JP7076589B2 - Positioning value classification device - Google Patents

Description

The present invention relates to a positioning value classification device for classifying positioning values.

The positioning value classification device performs positioning by GPS (Global Positioning System), TDOA (Time Difference Of Arrival) / FDOA (Frequency Difference Of Arrival), triangulation, radar method, etc., and classifies the positioning values. Here, a positioning classification device in a typical TDOA / FDOA method will be described.

In the TDOA / FDOA method, when a signal from a radio source passes through multiple satellites, the time difference (TDOA) caused by the difference in the position of each satellite and the Doppler frequency difference (FDOA) caused by the difference in the movement vector of each satellite. The radio wave source is positioned from the intersection of (TDOA / FDOA curve). However, it includes positioning values from a plurality of radio wave sources. Therefore, it is desired to automatically classify (cluster) the obtained positioning values for each radio wave source.

In TDOA / FDOA positioning, it is desirable that the number of satellites used for positioning is small in terms of operation. Therefore, positioning using two satellites is common, but since the DOP (Dilution of Precision) determined by the satellite position / velocity vector and the radio source position is large, positioning is performed over a long period of time, and the obtained positioning value is obtained. The positioning error is reduced by averaging. However, when performing clustering, the change in the error (DOP) of the positioning value due to this long-term observation becomes an issue.

Regarding the conventional clustering process of GPS positioning values, for example, Non-Patent Document 1 describes a method of clustering GPS positioning values in a three-dimensional space by the shortest distance method. However, in this method, the error distribution of each positioning point is not taken into consideration, and all the positioning points are uniformly clustered with the same weight. When this method is applied to TDOA / FDOA positioning, there is a problem that the same radio source is classified as a plurality of radio sources because the DOP (error multiplication factor) differs greatly for each positioning value in TDOA / FDOA positioning. Here, the positioning classification device in the TDOA / FDOA method has been described, but the same radio source is similarly classified as a plurality of radio sources for the positioning values by GPS, TDOA / FDOA, triangulation, radar method, etc. There's a problem.

For reference, as a method of weighting the positioning value, for example, in Patent Document 1, the position accuracy of the GPS positioning value (determined by the number of positioning satellites and the DOP) is obtained, and weighting is performed based on the accuracy. A method for calculating a highly accurate positioning value is described. However, there is no description about how to introduce weighting into the measured values when performing the clustering process.

Fumitaka Nakahara, Takahiro Murakami, "Behavioral Pattern Analysis Method Based on Irregular Positioning Position Information History", IPSJ Research Report, vol. 2011-MBL-57, no. 4, pp.1-8, 2011/3.

Japanese Patent Application Laid-Open No. 10-185600.

When positioning is performed over a long period of time, the error distribution differs for each positioning point because the DOP changes during that period. However, in the conventional positioning classification device, the error distribution of each positioning point is not taken into consideration, and all the positioning points are uniformly clustered with the same weight. As a result, when the conventional positioning classification device is applied to these positioning points, the classification success rate is lowered by classifying the positioning value having a large error as another radio wave source.

The present invention has been made to solve the above problems, and in an environment where the positioning points have an error distribution, the positioning values from the same radio source are classified into one cluster with higher classification accuracy than the prior art. The purpose is to obtain a positioning classification device that can be used.

The positioning classification device according to the present disclosure technology is a TDOA calculated from TDOA and FDOA using a plurality of satellites, expressed in a vector format, and output a plurality of positioning values for a plurality of radio wave sources stored in a memory. / FDOA positioning value output unit , error covariance matrix calculation unit that calculates the error covariance matrix for each positioning value, and weighting processing unit that weights each positioning value with the inverse matrix of the error covariance matrix as a weight. , A clustering processing unit that classifies the positioning value into one or a plurality of clusters using the processing result of the weighting processing unit, and the error covariance matrix receives the positioning value and the signal used for measuring the positioning value. Calculated using the position of the receiving antenna and the TDOA error and FDOA error obtained from the peak of the mutual correlation function between the signals received by the receiving antenna, or the positioning value and the position of the receiving antenna. , And the error with the preset TDOA and FDOA.

According to the present invention, in an environment where positioning points have an error distribution, positioning values from the same radio wave source can be classified into one cluster with higher classification accuracy than in the prior art.

