JP7011305B2 - Separation frequency synthesis radar device, distance estimation method and program - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act Published on November 16, 2017 http: // www. ieice. org / ken / program / index. php? mode = program & tgs_regid = 1492363d98741916c2ed704a5baa635bf9baf6b2979a963220bc39e7ac871325 & tgid = IEICE-SANE & playout = & lang = eng December 7, 2017 ieice. org / ken / program / index. php? tgs_regid = 875ab377fb0bdbdd982b074a3b88beaad904c5ecf593aa4fae1348ee21cf41a0 & tgid = IEICE-WBS & lang =

The present invention relates to a separated frequency synthesis radar device, a distance estimation method, and a program that enable high-performance measurement by synthesizing signals in a plurality of separated frequency bands.

In recent years, there has been a demand for higher resolution and higher reliability of in-vehicle radar devices and radar devices installed as infrastructure by using the millimeter wave band. That is, even with the existing in-vehicle radar device, it is possible to detect a relatively large object such as a vehicle, but it is necessary to be able to detect a pedestrian or a smaller object, and the millimeter wave radar is even higher. Higher resolution and higher reliability are required.

In order to increase the resolution of the radar device, for example, it is conceivable to increase the frequency bands used and synthesize the received signals in each frequency band.
For example, Patent Document 1 describes a technique of transmitting signals having two frequencies and detecting a target from the frequency components of the signal obtained by time-integrating and averaging the received signals of the reflected signals. The technique described in Patent Document 1 is performed to suppress unnecessary components contained in a received signal.

Japanese Unexamined Patent Publication No. 5-281344

By the way, the frequency band that can be used by the radar device is restricted by the regulations of the country in which it is used, and it is difficult for the in-vehicle radar device to use the entire millimeter wave band, for example. For example, in Japan, a 60 GHz band, a 76 GHz band, or the like is assigned as a frequency band that can be used by a radar device, and a band between these frequency bands cannot be used by a radar device. Therefore, in order to increase the frequency band used by the radar device, it is conceivable to combine the received signals of each frequency band by using a plurality of available separated frequency bands.

Here, as described in Patent Document 1, it has been conventionally proposed to synthesize received signals of a plurality of frequencies, but as the synthesis process, signals of each frequency are time-integrated and averaged. This is a relatively simple process, and the improvement of accuracy by averaging is limited to (1 / √2), and the improvement of high performance is limited.

The present invention has been made in view of these points, and is an isolated frequency synthetic radar device capable of improving resolution and reliability by using a plurality of available frequency bands, and distance estimation. The purpose is to provide methods and programs.

The separation frequency synthesis radar device of the present invention includes a phase difference estimation processing unit 3 to which a received signal of the separation frequency band radar group 1 is supplied, and a coherent distance estimation processing unit 4.
The phase difference estimation processing unit 3 estimates the phase difference of the received signals of a plurality of separated frequency bands based on the received signals of each frequency band obtained from the separated frequency band radar group 1.
The coherent distance estimation processing unit 4 determines the optimum evaluation value from the signal obtained by using the phase difference obtained by the phase difference estimation processing unit, and obtains the target distance from the optimum evaluation value. Here, the estimation of the phase difference in the phase difference estimation processing unit and the determination of the optimum evaluation value in the coherent distance estimation processing unit using the estimated phase difference are repeated a plurality of times to obtain the maximum optimum evaluation value. It searches and outputs the estimated distance at the maximum optimum evaluation value as the target distance.
Further, the distance estimation method of the present invention estimates the phase difference of the received signals of a plurality of separated frequency bands based on the received signals of each frequency band obtained from the separated frequency band radar group using a plurality of separated frequency bands. In the coherent distance estimation process and the phase difference estimation process, the optimum evaluation value is determined from the signal obtained by using the phase difference estimated process and the phase difference obtained in the phase difference estimation process, and the target distance is obtained from the optimum evaluation value. The estimation of the phase difference of the above and the determination of the optimum evaluation value in the coherent distance estimation process using the estimated phase difference are repeated multiple times to search for the maximum optimum evaluation value and the maximum optimum evaluation value. Includes a target distance search process that outputs the estimated distance in. As the target distance.
Further, the program of the present invention causes a computer to execute the phase difference estimation process, the coherent distance estimation process, and the target distance search process of the distance estimation method.

According to the present invention, it is possible to achieve higher resolution than when signals obtained in a plurality of frequency bands are simply averaged, and it is possible to improve reliability in detecting a target.

It is a block diagram which shows the structure of the separation frequency synthesis radar apparatus according to 1st Embodiment of this invention. It is a flowchart which shows the processing example by the separation frequency synthesis radar apparatus by the 1st Embodiment of this invention. It is explanatory drawing which shows the image example (without phase estimation) of the frequency band of the separation frequency synthesis radar apparatus. It is explanatory drawing which shows the image example (after phase estimation) of the frequency band of the separation frequency synthesis radar apparatus. It is explanatory drawing which shows the example of the relationship between the estimated distance and the likelihood value by the example of 1st Embodiment of this invention. It is a block diagram which shows the structure of the separation frequency synthesis radar apparatus by the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the separation frequency synthesis radar apparatus by 3rd Embodiment of this invention. It is a flowchart which shows the processing example in the separation frequency synthesis radar apparatus by the 3rd Embodiment example of this invention. It is a block diagram which shows the structure of the separation frequency synthesis radar apparatus by 4th Embodiment of this invention. It is a flowchart which shows the processing example in the separation frequency synthesis radar apparatus by 4th Embodiment of this invention. It is a block diagram which shows the structure (modification example 1) of the separation frequency synthesis radar apparatus by the modification of the embodiment of this invention. It is a block diagram which shows the structure (modification example 2) of the separation frequency synthesis radar apparatus by the modification of the embodiment of this invention. It is a block diagram which shows the structure (modification example 3) of the separation frequency synthesis radar apparatus by the modification of the embodiment of this invention. It is a block diagram which shows the structure (modification example 4) of the separation frequency synthesis radar apparatus by the modification of the embodiment of this invention. It is explanatory drawing which shows the example of the effect when the target is detected by the separation frequency synthesis radar apparatus by the Example of Embodiment of this invention. It is a block diagram which shows the structure of the separation frequency synthesis radar apparatus by 5th Embodiment of this invention. It is a flowchart which shows the processing example in the separation frequency synthesis radar apparatus by 5th Embodiment of this invention.

