobile collision avoidance radar, can accurately surveys
The information such as the azimuth where the speed and distance, and target of surrounding objects are measured, can accurately find unmanned vapour
The potential danger of car in the process of moving, and the obstacle information arrived according to detections of radar, take measures to eliminate danger automatically.
The distance-finding method being applied at present on automobile mainly has laser ranging, ultrasonic ranging, infrared distance measuring, millimeter wave
The several methods such as radar range finding.It is infrared, the first-class optical technology of shooting is cheap and technology is simple, but all weather operations effect
Bad, crashworthiness is limited;Ultrasonic wave is influenceed greatly by state of weather, and detection range is shorter.And millimetre-wave radar overcome it is above-mentioned
The shortcomings that several detection modes, there is stable detection performance and good environment applicability.It not only has frequency height, wavelength
Short, bandwidth, small volume, it is in light weight the features such as, and compared with above-mentioned several sensors, millimetre-wave radar penetrating fog, cigarette, ash
The ability of dirt is strong, strong antijamming capability, is not influenceed by light, and detection range is remote, has the characteristics that round-the-clock round-the-clock.Cost
It has been declined that, and the overall dimensions of radar can be made very small, and be easy to install on automobile, be pretended to be automatic both at home and abroad at present
The general choice mode of driving Anticollision Radar.
In summary:No matter for security standpoint or economic angle, the development of autonomous driving vehicle Anticollision Radar is all
Great application value and realistic meaning.
The content of the invention
The collision between barrier easily occurs during in order to solve the problems, such as pilotless automobile, the present invention proposes one kind
Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method, to carry out speed, distance, azimuth resolving to barrier, from
And barrier is driven a vehicle and detected, collision free.
In order to solve the above-mentioned technical problem, the technical scheme is that:
A kind of pilotless automobile anticollision MMW RADAR SIGNAL USING processing method, comprises the following steps:
S1.AD data acquisitions;
S2. direct current is removed;
S3. window function is handled;
S4.FFT is converted;
S5. Threshold detection;
S6. binary detection;
S7. one kind in computing speed, distance or angle or combination.
Beneficial effect:
The present invention gives one kind and realizes pilotless automobile anticollision millimetre-wave radar based on linear frequency modulation triangular wave first
The Waveform Design of system;
The present invention provides the pilotless automobile anticollision millimetre-wave radar high-performance letter realized based on linear frequency modulation triangular wave
Number processing method, this method can realize the detection of the relative distance and relative velocity to front obstacle, while can be real
The detection function at existing target direction angle.As a result of more signal processing methods, CAS can be caused, can be exported more
Prepare more stable target information, more accurately object judgement is made for pilotless automobile anticollision.
Brief description of the drawings
Frequency variation diagrams of Fig. 1 linear frequency modulation triangular wave FMCW in a frequency sweep cycle;
Fig. 2 pilotless automobile short distance CAS signal processing flow figures.
Embodiment
Embodiment 1:A kind of pilotless automobile anticollision MMW RADAR SIGNAL USING processing method, comprises the following steps:
S1.AD data acquisitions;
S2. direct current is removed;
S3. window function is handled;
S4.FFT is converted;
S5. Threshold detection;
S6. binary detection;
S7. one kind in computing speed, distance or angle or combination.
Wherein:
The specific method of the step S1 is:
(1) by the continuous I/Q data in passage 1 and passage 2, processing is digitized by AD samplings;
(2) data collected in passage 1 and passage 2 are divided into the upper frequency sweep data of triangular wave and lower frequency sweep data, gone
Except direct current is removed after forward part data point, the FFT of time-frequency is carried out, time domain data is converted into frequency data;Before the removal
Partial data point, exactly in the data that AD is collected, the forward part data point that AD is collected first is got rid of, typically 50~70
It is individual, such as, if collecting 700 points, preceding 50 points are got rid of, the data from 51 to 700 remove direct current and carry out FFT changes
Change.Why to get rid of this partial dot has two reasons, first, inside these data, partial data is due to that waveform is changing
When, pulse caused by voltage, cause the reason for this partial data is abnormal, and second reason is due to range ambiguity.This
Part is said before the reason for causing range resolution ratio to reduce, and is the linearity of transmitted waveform in fact, is caused this resolution
Rate reduces.
The specific method of the step S2 is:
(1) average of the upper and lower frequency sweep I/Q data of respective passage triangular wave in passage 1 and passage 2, will be calculated respectively;
(2) the upper and lower frequency sweep IQ of respective passage triangular wave each data are cut into the average that previous step is calculated.
The specific method of the step S3 is:By in passage 1 and passage 2, the upper and lower frequency sweep section of triangular wave each removes direct current
Time domain data afterwards carries out windowing process, selects Hanning window and/or hamming window.