The hardware configuration of the positioning value classification device 100 according to the first embodiment. The processing configuration of the processor 1 according to the first embodiment. The functional block diagram of the positioning value classification apparatus 100 which concerns on Embodiment 1. FIG. The flowchart which shows the operation of the positioning value classification apparatus 100 which concerns on Embodiment 1. Positioning values of two radio wave sources in the simulation example of the first embodiment. The clustering processing result when the unweighted positioning value is used in the first embodiment. Clustering processing result when the weighted positioning value is used in the first embodiment. The functional block diagram of the positioning value classification apparatus 100 which concerns on Embodiment 2. FIG. The functional block diagram of the positioning value classification apparatus 100 which concerns on Embodiment 3. FIG. The functional block diagram of the positioning value classification apparatus 100 which concerns on Embodiment 4. FIG.

Embodiment 1.
Hereinafter, embodiments of the present invention will be described. In the first embodiment, as an example of the positioning value classification device, a case where the positioning value classification device 100 classifies the TDOA / FDOA positioning value will be described.

The configuration of the positioning value classification device 100 for classifying the TDOA / FDOA positioning values in the first embodiment will be described. FIG. 1 shows the hardware configuration of the positioning value classification device 100 according to the first embodiment. In the figure, the positioning value classification device 100 includes a processor 1, a memory 2, and a display 3. In each of the following figures, the same reference numerals indicate the same or corresponding parts. FIG. 2 shows the processing inside the processor 1. As shown in FIG. 2, the processor 1 calculates an error covariance matrix for each positioning value, weights the positioning value according to the error of the positioning value, and performs clustering processing. The processor 1 realizes the above function by executing the program stored in the memory 2.

FIG. 3 is a functional configuration diagram of the positioning value classification device 100, and the positioning value classification device 100 includes a TDOA / FDOA positioning value output unit 11, an error covariance matrix calculation unit 12, a weighting processing unit 13, and a clustering processing unit 14. There is. FIG. 4 shows a flowchart showing the operation of the positioning value classification device 100.

Hereinafter, the operation of the positioning value classification device 100 according to the first embodiment will be described with reference to FIG.

The TDOA / FDOA positioning value output unit 11 outputs the TDOA / FDOA positioning value stored in the memory 2 (S41). TDOA / FDOA positioning is the time difference (TDOA) caused by the difference in the position of each satellite and the Doppler frequency difference (FDOA) caused by the difference in the movement vector of each satellite when the signal from the radio wave source passes through multiple satellites. It is a method of positioning the radio wave source from the intersection (TDOA / FDOA curve) of.

（１）

ここで、

（２）

（３）

（４）

（５）

である。cは光速、λは波長、p_intは電波源の位置ベクトル、p_s1, p_s2はそれぞれ衛星１，２の位置ベクトル、v_s1, v_s2はそれぞれ衛星１，２の速度ベクトル、σ_vτ はTDOAの誤差、σ_vf はFDOAの誤差である。また、lat.N, lon.E はそれぞれ、電波源の緯度、経度である。誤差共分散行列の算出方法は２つある。１つ目は、測位値と、その測位値の測定に用いられる信号を受信する受信アンテナ位置と、電波源からの信号が複数の衛星を経由し、複数の受信アンテナで受信した信号間の相互相関関数のピークから求まる該信号を受信する際の時間差/周波数差の誤差と用いて誤差共分散行列を計算する方法である。２つ目は、測位値と、その測位値の測定に用いられる信号を受信する受信アンテナ位置と、あらかじめ設定された該信号を受信する際の時間差/周波数差の誤差を用いて誤差共分散行列を計算する方法である。Ｓ４２で誤差共分散行列を算出した後、クラスタ数を測位値サンプル数（Ｓ４３）とし、次の重み付け処理部１３の処理に移る。The error covariance matrix calculation unit 12 calculates the error covariance matrix P for each positioning point (S42). Here, the error covariance matrix is a solution of the least squares method, and is calculated as follows in the TDOA / FDOA positioning method.