Hereinafter, examples of each embodiment of the present invention will be described in order with reference to the drawings. In the drawings illustrating the examples of each embodiment, the same parts are designated by the same reference numerals, and duplicate description of the configurations and processes already described in the examples of the other embodiments will be omitted.

<Example of the first embodiment>
First, the separation frequency synthesis radar device of the first embodiment of the present invention will be described with reference to FIGS. 1 to 5.
FIG. 1 shows the configuration of the separation frequency synthesis radar device 10a of the first embodiment.
The separation frequency synthesis radar device 10a includes a separation frequency band radar group 1, a non-coherent distance estimation processing unit 2, a phase difference estimation processing unit 3, and a coherent distance estimation processing unit 4.

The separated frequency band radar group 1 is a radar group that uses a plurality of separated frequency bands, and is a frequency direction signal before band synthesis of the synthetic band radar or an output signal of the pulse compression radar, and a reference function in the frequency domain. It is multiplied by the complex conjugate. The separated frequency band radar group 1 outputs a signal depending on the transmission frequency and the target distance. Here, as an example, two frequency bands of 60 GHz band and 76 GHz band are used as a plurality of separated frequency bands.
The separated frequency band radar group 1 can be applied by various methods as long as it is a radar group using a plurality of separated frequency bands. Further, three or more frequency bands may be used as a plurality of separated frequency bands.

The output of the separated frequency band radar group 1 consists of a carrier frequency of each band fc _m and an in-band frequency f _n .
At this time, the frequency axis signal X measured by the radar in each frequency band is represented by the following equation [Equation 1]. In this equation [Equation 1], f ₀ is the reference transmission frequency, f _c is the carrier frequency, Δf is the step size of the sample in the frequency direction, m is the frequency band number, n is the frequency number in each band, and s is the measurement. The snapshot number (CPI number) indicating the number of times, K is the number of targets, k is the target number (number set for each target), and R _k is the target frequency (target distance).

In the equation [Equation 1], the case where the frequency dependence between the bands and the frequency dependence within the band are taken into consideration for the target reflected power is shown. If there is no interband frequency in the target reflected power, the frequency axis signal X is expressed by the following equation [Equation 2]. In the equation [Equation 2], α _k is a constant determined by the shape of the target and the like.

を、α_ｍ，ｋとおけば、周波数軸上信号Ｘは、次の［数３］式で示される。 here,

If is set to α _{m, k} , the frequency axis signal X is represented by the following equation [Equation 3].

In the following explanation, for the sake of simplicity, it is assumed that there is no difference in frequency between and within the band in the target reflected power, the number of separated frequencies is 2 (here, 2 in the 60 Ghz band and the 76 Ghz band), and the target number is set. It will be described as 2.
At this time, the frequency axis signals X ₆₀ and X ₇₆ measured by the radar in each frequency band are represented by the following equations [Equation 4] and [Equation 5]. The [Equation 4] and [Equation 5] equations show the [Equation 1] equation in the form of a determinant, and show an example in which the distances R ₀ and R ₁ of two targets exist. The distances R ₀ and R ₁ of these two targets are the values that we want to finally obtain.

In this way, in the separated frequency band radar group 1, the frequency axis signals X ₆₀ and X ₇₆ measured by the radar in each frequency band are supplied to the non-coherent distance estimation processing unit 2.

As shown in FIG. 1, the non-coherent distance estimation processing unit 2 is obtained by the likelihood calculation unit 21 for calculating the evaluation value in each band (here, the evaluation value is the likelihood) and the likelihood calculation unit 21. It also has a maximum likelihood search unit 22 that searches for the maximum likelihood (optimal evaluation value) at which the sum of the likelihoods of each band is maximized. It is an example that the likelihood calculation unit 21 calculates the likelihood as an evaluation value and the maximum likelihood search unit 22 searches for the maximum likelihood as the optimum evaluation value, and calculates other evaluation values and the optimum evaluation value. You may try to do it. That is, in addition to the case of searching for the likelihood in maximum likelihood estimation, the case of searching for the optimum evaluation value using the square error in the least squares method as the evaluation value, and the case of searching for the optimum evaluation value using the posterior probability in MAP estimation as the evaluation value. In some cases, the optimum evaluation value is searched for using the coincidence of moments in the moment method as the evaluation value.

When the frequency axis signals X ₆₀ and X ₇₆ represented by the equations [Equation 4] and [Equation 5] are input to the likelihood calculation unit 21 of the non-coherent distance estimation processing unit 2, the following [Equation 6] Perform the operations shown in the equation and the equation [Equation 7] to calculate the likelihood in each band. A shown in these [Equation 6] and [Equation 7] equations is shown in [Equation 8]. In the equations [Equation 6] and [Equation 7], H indicates the complex conjugate transpose, and trace indicates the sum of the diagonal terms of the matrix.

Then, the maximum likelihood search unit 22 searches for the maximum likelihood at which the likelihood of the frequency axis signals X ₆₀ , X ₇₆ and each function m, n, s for each band becomes maximum. For example, the sum of the likelihoods of the two frequency axis signals X ₆₀ and X ₇₆ is obtained as shown in the following equation [Equation 9]. In the equation [Equation 9], arg max is a function that returns an argument that maximizes the value of the evaluation function (here, the value of the likelihood).

Then, the maximum likelihood search unit 22 finds the distances R ₀ and R ₁ at which the sum of the likelihoods is maximum by a non-linear search such as the quasi-Newton method and the Levenberg-Marquardt method, and finds the search result in the maximum likelihood search. As the search result in the part 22, the distances R ₀ and R ₁ of the two targets are obtained. In addition, the formula [Equation 9] tries to find the distance at which the sum of the likelihoods is maximum, but this is just an example. For example, the arithmetic mean value, the maximum search for the harmonic mean value, the reciprocal of the arithmetic mean value, and the harmonic mean. It is also possible to replace it with the minimum search for the reciprocal of the value.
The non-coherent distance estimation processing unit 2 outputs the distances R ₀ and R ₁ of the two obtained targets.