The specific method of the step S4 is:By in passage 1 and passage 2, the upper and lower frequency sweep hop count of the triangular wave after adding window
According to FFT is carried out, time domain data is converted into frequency data.
The specific method of the step S5 is:
(1) by the triangle in the plural modulus value and passage 2 of each point after frequency sweep FFT on the triangular wave in passage 1
Plural modulus value in corresponding points on ripple after frequency sweep FFT, is averaging processing, by frequency sweep FFT under the triangular wave in passage 1
The plural modulus value in corresponding points under triangular wave in the plural modulus value and passage 2 of each point after conversion after frequency sweep FFT,
It is averaging processing;
(2) data after will be average, carry out CFAR Threshold detections.CFAR Threshold detections selecting unit averagely selects small thresholding
Detection method SOCA-CFAR, SOCA-CFAR are that cell-average selects small CFAR to handle, and idiographic flow is as follows:
1) the long L of reference window is set, and its value can survey according to outfield to be changed, and is chosen for 15~20 points, protection location selection 2
~3 points;
2) some modulus value point after single sawtooth period data FFT is directed to, calculates L data in its preceding reference window respectively
Average β 1 and rear reference window in L data average β2If its front or rear window length is less than L, actual window length is taken to calculate average;
3) window average β before and after the point is compared1And β2, selection wherein smaller is as the estimation of its level α, i.e. α=min (β1,
β2);
4) thresholding Product-factor γ, the then point detection threshold T=α * γ are set;Thresholding Product-factor γ can be according to multiple
Field testing result adjusts accordingly;
5) compare the size of the modulus value and its threshold value, if its modulus value is more than thresholding, record the positional information of the point,
Otherwise it is assumed that it does not cross thresholding;
6) for other all modulus value points, above step 2~5 is performed respectively, i.e., carries out sliding window detection, note for all points
Record all positional informations for crossing threshold point.
The specific method of the step S6 is:
To the data after CFAR Threshold detections, it is a range cell to make each data, to each range cell
Data carry out binary detection, if the data of the range cell cross thresholding, are designated as 1, if not having thresholding, are designated as 0,
Then the progress multicycle adds up, if the number of the thresholding of some range cell accumulative 1 exports the point coordinates more than K
Value, otherwise exported not as the target for crossing thresholding, wherein K represents accumulative 1 number;
After binary detection, when the points for crossing thresholding for meeting requirement simultaneously are not unique, only output is selected to move into one's husband's household upon marriage
First peak point of limit, mainly consider maximum to unmanned degree of danger to be nearest apart from pilotless automobile
Object, so being not to look for all maximal peak points for crossing thresholding, but select first peak value for crossing thresholding.
The step S6, after by CFAR detections and binary detection, thresholding is crossed for upper frequency sweep and lower frequency sweep section
Point carry out pairing processing, if frequency sweep crosses the point coordinates value of thresholding and differs by more than threshold value up and down, it is impossible to be defined as same mesh
Above and below target during frequency sweep, handled without pairing.As a kind of scheme, the threshold value is arranged to the seat that upper and lower frequency sweep crosses the point of thresholding
Scale value differ 25 points and more than.
In the step S7, the calculation method for speed, distance and angle is:
(1) by peak point that is after successful matching or need not being matched, its corresponding frequency values is calculated, if in passage 1
Upper first peak coordinate for crossing threshold point of frequency sweep section is p1_up, then frequency values corresponding to the point are f1_up, and corresponding FFT becomes
Data after changing are a_p1_up+1j*b_p1_up, and phase isUpper frequency sweep section is right in passage 2
Data after the FFT answered are a_p2_up+1j*b_p2_up, phaseIf passage
Lower first peak coordinate for crossing threshold point of frequency sweep section is p1_down in 1, then frequency values corresponding to the point are f1_down;
Wherein:A represents the data value on I roads, and b represents the data value on Q roads, and a_p1 is represented in the array of a+j*b compositions, mistake
Coordinate corresponding to the peak point of thresholding is p1, and b_p1 is represented in the array of a+j*b compositions, crosses and is sat corresponding to the peak point of thresholding
It is designated as p1;
(2) frequency values f1_down corresponding to swept frequency value f1_up and lower frequency sweep will be gone up in obtained passage 1, according to public affairs
FormulaCalculate to the distance of obstacle target before pilotless automobile, wherein, T is triangle wave period, T
=20ms, B are modulating bandwidth, and B=200MHz, c are the light velocity, c=3.0 × 108;According to formulaIts
Middle f0Centered on frequency calculate pilotless automobile before to obstacle target speed, f0It is centre frequency, f0=24.125GHz,
The centre frequency of radar transmitter frequency, it is from 24.015 to 24.225,24.125 be exactly centre frequency if bandwidth 200MHz.