(1)

here,

(2)

(3)

(4)

(5)

Is. c is the speed of light, λ is the wavelength, p _int is the position vector of the radio source, p _s1 and p _s2 are the position vectors of satellites 1 and 2, v _s1 and v _s2 are the velocity vectors of satellites 1 and 2, respectively, and σ _{v τ} is. The error of TDOA, σ _vf , is the error of FDOA. In addition, lat.N and lon.E are the latitude and longitude of the radio wave source, respectively. There are two methods for calculating the error covariance matrix. The first is the positioning value, the position of the receiving antenna that receives the signal used to measure the positioning value, and the signal from the radio wave source passing through multiple satellites and crossing between the signals received by the multiple receiving antennas. It is a method of calculating an error cross-correlation matrix by using the error of the time difference / frequency difference when receiving the signal obtained from the peak of the correlation function. The second is an error covariance matrix using the positioning value, the receiving antenna position for receiving the signal used to measure the positioning value, and the preset time difference / frequency difference error when receiving the signal. Is a method of calculating. After calculating the error covariance matrix in S42, the number of clusters is set to the number of positioning value samples (S43), and the process proceeds to the next weighting processing unit 13.

The weighting processing unit 13 calculates the weight for each sample in the cluster using the error covariance matrix P (S45), and multiplies the weight for each sample (S46).

In the clustering processing unit 14, the distance matrix between all the clusters is calculated (S47), and the combination of the clusters having the closest distance is regarded as one cluster (S48). The clustering process is performed until the minimum value of the distance exceeds the threshold value (S49) or the number of clusters becomes 1 (S44). In the present embodiment, the weight of the positioning value is recalculated every time the clustering integration is repeated according to the error of the positioning value, and the clustering process is performed using the positioning value weighted to the positioning value. Here, there are a plurality of methods for calculating the distance matrix between clusters, and three typical Ward's method, center of gravity method, and shortest distance method are described.

（６）

同様に、他の全クラスタ間の距離を算出し、距離行列を算出する。ここで、あるクラスタCに対するE(C)の計算方法は以下の通りである。

（７）

ここで、x_iはクラスタCに含まれる各サンプルの座標であり、x^∧（xの上に^∧）は次式で求められる重心の座標である。

（８）

ここで、P_iは測位点毎の誤差共分散行列であり、P_i ^-1x_iは重み付けした測位値である。また、Iはクラスタ内のサンプル数であり、クラスタを統合する毎に変化する。In Ward's method, the distance between clusters is calculated as follows. This is a method of integrating clusters based on the sum of squares of the distances between the center of gravity of the samples in the cluster and the samples in the cluster. When using weighted positioning values, the distance between clusters in Ward's method is calculated as follows. The sum of the distance squares E (C1) between the center of gravity of the cluster C1 and each sample, the sum of the distance squares E (C2) between the center of gravity of the cluster C2 and each sample, and the sum of the distance squares E of the center of gravity and each sample when the clusters C1 and C2 are integrated. Calculate (C1∪C2) and calculate the distance D (C1, C2) of Ward's method.

(6)

Similarly, calculate the distance between all other clusters and calculate the distance matrix. Here, the calculation method of E (C) for a certain cluster C is as follows.

(7)

Here, x _i is the coordinates of each sample contained in cluster C, and x ^∧ ( ^∧ on x) is the coordinates of the center of gravity obtained by the following equation.

(8)

Here, P _i is an error covariance matrix for each positioning point, and P _i ^-1 x _i is a weighted positioning value. Also, I is the number of samples in the cluster, which changes each time the cluster is integrated.

（９）

ここで、Ｐ_ｉは、測位点毎の誤差共分散行列であり、Ｐ_ｉ ^-1ｘ_ｉは重み付けした測位値である。また、Ｉはクラスタ内のサンプル数であり、クラスタを統合する毎に変化する。重心値から、クラスタの重心間距離D(C1, C2)を算出して距離行列を算出する。

（１０）

ここで、x_C1 ^∧（x_C1の上に^∧）はクラスタC1の重心値、x_C2 ^∧（x_C2の上に^∧）はクラスタC2の重心値である。Next, a method of calculating the distance between clusters by the center of gravity method will be described. This is a method of integrating clusters that have the shortest distance between the centers of gravity of the clusters among all the combinations of clusters. When the weighted positioning value is used, the distance between the clusters by the center of gravity method is calculated as follows. The center of gravity of the weighted positioning value is

(9)

Here, _{Pi is an error covariance matrix for each positioning point, and Pi -1 x i is} ^a _weighted positioning value. Further, I is the number of samples in the cluster, and changes each time the cluster is integrated. From the center of gravity value, the distance D (C1, C2) between the centers of gravity of the cluster is calculated to calculate the distance matrix.