及び、距離R₁の目標に対応する以下の複素振幅

を、最小二乗法等によりそれぞれ推定する。 The information of the distances R ₀ and R ₁ output by the non-coherent distance estimation processing unit 2 is supplied to the phase difference estimation processing unit 3.
The phase difference estimation processing unit 3 takes the distances R ₀ and R ₁ , which are the outputs of the non-coherent distance estimation processing unit 2, as inputs, and has the following complex amplitude corresponding to the target of the distance R ₀ .

And the following complex amplitudes corresponding to the goal of distance R ₁

Are estimated by the least squares method or the like.

Here, the amplitude phase difference α _{0, s} , α _{1, s} between the target separation frequency bands of each snapshot corresponding to the distances R ₀ and R ₁ is obtained as shown in the following equation [Equation 10].

In this way, the phase difference estimation processing unit 3 obtains the amplitude phase difference α _{0, s} , α _{1, s} between the target separation frequency bands corresponding to the distances R ₀ and R ₁ , and the obtained amplitude phase difference. α _{0, s} and α _{1, s} are supplied to the coherent distance estimation processing unit 4.

The coherent distance estimation processing unit 4 performs the maximum likelihood search processing with the amplitude phase differences α _{0, s} , α _{1, s} obtained by the phase difference estimation processing unit 3 as the initial phase, and from the maximum likelihood search processing result. , Distances R ₀ , R ₁ are estimated. Here, the maximum likelihood search process is repeatedly executed a preset number of times. Further, after repeating the maximum likelihood search process a preset number of times, the maximum likelihood search process is repeated from the phase difference estimation by the phase difference estimation processing unit 3 for a preset number of global trials. Execute.

The flowchart of FIG. 2 shows the flow of the phase difference estimation in the phase difference estimation processing unit 3 and the maximum likelihood search processing in the coherent distance estimation processing unit 4. The separation frequency synthesis radar device 10a shown in FIG. 1 includes a control unit (not shown) that controls the processing operations of the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4, and is controlled by this control unit. In the procedure shown in the flowchart of FIG. 2, the phase difference estimation processing unit 3 performs the phase difference estimation and the coherent distance estimation processing unit 4 performs the maximum likelihood search processing.
First, the phase difference estimation processing unit 3 estimates the phase difference from the distances R ₀ and R ₁ (step S11). The phase difference estimation process here is executed by the operation shown in the above-mentioned equation [Equation 10].

The phase difference φ _α between the separation frequencies obtained by the phase difference estimation processing unit 3 is set in the coherent distance estimation processing unit 4 as the initial phase (step S12). Then, the coherent distance estimation processing unit 4 performs a search process for the maximum likelihood from the stored phase difference φ _α and the distances R ₀ and R ₁ used for estimating the phase difference φ _α (step S13). ..
Here, the coherent distance estimation processing unit 4 inputs an X in which the frequency axis signal X ₆₀ shown in the equation [Equation 4] and the frequency axis signal X ₇₆ shown in the equation [Equation 5] are vertically combined. The operations shown in the equations [Equation 11] and [Equation 12] are performed.

を用いることも可能であり。これにより、より高分解能化が期待される。

It is also possible to use. As a result, higher resolution is expected.

The coherent distance estimation processing unit 4 uses the frequency axis signals X ₆₀ and X ₇₆ as inputs to perform a non-linear search for the distances R ₀ and R ₁ having the maximum likelihood by the quasi-Newton method or the Levenberg-Marquardt method. This is performed, and the estimated values of the distances R ₀ and R ₁ are obtained (step S13). Here, the coherent distance estimation processing unit 4 temporarily stores the obtained likelihood values and the estimated distances R ₀ and R ₁ (step S14), and tries to obtain the estimated values in step S13 in a predetermined local manner. It is determined whether or not the execution has been performed a set number of times (step S15).
When it is determined in step S15 that the number of trials is less than the number of local settings (NO in step S15), the coherent distance estimation processing unit 4 returns to step S13 with respect to the distances R ₀ and R ₁ which are the initial search values. The initial value of the search is updated by adding a random value, and the non-linear search of the distances R ₀ and R ₁ that maximizes the likelihood is performed again.

Further, when it is determined in step S15 that the number of trials has reached the locally set number (YES in step S15), the coherent distance estimation processing unit 4 has a distance R ₀ which is the maximum likelihood from the results stored in step S14. , R ₁ are taken out and stored as the maximum likelihood determination result (step S16).

After that, the coherent distance estimation processing unit 4 determines whether or not the trial for extracting the distances R ₀ and R ₁ having the maximum likelihood in step S16 has been executed a predetermined number of globally set times (step S17).
When it is determined in step S17 that the number of trials is less than the number of global settings (NO in step S17), the coherent distance estimation processing unit 4 determines the estimated distance R ₀ , R ₁ of the maximum likelihood determination result stored in step S16. Is supplied to the phase difference estimation processing unit 3, and the processing is executed again from the phase difference estimation in step S11 using the estimated distances R ₀ and R ₁ of the maximum likelihood determination result.

Further, when it is determined in step S17 that the number of trials has reached the global set number (YES in step S17), the coherent distance estimation processing unit 4 has a distance R ₀ which is the maximum likelihood from the results stored in step S16. , R ₁ are taken out, and the distances R ₀ and R ₁ having the maximum likelihood taken out are output as the estimation result by the coherent distance estimation processing unit 4 (step S18).