(3) in passage 1 and passage 2, according to the respective phase that above frequency sweep is calculated respectively
WithFurther according to calculation formula
Phase difference ψ is calculated;According to formulaComputer azimuth angle, wherein, λ is wavelength, and d is antenna spacing.
Embodiment 2:Supplement of the present embodiment as embodiment 1, what the present embodiment was mainly introduced is to use millimetre-wave radar
Realize the barrier avoiding function of pilotless automobile.Because millimetre-wave radar operation wavelength is between 1mm~10mm, visited with others
Survey mode is compared, and mainly has that detection performance is stable, environment is well-adjusted, size is small, price is low, can be in the sleet of rather harsh
The advantages that weather uses.Therefore, the present invention introduces the pilotless automobile barrier avoiding function system letter based on millimetre-wave radar
The realization of number processing method.
The present embodiment is mainly to complete distance, speed and side that pilotless automobile travels front Environment Obstacles thing to it
Position measures.Mainly by using millimetre-wave radar, the ultimate range of pilotless automobile avoidance is reached for the present embodiment patent
To 200m, simultaneously because perceptual performance of the radar to environment, it is possible to achieve to the quick sensing of surrounding objects, can accurately sentence
Break before to the relative distance of risk object, relative velocity and azimuth.
The working frequency of millimetre-wave radar designed by the present embodiment is in 24GHz or 77GHz, using FMCW continuous wave bodies
System, using linear frequency modulation, its range resolution ratio is high.Waveform uses linear frequency modulation triangular wave FMCW, and being primarily due to the present embodiment will
Realize the calculating to target range and speed.Target range and speed can be realized by the upper frequency sweep and lower frequency sweep of triangular wave
Degree resolves.The maximal rate of the pilotless automobile of the present embodiment design is 250km/h, and the maximum of pilotless automobile anticollision is surveyed
Away from for 200m.
The present embodiment is mainly the design and letter for providing pilotless automobile anticollision MMW RADAR SIGNAL USING process part
Number processing method.
The radar center frequency f of the present embodiment design is 24.125GHz.Transmitted waveform selection triangular wave, cycle 20ms,
With a width of 200MHz.Transmitted waveform is as shown in Figure 1.
The present embodiment realizes the resolving to target range speed by single channel I/Q data, because the present embodiment realizes target side
The calculating of parallactic angle, so the present embodiment is by the way of double reception antenna, i.e. binary channels I/Q data, by binary channels each on
The angle measurement function of the target is realized in the calculating of frequency sweep section.
Pilotless automobile anticollision MMW RADAR SIGNAL USING process chart, as shown in Figure 2:It is as follows to implement step:
1st, AD data acquisitions are data processing
(1) by the continuous I/Q data in passage 1 and passage 2, processing is digitized by AD samplings;
(2) data collected in passage 1 and passage 2 are divided into the upper frequency sweep data of triangular wave and lower frequency sweep data, and
The good data of the linearity are chosen respectively does subsequent treatment;
2nd, direct current is removed
(1) average of the upper and lower frequency sweep I/Q data of respective passage triangular wave in passage 1 and passage 2, will be calculated respectively;
(2) the upper and lower frequency sweep IQ of respective passage triangular wave each data are cut into the average that previous step is calculated, from
And complete to go the purpose of direct current, reduce the influence that direct current component detects to target gate.
3rd, window function is handled
The time domain data that in passage 1 and passage 2, the upper and lower frequency sweep section of triangular wave is each gone after direct current is carried out at adding window
Reason, Hanning window, hamming window etc. can be selected, secondary lobe be reduced, so as to improve the detection performance of target;Hanning window can cause main lobe to add
Width is simultaneously reduced, but secondary lobe can be substantially reduced.Hamming window and Hanning window are all Cosine Windows, and simply weight coefficient is different.Hamming window
The coefficient of weighting can make secondary lobe reach smaller.
4th, FFT
By in passage 1 and passage 2, the upper and lower frequency sweep segment data of the triangular wave after adding window carries out FFT, by time domain number
According to being converted into frequency data.
5th, CFAR Threshold detections
(1) by the triangle in the plural modulus value and passage 2 of each point after frequency sweep FFT on the triangular wave in passage 1
Plural modulus value in corresponding points on ripple after frequency sweep FFT, is averaging processing, and will similarly be swept under the triangular wave in passage 1
The plural number in corresponding points under triangular wave in the plural modulus value and passage 2 of each point after frequency FFT after frequency sweep FFT
Modulus value, it is averaging processing;
(2) data after will be average, carry out CFAR Threshold detections.CFAR Threshold detections can averagely be selected small with selecting unit
Threshold detection method SO-CFAR, protection location can select 1 to 2 points, and window points can select 15~20.