(10)

Here, x _C1 ^∧ ( ^∧ on x _C1 ) is the centroid value of cluster C1, and x _C2 ^∧ ( ^∧ on x _C2 ) is the centroid value of cluster C2.

（１１）

ここで、Ｐ_ｉ ^-1ｘ_ｉはクラスタC1の重み付けを行った測位値のサンプル、Ｐ_j ^-1ｘ_j はクラスタC2の重み付けを行った測位値のサンプルである。Next, a method of calculating the inter-cluster distance by the shortest distance method will be described. The shortest distance method is a method in which the smallest distance between samples in two clusters is used as the distance between clusters. When the weighted positioning value is used, the distance between clusters by the shortest distance method is calculated as follows.

(11)

Here, P _i ^-1 x _i is a sample of the weighted positioning value of the cluster C1, and P _j ^-1 x _j is a sample of the weighted positioning value of the cluster C2.

In order to confirm the effect of the positioning value classification device 100 in this embodiment, a simulation example is shown. In the simulation, TDOA / FDOA positioning is performed for two radio sources, and clustering is performed using the Ward method. The error distribution of the positioning value differs for each positioning point. The classification accuracy is compared between the case where the weighted positioning value is used and the case where the unweighted positioning value is used.

FIG. 5 shows the positioning values of the two radio wave sources, FIG. 6 shows the clustering processing result when the unweighted positioning value is used, and FIG. 7 shows the clustering processing result when the weighted positioning value is used. From this result, it can be seen that the number of classifications of radio sources is changed from 5 to 2 by performing the clustering process using the weighted positioning values. In addition, the success rate of classification of radio sources (radio source # 1) has increased from 0.9 to 1, indicating that the classification accuracy is improved. This is because by performing clustering processing using the positioning values weighted to the positioning values, even the positioning values having a relatively large error due to the DOP are classified as the same radio wave source. Here, the case where the Ward method is used is shown, but even when the center of gravity method and the shortest distance method are used, the classification accuracy can be improved by using the positioning value classification device 100.

As described above, the positioning value classification device 100 according to the first embodiment has the TDOA / FDOA positioning value output unit 11 which is an output unit for outputting a plurality of positioning values, and the error distribution of the positioning value for each positioning value. It is characterized by including a weighting processing unit 13 that performs weighting according to the weight, and a clustering processing unit 14 that classifies the plurality of positioning values into one or a plurality of clusters by using the weighted positioning values. With this configuration, the positioning values can be classified into one or more clusters in consideration of the error distribution for each positioning value, and in an environment where the positioning points have an error distribution, the same radio wave source with higher classification accuracy than the conventional technique. Positioning values from can be classified into one cluster.

Further, in the positioning value classification device 100 according to the first embodiment, the positioning value is represented in a vector format, and the positioning value classification device 100 is an error covariance unit that calculates an error covariance matrix for each positioning value. The matrix calculation unit 12 is provided, and the weighting processing unit 13 is characterized in that the positioning value is weighted based on the error covariance matrix. With this configuration, the positioning values can be expressed in vector format, and the positioning values from the same radio source can be classified into one cluster with higher classification accuracy than the conventional technique while supporting multidimensional positioning. ..

Further, in the positioning value classification device 100 according to the first embodiment, the clustering processing unit 14 makes each of the plurality of positioning values output from the output unit into a cluster, and integrates two clusters having a short distance into one cluster. It is characterized in that the positioning value is classified into one or a plurality of clusters by repeating the processing. With this configuration, it is possible to efficiently classify the positioning values having a short distance as one cluster while considering the error distribution for each positioning value.

Further, in the positioning value classification device 100 according to the first embodiment, the distance is a distance based on the Ward method, the center of gravity method, or the shortest distance method. With this configuration, the distance between clusters can be accurately measured, and positioning values with close distances can be efficiently classified as one cluster.