3 and 4 are diagrams illustrating an outline of phase difference estimation according to the embodiment of the present embodiment.
Two bands can be connected by performing phase difference estimation in the phase difference estimation processing unit 3 and distance estimation in the coherent distance estimation processing unit 4 using the result, and the average of the distances obtained in each band. Higher distance resolution and accuracy can be obtained than simply finding.
That is, for example, as shown in FIG. 3A, the received signal of the first frequency band Band 1 and the received signal of the second frequency band Band 2 exist at different frequencies, and the phases of the received signals of the two frequency bands Band 1 and Band 2 exist. If not, as shown in FIG. 3B, even if the first frequency band Band 1 and the second frequency band Band 2 are connected as they are as shown by the broken line, the phases of the two received signals do not match. The information between the two bands shown by the broken line cannot be used for distance estimation.

On the other hand, by estimating the phase difference α of the two bands by the phase difference estimation processing unit 3, the phase shift of the signals of the two frequency bands Band1 and Band2 is compensated, and as shown in FIG. 4A, the first frequency band. Even if the received signal of Band 1 and the received signal of the second frequency band Band 2 exist at different frequencies, the received signals of the two frequency bands Band 1 and Band 2 have almost one phase as shown by the broken line in FIG. 4B. It will be connected. Therefore, the information between the two bands shown by the broken line can be used for distance estimation, and high distance resolution and accuracy can be obtained.

FIG. 5 shows an outline of a state in which the detection accuracy of the estimated distances R ₀ and R ₁ is improved by repeatedly executing the distance estimation process in the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4.
In FIGS. 5A, B, and C, the vertical axis represents the likelihood value and the horizontal axis represents the estimated distance.
When the non-coherent distance estimation processing unit 2 calculates the distance for each band, the likelihood value is detected in the state shown in FIG. 5A, and the distance of the point where the likelihood value is the highest is determined. However, when the distance is estimated for each band by the non-coherent distance estimation processing unit 2, the accuracy of the estimated distance is low. The state in which the accuracy of the estimated distance is low here is a state in which the peak of the estimated distance (the point having the highest likelihood value) is not clear, as shown in FIG. 5A.

On the other hand, FIGS. 5B and 5C are examples in which the distance estimation is performed by the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4. FIG. 5B shows a case where the estimation accuracy of the complex amplitude coefficient is poor, and corresponds to a state in which the number of trials in the flowchart of FIG. 2 is small (that is, a state in which the number of global settings is determined to be less than the number of sets in step S17).
As shown in FIG. 5B, when the estimation accuracy of the complex amplitude coefficient is poor, the estimated distance and the likelihood value do not have a constant relationship, and the likelihood value fluctuates repeatedly. In this state, even if a state having a high likelihood value is searched, a state different from the maximum likelihood value may be searched. That is, as shown in FIG. 5B, the peak position of a specific wave in the waveform that undulates and repeats fluctuations is searched for as the maximum likelihood value, and a position different from the original maximum likelihood value is searched for. There is a possibility that it will end up.

here. In the present embodiment, as described in the flowchart of FIG. 2, the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4 repeat the local setting number and the global setting number as the maximum likelihood search processing. By repeating the above, a value close to the true maximum likelihood value is gradually searched.
FIG. 5C shows an example of the relationship between the likelihood value and the estimated distance when the search for the maximum likelihood is repeated a global set number of times.
The state shown in FIG. 5C is a state in which the estimation accuracy of the complex amplitude coefficient is good, there is no wavy fluctuation of the likelihood value as shown in FIG. 5B, and the estimated distance with the maximum likelihood value is selected. , The optimum estimated distance can be obtained.

Specifically, in the non-coherent distance estimation processing unit 2 that averages the signals in each frequency band, the distance measurement accuracy is indicated by 1 / √2 when the number of samples is doubled. On the other hand, when the distance estimation is performed by the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4, the vector length is doubled and the distance measurement accuracy is expected to be 1 / 2√2, and the individual frequency bands are expected. It is possible to achieve high-precision and high-resolution distance measurement performance that cannot be obtained by averaging the processing results of.
The fact that high-precision and high-resolution ranging performance can be obtained means that, for example, the separation frequency synthesis radar device of the present embodiment is mounted on a moving body such as an automobile, and the distance measuring performance is located close to the moving body. When there are two targets (car, person, bicycle, etc.), the two targets can be accurately separated and their positions can be estimated.
For example, as shown in FIG. 15, when two people (targets) P1 and P2 are in close proximity to the front of the vehicle 100, the signal S1 (FIG. 15) for estimating the distance with the conventional radar device. In the lower left), the peak in which the two targets P1 and P2 were combined was detected, and it was difficult to detect the distances of the respective targets P1 and P2 separately. On the other hand, in the signal S2 (lower right of FIG. 15) obtained when the separation frequency synthesis radar device 10a of the present embodiment is mounted on the vehicle 100, the peaks of the two adjacent targets P1 and P2 are individually. The distance between the targets P1 and P2 can be estimated accurately.

<Example of the second embodiment>
Next, the separation frequency synthesis radar device of the second embodiment of the present invention will be described with reference to FIG.
FIG. 6 shows the configuration of the separation frequency synthesis radar device 10b of the second embodiment.
The separation frequency synthesis radar device 10b includes a separation frequency band radar group 1, a phase difference estimation processing unit 3, and a coherent distance estimation processing unit 4. That is, in the separation frequency synthesis radar device 10a (FIG. 1) described in the first embodiment, the non-coherent distance estimation processing unit 2 is provided, distance estimation is performed for each frequency band, and then the phase difference is obtained. The phase difference estimation in the estimation processing unit 3 and the coherent distance estimation processing in the coherent distance estimation processing unit 4 are performed.

On the other hand, in the separation frequency synthesis radar device 10b shown in FIG. 6, the carrier frequency and the in-band frequency f _n of each band fc _m obtained by the separation frequency band radar group 1 are supplied to the phase difference estimation processing unit 3. Then, the phase difference α is estimated. For example, as in the first embodiment, the phase difference estimation processing unit 3 obtains the amplitude phase differences α ₀ and α ₁ between the target separation frequency bands corresponding to the distances R ₀ and R ₁ . Then, the obtained amplitude phase differences α ₀ and α ₁ are supplied to the coherent distance estimation processing unit 4. As described in the flowchart of FIG. 2, the phase difference estimation process in the phase difference estimation processing unit 3 and the distance estimation process in the coherent distance estimation processing unit 4 perform the local setting number of times when searching for the maximum likelihood. Repeat and repeat the global setting number of times.