6th, binary detection
To the data after CFAR Threshold detections, it is a range cell to make each data.To each range cell
Data carry out binary detection, i.e., if the data of the range cell cross thresholding, are then designated as 1, if not having thresholding, are designated as
0.Then the progress multicycle adds up, if the number of the thresholding of some range cell accumulative 1 more than K, exports point seat
Scale value, otherwise exported not as the target for crossing thresholding.
(3) after binary detection, when the points for meeting to require thresholding simultaneously are a lot, only selection exported thresholding
First peak point, mainly consider that maximum to unmanned degree of danger is the thing nearest apart from pilotless automobile
Body, so being not to look for all maximal peak points for crossing thresholding, but select first peak value for crossing thresholding.
7th, pairing is handled
Carried out by CFAR detections and binary detection, the point that thresholding is crossed for upper frequency sweep and lower frequency sweep section at pairing
Reason.If the point coordinates value difference that upper and lower frequency sweep crosses thresholding is too big, it is impossible to when being defined as the frequency sweep up and down of same target, does not enter
Row pairing is handled.
8th, speed, distance resolve
(1) by the peak point after successful matching, its corresponding frequency values is calculated, if upper frequency sweep first mistake of section in passage 1
The peak coordinate of threshold point is p1_up, then frequency values corresponding to the point are f1_up, and the data after corresponding FFT are a_p1_up+
1j*b_p1_up, phase, the data in passage 2 after point FFT corresponding to upper frequency sweep section are a_p2_up+1j*b_p2_up, phase
Position;If lower first peak coordinate for crossing threshold point of frequency sweep section is p1_down in passage 1, then frequency values corresponding to the point are f1_
down;
(2) by frequency values f1_ corresponding to upper swept frequency value f1_up in the passage 1 obtained in step 3 and lower frequency sweep
Down, according to formula, wherein, T is triangle wave period, and T=20ms, B are modulating bandwidth, and B=200MHz, c are the light velocity,;According to
Formula, wherein centered on frequency ,=24.125GHz.According to the two formula, obtain before pilotless automobile to obstacle target
Distance and speed;
9th, angle resolves
In passage 1 and passage 2, the phase that is calculated respectively according to respective upper frequency sweep and, calculate and obtained according to calculation formula
It is to phase difference.
According to formula, computer azimuth angle, wherein, d is antenna spacing.
So far, complete single detection and complete pilotless automobile anticollision millimetre-wave radar to pilotless automobile operation front
The resolving function of the information such as obstacle distance, speed and azimuth.
In order to improve the distance of solving target, speed and the accuracy of angle information, using multiple cycle data sliding window
Processing mode, i.e., the AD of each passage in the multiple cycles I/Q datas collected are averaging processing.Using multicycle sliding window
The processing method of formula can effectively improve the degree of accuracy of detection target.Using sliding window periodicity number selection mainly according to
According to target in the periodicity, premised on span does not occur from cell cases, then it can be reached in view of chip processing capabilities
To the principle of real-time.
Embodiment 3:For in above-mentioned each scheme, peak value processing, the present embodiment provides one kind and is applied to pilotless automobile
The peak value processing method of signal:
One peak point threshold factor α is set, and it is used to limit the thresholding maximal peak point excessively detected and a upper cycle
The absolute difference of the maximal peak point of appearance so that the absolute difference cannot be greater than peak point threshold factor α:
Expression formula is as follows:
|L_max(k)-L_max(k-1)|≤α;
Wherein:L_max (k) crosses thresholding maximal peak point coordinate for the k cycles, and L_max (k-1) was the maximum in a upper cycle
Peak value point coordinates, k represent the kth moment;vmaxFor pilotless automobile maximal rate, λ is millimetre-wave radar wavelength, and fs is sampling
Rate, N are FFT points;
If the k moment, cross thresholding maximal peak point and cross the absolute value differences of thresholding maximal peak point set with the k-1 moment
In the range of the peak point threshold factor α put, then it is assumed that the peak point in kth cycle is effective;If the k moment, thresholding peak-peak is crossed
Point exceedes set peak point threshold factor α, then the peak point that the k moment exports is replaced with the peak point at k-1 moment.
As the explanation of above-mentioned technological means, in a time quantum of adjacent periods, peak that current period calculates
It is worth point, the peak point with the last cycle, if in adjacent periods, speed does not change, then peak point is in adjacent periods
It inside can also keep constant, but if within the adjacent periods time, pilotless automobile speed changes, and can cause current week
Certain change occurs for peak point of the peak point of phase in a upper cycle, if target is remote pilotless automobile, currently
The points in cycle can be more than the points in a upper cycle, if target is proximate to pilotless automobile, the points meeting of current period
Less than the points in a upper cycle, the excursion of the peak point is designed peak point threshold factor α, the predictor selection
Span, depend primarily in adjacent periods, the maximal rate of pilotless automobile, i.e. formulaIts
Middle vmaxFor pilotless automobile maximal rate, λ is millimetre-wave radar wavelength, and fs is sample rate, and N is FFT points.
But if after pilotless automobile environment undergos mutation, the corresponding peak value points for crossing thresholding may also can be continuous
Generation exceeds designed threshold factor.If without amendment, after undergoing mutation, it is maximum that what each cycle detection was arrived crosses thresholding
Peak point can all exceed the threshold factor set, and each thresholding maximal peak point coordinate of crossing can all be corrected for the peak of last moment
It is worth point coordinates, i.e., the value being similarly worth before also keeping mutation, it is impossible to the value after aristogenesis.In order to improve pilotless automobile thunder
Adaptability up to table to various environment, a peak value point mutation is introduced for this and adds up factor φ.
Set peak value point mutation to add up factor φ, the definition that the peak value point mutation adds up factor φ is, if from k when
Quarter, continuous b cycle, b span are 5~10, cross thresholding maximal peak point and previous cycle to cross thresholding maximum
Peak point is compared, above threshold value threshold factor a, then kth+b the moment, the thresholding maximal peak point excessively that current time is calculated
Thresholding maximal peak point is crossed as current time.In order to ensure the real-time of tracking, it is proposed that b value is 5~10.
It must be gone out by previous step after limiting maximal peak point, in order to improve the precision of system value measurement, propose to improve and survey
Spectrum maximum estimated algorithm away from precision.
Ideally, the frequency spectrum of echo difference frequency signal only has a spectral line, but reality is in use, due to adopting
There is fence effect in sample, discrete spectrum maximum amplitude spectral line will necessarily shift spectrum peak position, so as to be calculated by peak point
The distance value gone out will have certain error with actual range.When spectral peak shifts, relative to main lobe peak value institute
Corresponding central spectral line will two kinds of situations, i.e. left avertence or right avertence.If crossed in the left and right peak value of thresholding maximum peak point,
Left side peak value is more than the right peak value, then the position where central spectral line, between maximal peak point and left side peak point, conversely,
Then between maximal peak point and the right peak point.
Because the frequency spectrum that FFT is calculated apart from general equidistantly sampling, its spectral magnitude maximum point to must continuously be located at
Have in the main lobe of its curve, in main lobe and only two sampled points.If the coordinate for crossing thresholding maximal peak point A1 is (a1, k1),
Wherein, a1 represented the value of thresholding maximal peak point, and k1 represented range value corresponding to thresholding peak point;Maximal peak point or so
Both sides, minor peaks point coordinates is A3 (a3, k3), if required center peak point A is (amax, kmax), then e=amax-a1, then
A1 points, be (a2, k1)=(a1+2e, k1) on A point symmetries point A2 coordinates, the zero point A4 of complex envelope for (a4, k1)=(a3+e,
0);
Wherein:A2, a3, a4 are the values of the thresholding maximal peak point excessively of corresponding points, and k3, k4 are the threshold peaks excessively of corresponding points
Range value corresponding to point;
A2, A3 and A4 are approximately straight line, and its linear relationship is:
OrderThen
Setting error E and deviation e are compared, if | e |<E, then the value for crossing thresholding peak point now is then required
Center peak point value, if deviation e is more than set error E,β is modifying factor, value model
Enclose for 1.5~1.9, the selection reason of the modifying factor is:Due to it is initial whenA point symmetries point A2 is sat
It is designated as (a2, k1)=(a1+2e, k1), A point transverse axis coordinate points and A2 transverse axis coordinates are on maximal peak point pair during primary condition
Claiming, i.e., A2 coordinate points are a1+2e, if deviation e is more than set error E, illustrate that A2 coordinate selection is excessive,
Be maximal peak point between a1+2e, 2 times of deviation e needs to carry out to take small, and the modification method that the present invention uses is to pass through
Change modifying factor β size so as to change l values, then carry out e continuous iteration, untill e is less than the error E of setting.