The separation frequency synthesis radar device 10b having the configuration shown in FIG. 6 can also achieve high-precision and high-resolution distance measurement performance as in the case of the separation frequency synthesis radar device 10a of the first embodiment.

<Example of the third embodiment>
Next, the separation frequency synthesis radar device of the third embodiment of the present invention will be described with reference to FIGS. 7 and 8.
FIG. 7 shows the configuration of the separation frequency synthesis radar device 10c of the third embodiment.
The separation frequency synthesis radar device 10c includes a separation frequency band radar group 1, a non-coherent distance estimation processing unit 2, an estimation target selection distance estimation unit 5, a phase difference estimation processing unit 3, and a coherent distance estimation processing unit 4. Be prepared.

The separation frequency band radar group 1 and the non-coherent distance estimation processing unit 2 are the same as the separation frequency band radar group 1 and the non-coherent distance estimation processing unit 2 described in the first embodiment (FIG. 1). Further, the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4 are the same as the phase difference estimation processing unit 3 and the coherent distance estimation processing unit 4 described in the first embodiment.

Then, the separation frequency synthesis radar device 10c supplies the information of the target distance obtained by the non-coherent distance estimation processing unit 2 to the estimation target selection distance estimation unit 5. The estimation target selection distance estimation unit 5 uses a projection matrix to perform processing for removing information other than the target, and then performs processing for estimating the distance. For example, when two target distances R ₀ and R ₁ exist, when estimating the target distance R ₀ , the information about the target distance R ₁ is removed (suppressed) to estimate, and the target distance R ₁ is estimated. At this time, the information about the target distance R ₀ is removed (suppressed) and estimated. The details of the processing performed by the estimation target selection distance estimation unit 5 will be described later (FIG. 8).

Then, the information of the target distance in which the unnecessary component is suppressed by the estimation target selection distance estimation unit 5 is supplied to the phase difference estimation processing unit 3. Other configurations are the same as those of the separation frequency synthesis radar device 10a shown in the first embodiment.

The flowchart of FIG. 8 shows the flow of processing performed by the estimation target selection distance estimation unit 5.
The estimation target selection distance estimation unit 5 first adds a minute value to the target distances R ₀ and R ₁ obtained by the non-coherent distance estimation processing unit 2, and sets the initial value to a randomly shifted value ( Step S21).
The minute value added here is, for example, a value of about 1/100 with respect to the search range of the distance.
Then, the estimation target selection distance estimation unit 5 updates the initial value (step S22). In the first trial, the initial value given in step S21 is used as it is as the update value in step S22.

が入力される。そして、減算対象目標距離

に対応する複素振幅ａ_m,kをフーリエ変換等により求める。
離隔周波数帯レーダ群１で得られる２つの周波数軸上信号Ｘ_６０及びＸ_７６での複素振幅ａ_m,kを、［数１３］式及び［数１４］式に示す。 After that, the updated target distance (target distance updated in step S27 described later) is updated as the subtraction target target distance (step S23). In the first trial, the initial value given in step S21 is used as it is as the target distance to be subtracted in step S23.
Next, the estimation target selection distance estimation unit 5 performs an amplitude estimation process of the subtraction target (step S24). Here, the target estimated distance

Is entered. And the target distance to be subtracted

The complex amplitudes a _{m and k} corresponding to are obtained by Fourier transform or the like.
The complex amplitudes _{am and k} of the two frequency axis signals X ₆₀ and X ₇₆ obtained by the separated frequency band radar group 1 are shown in the equations [Equation 13] and [Equation 14].

Here, a (R) is defined by the following equation.

Next, the estimation target selection distance estimation unit 5 generates a subtraction waveform in each frequency band corresponding to the subtraction target (for example, target distances R ₀ and R ₁ ) (step S25). The subtraction target waveform in each frequency band is represented by the following equations [Equation 16] and [Equation 17].

Then, the estimation target selection distance estimation unit 5 performs a subtraction process of subtracting the subtraction waveform from the original signal (frequency axis signal (X ₆₀ , X ₇₆ ) measured by the radar of each frequency band) (step S26). ), Update to the subtracted signal as shown in the following equations [Equation 18] and [Equation 19].

After obtaining the subtracted signal, the estimation target selection distance estimation unit 5 sets the steering vector in each band as B60_1 and B76_1, and uses the quasi-Newton method, the Levenberg-Marquardt method, or the like to obtain the target distance (^) R. _j is estimated, and the target distance is updated with the estimation result (step S27). In addition, in this specification, (^) means that ^ is added above the symbol after that. In the formula described as image data, ^ is described in the original position.
The following equations [Equation 20] to [Equation 25] show the estimation process of the target distance, and the target distance is estimated from the maximum search of the evaluation equation shown in the equation [Equation 25].

を用いることで距離推定精度の向上が期待される。 It should be noted that the equations [Equation 20] to [Equation 21] are different from the above when the distances between the targets are particularly close.

Is expected to improve the distance estimation accuracy.

When the number of subtraction targets is known, the complex amplitudes a _{0, k} , a _{1, k} of the subtraction waveforms in the [Equation 16] and [Equation 17] equations are obtained by the following equations.

When the above equation is used, B60_1 and B76_1 of the equations [Equation 20] and [Equation 21]
Is B60_1 = B76_1 = a (R), respectively. The above is also expected to improve the distance estimation accuracy.

Next, the estimation target selection distance estimation unit 5 determines whether the number of trials tr1 in which the target distance is estimated and updated in step S27 is the preset number of trials or less than the number of trials (the number of trials). Step S28). Here, when the number of trials is less than the set number of trials (NO in step S28), the estimation target selection distance estimation unit 5 returns to the process of step S23 and subtracts by the estimated value of the target distance obtained in step S27. Update the target distance.