Modifying factor β value principle can be chosen according to the required E values reached, if E demand precision is not high, modifying factor
β can select 1.9 to be modified, if E demand precision is very high, it may be necessary to which successive ignition reaches requirement, then needs modifying factor
Sub- β selects a little bit smaller as far as possible, and 1.5 can be selected to be modified, quickly maximal peak point is calculated The present invention gives one
The interval range value of modifying factor, i.e. modifying factor β=1.5~1.9.Change the value that modifying factor calculates e, to be calculated
The value amax=a1+e of center peak point.
As another embodiment, in addition to step:Distance tracking:One threshold factor ε is set, and it is used to limit currently
The absolute difference for the data H (k-1) that data H (k) occurred with a upper cycle so that the absolute difference cannot be greater than the threshold value
Factor ε;
Expression formula is as follows:
| H (k)-H (k-1) |≤ε, ε span are 0.8~1.3;
If the data at k moment and the absolute value differences at k-1 moment, in the range of set threshold factor ε, then it is assumed that
The peak point in kth cycle is effective;If the k moment, data exceed set threshold factor ε, then the data k- that the k moment exports
The data at 1 moment are replaced.
The accumulative factor θ of one mutation is set, and the definition that the mutation adds up factor θ is if that, since the k moment, continuous b is individual
Cycle, data above threshold value threshold factor θ, then at the kth+b moment, will resolve at current time compared with the data in previous cycle
Data of the data gone out as current time.
As a kind of embodiment, specific in the present embodiment, for it is above-mentioned be not carried out distance tracking or perform distance with
Track, during output, for the data of single output, using the output of sliding window algorithm value;
The data at kth moment are equal to the N in sliding windowcIndividual value removes the average after maximum and minimum value, as last
Data output, its calculation formula are
Wherein
NcData are counted used by representing sliding window.
Using peak-tracking algorithm and track algorithm, it is possible to prevente effectively from the mistake due to single or multiple peak value searching
And cause once or repeatedly data calculation anomaly, such as in single peak search procedure, generation peak value saltus step, it is adjacent
Peak difference values between cycle are very big, while by the saltus step with peak value, caused by very big saltus step occurs, i.e., in the cycle,
Saltus step scope caused by peak value saltus step, has been far longer than caused by a cycle as caused by pilotless automobile speed
Distance change scope.Thus peak time tracking and tracking are it is possible to prevente effectively from exceptional value caused by this anomaly peak, so as to have
The stability of the data of the raising tracking on effect ground.
It is described above, the only preferable embodiment of the invention, but the protection domain of the invention not office
It is limited to this, any one skilled in the art is in the technical scope that the invention discloses, according to the invention
Technical scheme and its inventive concept be subject to equivalent substitution or change, should all cover within the protection domain of the invention.
Claims (10)
1. a kind of pilotless automobile anticollision MMW RADAR SIGNAL USING processing method, it is characterised in that comprise the following steps:
S1.AD data acquisitions;
S2. direct current is removed;
S3. window function is handled;
S4.FFT is converted;
S5. Threshold detection;
S6. binary detection;
S7. one kind in computing speed, distance or angle or combination.
2. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S1 specific method is:
(1)By the continuous I/Q data in passage 1 and passage 2, processing is digitized by AD samplings;
(2)The data collected in passage 1 and passage 2 are divided into the upper frequency sweep data of triangular wave and lower frequency sweep data, before removal
Direct current is removed after partial data point, carries out the FFT of time-frequency, time domain data is converted into frequency data.
3. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S2 specific method is:
(1)In passage 1 and passage 2, the average of respective passage triangular wave upper and lower frequency sweep I/Q data will be calculated respectively;
(2)The upper and lower frequency sweep IQ of respective passage triangular wave each data are cut into the average that previous step is calculated.
4. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S3 specific method is:By in passage 1 and passage 2, time domain data that the upper and lower frequency sweep section of triangular wave is each gone after direct current
Windowing process is carried out, selects Hanning window and/or hamming window.
5. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S4 specific method is:By in passage 1 and passage 2, the upper and lower frequency sweep segment data of the triangular wave after adding window carries out FFT changes
Change, time domain data is converted into frequency data.
6. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S5 specific method is:
(1)By on the triangular wave in the plural modulus value and passage 2 of each point after frequency sweep FFT on the triangular wave in passage 1
Plural modulus value in corresponding points after frequency sweep FFT, is averaging processing, by frequency sweep FFT under the triangular wave in passage 1
The plural modulus value in corresponding points under triangular wave in the plural modulus value and passage 2 of each point afterwards after frequency sweep FFT, carry out
Average treatment;
(2)Data after will be average, carry out CFAR Threshold detections.
7. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 6, it is characterised in that CFAR
Threshold detection selecting unit averagely selects small Threshold detection method, and idiographic flow is as follows:
1)The long L of reference window is set, and its value can survey according to outfield to be changed, and is chosen for 15~20 points, protection location selection 2~3
Individual point;
2)For some modulus value point after single sawtooth period data FFT, the equal of L data is calculated in its preceding reference window respectively
Value β1With the average β of L data in rear reference window2If its front or rear window length is less than L, actual window length is taken to calculate average;
3)Compare window average β before and after the point1And β2, selection wherein smaller is as the estimation of its level α, i.e. α=min (β1,β2);
4)Thresholding Product-factor γ, the then point detection threshold T=α * γ are set;
5)Compare the size of the modulus value and its threshold value, if its modulus value is more than thresholding, record the positional information of the point, otherwise
Think that it does not cross thresholding;
6)For other all modulus value points, above step 2 is performed respectively)~5), i.e., carry out sliding window detection, record for all points
All positional informations for crossing threshold point.
8. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that described
Step S6 specific method is:
To the data after CFAR Threshold detections, it is a range cell to make each data, to the data of each range cell
Binary detection is carried out, if the data of the range cell cross thresholding, is designated as 1, if not having thresholding, is designated as 0, then
The progress multicycle adds up, no if the number of the thresholding of some range cell accumulative 1 exports the point coordinates value more than K
Then exported not as the target for crossing thresholding, wherein K represents accumulative 1 number;
After binary detection, when the points for crossing thresholding for meeting requirement simultaneously are not unique, only selection exported thresholding
First peak point.
9. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 7, it is characterised in that described
Step S6, after by CFAR detections and binary detection, the point that thresholding is crossed for upper frequency sweep and lower frequency sweep section matches
Processing, if the point coordinates value that upper and lower frequency sweep crosses thresholding differs by more than threshold value, it is impossible to be defined as the frequency sweep up and down of same target
When, handled without pairing.
10. pilotless automobile anticollision MMW RADAR SIGNAL USING processing method as claimed in claim 1, it is characterised in that institute
State in step S7, the calculation method for speed, distance and angle is:
(1)By peak point that is after successful matching or need not being matched, its corresponding frequency values is calculated, if being swept in passage 1
The peak coordinate that frequency range first crosses threshold point is p1_up, then frequency values corresponding to the point are f1_up, after corresponding FFT
Data be a_p1_up+1j*b_p1_up, phase isIn passage 2 corresponding to upper frequency sweep section
Data after the FFT are a_p2_up+1j*b_p2_up, phaseIf in passage 1
Lower first peak coordinate for crossing threshold point of frequency sweep section is p1_down, then frequency values corresponding to the point are f1_down;
Wherein:Wherein:A represents the data value on I roads, and b represents the data value on Q roads, and a_p1 is represented in the array of a+j*b compositions,
It is p1 to cross coordinate corresponding to the peak point of thresholding, and b_p1 is represented in the array of a+j*b compositions, corresponding to the peak point for crossing thresholding
Coordinate is p1;
(2)Frequency values f1_down corresponding to swept frequency value f1_up and lower frequency sweep will be gone up in obtained passage 1, according to formulaTo the distance of obstacle target before calculating unmanned plane, wherein, T is triangle wave period, and B is frequency modulation band
Width, c are the light velocity, c=3.0 × 108;According to formulaWherein f0Centered on frequency calculate unmanned plane before
To the speed of obstacle target, f0It is centre frequency;
(3)In passage 1 and passage 2, according to the respective phase that above frequency sweep is calculated respectivelyWithFurther according to calculation formula
Phase difference ψ is calculated;According to formulaComputer azimuth angle, wherein, λ is wavelength, and d is antenna spacing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610724241.XA CN107783117A (en) | 2016-08-25 | 2016-08-25 | Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610724241.XA CN107783117A (en) | 2016-08-25 | 2016-08-25 | Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107783117A true CN107783117A (en) | 2018-03-09 |
Family
ID=61438616