Further, when the number of trials tr1 reaches the preset number of trials in step S28 (YES in step S28), the estimation target selection distance estimation unit 5 determines that the number of trials tr2 which updated the initial value in step S22 is determined. It is determined whether the number of trials is preset or less than the number of trials (step S29). Here, when the number of trials is less than the set number of trials (NO in step S29), the estimation target selection distance estimation unit 5 returns to the initial value update process in step S22, and updates the initial value for each trial number. Obtain the target distance obtained as a process.

Further, when the number of trials tr2 reaches the preset number of trials in step S29 (YES in step S29), the estimation target selection distance estimation unit 5 presets the number of trials tr3 that generated the initial value in step S21. It is determined whether the number of trials has been made or the number of trials is less than the number of trials (step S30). Here, when the number of trials for which the initial value is generated is less than the set number of trials (NO in step S30), the estimation target selection distance estimation unit 5 returns to the process of generating the initial value in step S21 and returns to the initial value generation process. Repeat from the generation process.

Further, when the number of trials tr3 reaches the preset number of trials in step S30 (YES in step S20), the estimation target selection distance estimation unit 5 uses the set of target distances stored in step S27 to target. The distance is determined (step S31). Here, for example, the target distance (^) R _j is determined by the median value or the mode value of the set of target distances to be input.
The estimation target selection distance estimation unit 5 outputs the target distance thus determined to the phase difference estimation processing unit 3.

By providing the estimation target selection distance estimation unit 5, even if the output of the separated frequency band radar group 1 includes a large number of target signals, the target signal to be the target is extracted and the phase difference estimation process is performed. The phase difference estimation in the unit 3 and the distance estimation can be performed in the coherent distance estimation processing unit 4, and high-precision and high-resolution distance measurement performance can be achieved.

<Example of the fourth embodiment>
Next, the separation frequency synthesis radar device of the fourth embodiment of the present invention will be described with reference to FIG.
FIG. 9 shows the configuration of the separation frequency synthesis radar device 10d of the fourth embodiment.
The separation frequency synthesis radar device 10d includes a separation frequency band radar group 1, a non-coherent distance estimation processing unit 2, an estimation target selection distance estimation unit 5, a phase difference estimation processing unit 3', and an estimation target selection distance estimation unit 5. ′ And.
The processing by the non-coherent distance estimation processing unit 2 and the estimation target selection distance estimation unit 5 is the same as that of the separation frequency synthesis radar device 10c of the third embodiment.

Then, in the separation frequency synthesis radar device 10d of the present embodiment, the phase difference estimation processing unit 3'estimates the phase difference from the estimated distance obtained by the estimation target selection distance estimation unit 5, and the phase difference estimation processing unit 3 When the estimation target selection distance estimation unit 5 ′ estimates the distance using the phase difference estimated by ′, each of them is subtracting information other than the target as in the case of the estimation target selection distance estimation unit 5. Performs the process of estimating the target distance.

及び

を、最小二乗法（一般化逆行列）等にてそれぞれ推定する。この複素振幅において、Ｓはスナップショット数、ｓはスナップショット番号である。
ここで、α６０′＜Ｓ＞及びα７６′＜Ｓ＞は、次の［数２６］式及び［数２７］で定義される。［数２６］式及び［数２７］に示すＡは、［数８］式に示すＡである。 The flowchart of FIG. 10 shows the flow of processing performed by the phase difference estimation processing unit 3'and the estimation target selection distance estimation unit 5'.
First, the phase difference estimation processing unit 3'estimates the phase difference (step S41).
Here, the phase difference estimation processing unit 3'takes the distances R ₀ and R ₁ as inputs, and the complex amplitude with respect to the target of the distances R ₀ and R ₁ .

as well as

Are estimated by the least squares method (generalized inverse matrix) or the like. In this complex amplitude, S is the number of snapshots and s is the snapshot number.
Here, α60 ′ <S> and α76 ′ <S> are defined by the following [Equation 26] equation and [Equation 27]. The A shown in the [Equation 26] and [Equation 27] is the A shown in the [Equation 8].

At this time, the amplitude phase difference α ₀ and α ₁ between the target separation frequency bands with respect to the distances R ₀ and R ₁ are expressed by the following equation [Equation 28].

In the phase difference estimation process in step S41, the phase difference is initialized to a value such as 1 in the first loop in which the input distances R ₀ and R ₁ are not entered.
The phase difference estimation processing unit 3 ′ outputs the amplitude phase difference α ₀ , α ₁ calculated in this way and the complex amplitude α 60 ′, α 76 ′.
The estimation target selection distance estimation unit 5 ′ to which the amplitude phase difference α ₀ , α ₁ and the complex amplitude α 60 ′, α 76 ′ are input performs the subtraction waveform generation process (step S42). Here, the subtraction target waveforms shown in the equations [Equation 29] and [Equation 30] are generated for each frequency band.

Then, the estimation target selection distance estimation unit 5'performs the subtraction process (step S43). Here, as shown in the following equations [Equation 31] and [Equation 32], the subtraction waveform is subtracted from the original signal (frequency axis signal (X ₆₀ , X ₇₆ ) measured by the radar of each frequency band). Subtract.

Further, the estimation target selection distance estimation unit 5'performs the target distance estimation / update process (step S44). Here, the frequency axis signal (^) X ₆₀ shown in the equation [Equation 31] obtained in step S43 and the frequency axis signal (^) X ₇₆ shown in the equation [Equation 32] are vertically coupled (^). Enter X.
At this time, the target distance is obtained from the maximum search of the evaluation formula of the following formula [33] by using the quasi-Newton method, the Lebenberg-Marquat method, etc., with B as the steering vector obtained by vertically connecting the steering vectors B60_1 and B76_1. (^) Estimate R _j and update it. The TR of the formula [33] is represented by the formula [Equation 34].

としてステアリングベクトルを、各帯域のｂ_ifを縦に結合したＢをステアリングベクトルとして用いることも可能であり、これにより、より高分解能化が期待される。 Especially when the distance between targets is close.