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610724241.XA Pending CN107783117A (en) | 2016-08-25 | 2016-08-25 | Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107783117A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109658715A (en) * | 2019-01-31 | 2019-04-19 | 厦门精益远达智能科技有限公司 | Statistical method of traffic flow, device, equipment and the storage medium of multilane |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102890272A (en) * | 2012-11-05 | 2013-01-23 | 中国航天科工集团第二研究院二十三所 | Method for processing millimeter wave cloud radar signal |
CN104199034A (en) * | 2014-09-19 | 2014-12-10 | 黄国鹏 | Radar sensor based automobile collision avoidance method and device |
US20140375491A1 (en) * | 2013-06-20 | 2014-12-25 | Infineon Technologies Ag | Method, device and system for processing radar signals |
CN104808189A (en) * | 2015-05-04 | 2015-07-29 | 奇瑞汽车股份有限公司 | Millimeter-wave radar signal processing system and method |
CN105445714A (en) * | 2015-11-24 | 2016-03-30 | 大连楼兰科技股份有限公司 | Automobile forward direction anticollision system signal processing method |
CN105629211A (en) * | 2015-12-29 | 2016-06-01 | 大连楼兰科技股份有限公司 | Multi-target detection combined waveform automobile lane change auxiliary system signal processing method and automobile lane change auxiliary system |
-
2016
- 2016-08-25 CN CN201610724241.XA patent/CN107783117A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102890272A (en) * | 2012-11-05 | 2013-01-23 | 中国航天科工集团第二研究院二十三所 | Method for processing millimeter wave cloud radar signal |
US20140375491A1 (en) * | 2013-06-20 | 2014-12-25 | Infineon Technologies Ag | Method, device and system for processing radar signals |
CN104199034A (en) * | 2014-09-19 | 2014-12-10 | 黄国鹏 | Radar sensor based automobile collision avoidance method and device |
CN104808189A (en) * | 2015-05-04 | 2015-07-29 | 奇瑞汽车股份有限公司 | Millimeter-wave radar signal processing system and method |
CN105445714A (en) * | 2015-11-24 | 2016-03-30 | 大连楼兰科技股份有限公司 | Automobile forward direction anticollision system signal processing method |
CN105629211A (en) * | 2015-12-29 | 2016-06-01 | 大连楼兰科技股份有限公司 | Multi-target detection combined waveform automobile lane change auxiliary system signal processing method and automobile lane change auxiliary system |
Non-Patent Citations (3)
Title |
---|
刘炜: "基于TMS320VC5402的汽车防撞警示雷达研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
盂祥伟等: "高斯背景下距离扩展目标的恒虚警率检测", 《系统工程与电子技术》 * |
陈伯孝: "《现代雷达系统分析与设计》", 30 September 2012, 西安电子科技大学出版社 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109658715A (en) * | 2019-01-31 | 2019-04-19 | 厦门精益远达智能科技有限公司 | Statistical method of traffic flow, device, equipment and the storage medium of multilane |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8299957B2 (en) | Method for detecting a vehicle type, a vehicle speed and width of a detecting area by a vehicle radar sensor | |
US10054680B2 (en) | Radar detection of endo-clutter high-value targets using tracker feedback | |
CN110058239B (en) | Vehicle-mounted millimeter wave radar device and target detection method | |
EP3415945A1 (en) | Method of determining the yaw rate of a target vehicle | |
CN106405556A (en) | Vehicle object information detection and identification system and signal processing method thereof | |
Suganuma et al. | Precise position estimation of autonomous vehicle based on map-matching | |
CN107783123A (en) | Pilotless automobile complex environment anticollision MMW RADAR SIGNAL USING processing system and method | |
CN110596731A (en) | Active obstacle detection system and method for metro vehicle | |
CN103487810A (en) | Method for detecting terrain obstacles with unmanned vehicle-borne radar based on echo characteristics | |
CN103176187A (en) | Filtering method of ground expressway target of onboard warning radar | |
CN107783121A (en) | Pilotless automobile anti-collision radar system signal processing system and method based on combined waveform | |
CN107783130B (en) | Signal processing method of unmanned vehicle complex environment anti-collision system based on combined waveform | |
CN107783098A (en) | Rotor wing unmanned aerial vehicle anticollision MMW RADAR SIGNAL USING processing unit | |
CN107783132A (en) | Autonomous driving vehicle anticollision millimetre-wave radar system and signal processing method | |
CN111175715B (en) | Auxiliary driving system and method capable of inhibiting radar close-range harmonic waves | |
CN107783117A (en) | Pilotless automobile anticollision MMW RADAR SIGNAL USING processing method | |
CN107783120A (en) | Pilotless automobile anticollision MMW RADAR SIGNAL USING processing unit | |
CN107783126B (en) | Signal processing method and device of automatic driving automobile anti-collision radar system based on combined waveform | |
CN107783090A (en) | Fixed-wing unmanned plane CAS method for processing radar signals based on millimetre-wave radar | |
CN112455503A (en) | Train positioning method and device based on radar | |
Mertens et al. | Ground target tracking with experimental data of the PAMIR system | |
US8237606B2 (en) | UHF radar system for implementing a method of eliminating vegetation clutter | |
CN107783131B (en) | Signal processing method of automatic driving automobile anti-collision radar system based on combined waveform | |
Zhong et al. | Dynamic obstacle detection and tracking based on 3D lidar | |
CN110542902B (en) | Simple height measurement method suitable for vehicle millimeter wave radar |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180309 |