It is also possible to use the steering vector as the steering vector and B, which is a vertically connected bi _if of each band, as the steering vector, which is expected to have higher resolution.

上式を用いた場合、Ｂはａ_ifを縦に結合したものとなる。 The complex amplitude (^) a _{0, k} , (^) a _{1, k} of the subtraction waveform can also be obtained by the following equation.

When the above equation is used, B is a vertically connected a _if .

Then, the estimation target selection distance estimation unit 5'determines whether or not the number of trials tr1 has reached the set number of times (step S45). Here, when the number of trials tr1 is less than the set number of times (NO in step S45), the process returns to step S42, and the update of the target distance to be subtracted is repeated. Further, when the number of trials tr1 reaches the set number of times (YES in step S45), the process proceeds to the process of determining the number of times of initial value update in step S46.
In the process of determining the number of initial value updates in step S46, it is determined whether or not the number of trials tr2 has reached the set number of times. Here, when the number of trials tr2 is less than the set number (NO in step S45), the process returns to step S41, the target distance (^) R _j is stored, and the input distance R ₀ to the phase difference estimation processing unit 3'. , R ₁ is given a random value, and the phase difference estimation processing unit 3'is executed again.
Then, when the number of trials tr2 reaches the set number of times in step S46 (YES in step S46), the stored target distance (^) R _j and TR are output.

Further, the estimation target selection distance estimation unit 5'outputs the target distance (^) R _j that takes the maximum value of TR among the target distance (^) R _j and TR output in step S46 (step S47). ). The target distance (^) R _j output in step S47 is the target distance output by the separation frequency synthesis radar device 10d of the present embodiment.
Even in the case of the separation frequency synthesis radar device 10d of the present embodiment, even if the output of the separation frequency band radar group 1 includes a large number of target signals, the target signal of the target is taken out and the distance is obtained. Estimates can be made, and high-precision and high-resolution ranging performance can be achieved.

<Example of the fifth embodiment>
Next, the separation frequency synthesis radar device of the fifth embodiment of the present invention will be described with reference to FIG.
FIG. 16 shows the configuration of the separation frequency synthesis radar device 10i of the fifth embodiment.
The separation frequency synthesis radar device 10i includes a separation frequency band radar group 1, a phase difference estimation processing unit 3 ″, and an estimation target selection distance estimation unit 5 ″.

The flowchart of FIG. 17 shows the flow of processing performed by the phase difference estimation processing unit 3 ″ and the estimation target selection distance estimation unit 5 ″.
First, the target number is set as 1 by updating the target number (step S40). After that, the processes from steps S41 to S47 (FIG. 10) in the fourth embodiment already described are carried out, and further, the end determination by the following formula is carried out (step S48). In the end determination of step S48, the process ends when the value is smaller than the set value ε (YES in step S48).

If the condition that the set value is smaller than the set value ε is not satisfied in the end determination of step S48 (NO in step S48), the process returns to the update of the target number in step S40, and the target number plus 1 is newly set. ..
Also in the case of the flowchart shown in FIG. 17, in the first loop in which the input distance R ₀ does not have a value during the phase difference estimation process in step S41, (^) a _0,0 , (^) Initialize a _1,0 to a value such as 1.
According to the separation frequency synthesis radar device according to the fifth embodiment, the target number of the phase difference estimation processing unit and the estimation target selection distance estimation unit may be unknown, and the initial value of the target distance is not required.

<Modification example>
Next, a modified example of each embodiment of the present invention will be described with reference to FIGS. 11 to 14.
The separation frequency synthesis radar device 10e shown in FIG. 11 includes a separation frequency band radar group 1 and a coherent phase difference / distance estimation processing unit 6. The coherent phase difference / distance estimation processing unit 6 takes the signals of the respective frequency bands obtained by the separated frequency band radar group 1 as inputs, and becomes the maximum for obtaining the phase difference α and the target distances R ₀ and R ₁ . The process of searching for the likelihood is performed by simultaneous operations.
Here, when the separated frequency band radar group 1 has two frequency bands, the initial value of the phase difference α is obtained from the input signal. When the separated frequency band radar group 1 has three or more frequency bands, the phase difference α is a random value.

The separation frequency synthesis radar device 10f shown in FIG. 12 includes a separation frequency band radar group 1, a non-coherent distance estimation processing unit 2, and a coherent phase difference / distance estimation processing unit 6.
In the example of FIG. 12, the coherent phase difference / distance estimation processing unit 6 inputs a value obtained from the distance individually calculated for each frequency band by the non-coherent distance estimation processing unit 2.

The separation frequency synthesis radar device 10g shown in FIG. 13 includes a separation frequency band radar group 1, a phase difference estimation processing unit 3, and a coherent phase difference / distance estimation processing unit 6.
In the example of FIG. 13, the phase difference estimation processing unit 3 estimates the phase difference from the signals of each frequency band output by the separated frequency band radar group 1.

The separation frequency synthesis radar device 10h shown in FIG. 14 includes a separation frequency band radar group 1, a phase difference estimation processing unit 3, and a coherent distance estimation processing unit 4.
In the example of FIG. 14, the phase difference estimation processing unit 3 estimates the phase difference from the signals of each frequency band output by the separated frequency band radar group 1, and estimates the coherent distance using the estimated phase difference. The processing unit 4 performs distance estimation processing. The estimation of the phase difference in the phase difference estimation processing unit 3 and the estimation of the distance in the coherent distance estimation processing unit 4 are repeatedly executed as described in the first embodiment and the like.

In each of the embodiments described so far, the distances R ₀ and R ₁ of the two targets are calculated in two frequency bands as the separated frequency bands. The number of these separated frequency bands and targets is an example, and the present invention can be applied to the case of using a larger number of frequency bands and the case of calculating the distance of a larger number of targets. Further, the arithmetic processing described using the mathematical formula shows an example, and the calculation may be performed using a mathematical formula different from the illustrated mathematical formula.
Further, in each embodiment, the likelihood is calculated as one of the optimum evaluation values, and the process of maximizing the likelihood is performed. However, an evaluation value other than the likelihood is obtained and the evaluation is performed. Processing may be performed so as to obtain the optimum evaluation value having the maximum value. As the evaluation value other than the likelihood, as described in the first embodiment, for example, when the square error in the least squares method is used as the evaluation value, or when the posterior probability in the MAP estimation is used as the evaluation value, the moment In some cases, the consistency of moments in the method is used as the evaluation value.

Further, in each embodiment described so far, a processing unit (phase difference estimation unit 3, coherent distance estimation processing unit 4, etc.) that estimates the distance from the separated frequency band radar group 1 and the signal obtained from the radar group 1 and the like are used. ) Is integrated with the radar device, but it may be configured as a device for processing the signal obtained from the existing separated frequency band radar group 1.
Further, the device for performing the distance estimation described so far may be configured by a computer that executes the processing described in each embodiment by calculation. In this case, a program for executing each process described in each embodiment may be created and the program may be implemented in the computer.

Further, in the fourth and fifth embodiments, those having poor accuracy of the complex amplitude (^) a _if in each band (for example, a large deviation from the average value of all frequency bands) are removed and the average value is used. This is expected to improve the speed of convergence in the processing of the separation frequency synthesis radar device. The frequency band with a small S / N is removed by threshold processing.
Further, in steps S41 to S44 of the fourth and fifth embodiments, steps S23 to S27 (FIG. 8) are simultaneously executed, and the output of step S27 is set as the initial value of the search in step S44. It is effective for reducing the load.
In steps S41 to S44 of the fourth and fifth embodiments, the use of a plurality of snapshots can be considered as a countermeasure against the signal degeneration when the snapshot is 1.

1 ... Separation frequency band radar group, 2 ... Non-coherent distance estimation processing unit, 3 ... Phase difference estimation processing unit, 4 ... Coherent distance estimation processing unit, 5, 5', 5 "... Estimating target selection distance estimation processing unit, 6 ... Coherent phase difference / distance estimation processing unit, 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i ... Separation frequency synthesis radar device, 21 ... Probability calculation unit, 22 ... Maximum likelihood search unit

Claims (9)

A phase difference estimation processing unit that estimates the phase difference of the received signals of a plurality of separated frequency bands based on the received signals of each frequency band obtained from the separated frequency band radar group using a plurality of separated frequency bands.
It is provided with a coherent distance estimation processing unit that determines an optimum evaluation value from a signal obtained by using the phase difference obtained by the phase difference estimation processing unit and obtains a target distance from the optimum evaluation value.
The estimation of the phase difference by the phase difference estimation processing unit and the determination of the optimum evaluation value by the coherent distance estimation processing unit using the estimated phase difference are repeated a plurality of times to search for the maximum optimum evaluation value. Then, the separation frequency synthesis radar device that outputs the estimated distance at the maximum optimum evaluation value as the target distance. From the received signal of each frequency band obtained from the separated frequency band radar group, the optimum evaluation value is calculated for each received signal of each frequency band, and the optimum evaluation value that maximizes the sum of the calculated optimum evaluation values of each frequency band is obtained. Further equipped with a non-coherent distance estimation processing unit that searches and obtains target distance information,
The separation frequency synthesis radar device according to claim 1, wherein the phase difference estimation processing unit estimates the phase difference of received signals in a plurality of separated frequency bands from the target distance information obtained by the non-coherent distance estimation processing unit. .. Further provided with an estimation target selection distance estimation unit that estimates the distance by excluding distance information other than the target from the target distance information obtained by the non-coherent distance estimation processing unit.
The separation frequency synthesis radar device according to claim 2, wherein the phase difference estimation processing unit estimates the phase difference of received signals in a plurality of separated frequency bands from the target distance information obtained by the estimation target selection distance estimation unit. .. The separation frequency synthesis radar device according to claim 3, wherein the coherent distance estimation processing unit performs distance estimation excluding distance information other than the target. The separation frequency synthesis radar according to claim 1, wherein the phase difference estimation processing by the phase difference estimation processing unit and the processing for searching for the maximum optimum evaluation value by the coherent distance estimation processing unit are performed by simultaneous calculation. Device. When the phase difference estimation process by the phase difference estimation processing unit and the process of searching for the maximum optimum evaluation value by the coherent distance estimation processing unit are performed at the same time, the initial value of the phase difference is set to
The separation frequency synthesis radar device according to claim 5, which is a random value. Equipped with a group of separated frequency band radars that obtain the received signal reflected from the target in multiple separated frequency bands.
The separation frequency synthesis radar device according to any one of claims 1 to 6, wherein the phase difference estimation processing unit obtains a received signal from the separation frequency band radar group. Phase difference estimation processing that estimates the phase difference of the received signals of multiple separated frequency bands based on the received signals of each frequency band obtained from the separated frequency band radar group using multiple separated frequency bands.
A coherent distance estimation process in which the optimum evaluation value is determined from a signal obtained by using the phase difference obtained in the phase difference estimation process and a target distance is obtained from the optimum evaluation value.
The estimation of the phase difference in the phase difference estimation process and the determination of the optimum evaluation value in the coherent distance estimation process using the estimated phase difference are repeated a plurality of times to search for the maximum optimum evaluation value. A distance estimation method that includes a target distance search process that outputs the estimated distance at the maximum optimum evaluation value as the target distance. A phase difference estimation processing step for estimating the phase difference of received signals in a plurality of separated frequency bands based on the received signals in each frequency band obtained from a group of separated frequency band radars using a plurality of separated frequency bands.
The optimum evaluation value is determined from the signal obtained by using the phase difference obtained in the phase difference estimation processing step.
A coherent distance estimation process step to obtain the target distance from the optimum evaluation value, and
The estimation of the phase difference in the phase difference estimation processing step and the determination of the optimum evaluation value in the coherent distance estimation processing step using the estimated phase difference are repeated a plurality of times to search for the maximum optimum evaluation value. Then, a program that causes a computer to execute a target distance search processing step that outputs the estimated distance at the maximum optimum evaluation value as the target distance.

Images