Further, in the positioning value classification device 100 according to the first embodiment, the distance is calculated by using the sum of squares of the distances between the center of gravity of the positioning values included in the cluster and the individual positioning values included in the cluster. It is characterized by. With this configuration, it is possible to calculate the distance between the case where the two clusters are integrated and the case where the two clusters are not integrated, and it is possible to appropriately determine whether or not to integrate the two clusters.

Further, in the positioning value classification device 100 according to the first embodiment, the distance is the distance between the center of gravity of the positioning value included in one of the two clusters and the center of gravity of the positioning value included in the other cluster. It is characterized by being. With this configuration, the distance between the centers of gravity of the two clusters can be calculated, and it becomes possible to appropriately determine whether or not to integrate the two clusters.

Further, in the positioning value classification device 100 according to the first embodiment, the distance is the distance between the positioning value included in one of the two clusters and the positioning value included in the other cluster. It is a feature. With this configuration, the shortest distance among the positioning values included in the two clusters can be calculated as the distance between the clusters, and it is possible to appropriately determine whether or not to integrate the two clusters. Will be.

Further, in the positioning value classification device 100 according to the first embodiment, the clustering processing unit 14 integrates the two clusters into one cluster when the distance between the two clusters meets a preset condition. It is characterized by that. With this configuration, it is possible to integrate two clusters when the preset conditions are met, and it is possible to appropriately determine whether or not to integrate the two clusters.

Further, in the positioning value classification device 100 according to the first embodiment, the calculation unit is placed between the positioning value, the position of the receiving antenna that receives the signal used for measuring the positioning value, and the receiving antenna. It is characterized in that the error covariance matrix is calculated by using the error of the time difference or the frequency difference of the signal obtained from the cross-correlation function of the signal. With this configuration, the error covariance matrix can be accurately calculated, and by weighting the positioning value based on the calculated error covariance matrix, it becomes possible to appropriately perform the clustering process.

Further, in the positioning value classification device 100 according to the first embodiment, the calculation unit determines the positioning value, the position of the receiving antenna for receiving the signal used for measuring the positioning value, and the preset signal. It is characterized in that the error covariance matrix is calculated by using the error of time difference or frequency difference. According to this configuration, since the error of the time difference or the frequency difference of the signal is set in advance in the own device, it is not necessary to calculate the error of the time difference or the frequency difference individually with respect to the signal, and the above-mentioned processing amount is small. The error covariance matrix can be calculated.

Embodiment 2.
In the second embodiment, as an application example of the positioning value classification device 100 different from the first embodiment, a case where the positioning value classification device 100 classifies the GPS positioning value will be described.

The configuration of the positioning value classification device 100 according to the second embodiment will be described. FIG. 8 shows a functional configuration diagram of the positioning value classification device 100 of the second embodiment. The positioning value classification device 100 according to the second embodiment includes a GPS positioning value output unit 15, an error covariance matrix calculation unit 16, a weighting processing unit 13, and a clustering processing unit 14.

The hardware configuration diagram, the processing inside the processor, and the flowchart of the positioning value classification device 100 according to the second embodiment are shown in FIGS. 1, 2, and 4 as in the actual embodiment 1. However, the functional configuration diagram shown on the left side of the flowchart of FIG. 4 is changed to the configuration of FIG.

、以下のように算出される。

（１２）

ここで、

（１３）

（１４）

（１５）

(i=1, …, 4) （１６）

である。α_ｉ，β_ｉ，γ_ｉは観測地点から衛星方向への視線の方向余弦（の推定値）である。また、ρ_ｉは衛星と電波源間の距離、(X_ｉ,Y_ｉ, Z_ｉ)は衛星の座標、(x,y,z)は測位値の座標である。なお、式（１２）の誤差共分散行列Pの算出方法は、以下の非特許文献２を参照されたい。
土屋 淳，辻 宏道，「新・GPS測量の基礎」 日本測量協会． In FIG. 8, the GPS positioning value output unit 15 outputs the GPS positioning value stored in the memory 2. The error covariance matrix calculation unit 16 calculates the error covariance matrix for each positioning point. Error covariance matrix

, Is calculated as follows.

(12)

here,

(13)

(14)

(15)

(i = 1,…, 4) (16)

Is. α _i , β _i , and γ _i are (estimated values) of the direction cosine of the line of sight from the observation point toward the satellite. Further, ρ _i is the distance between the satellite and the radio wave source, (X _i , Y _i, Z _i ) is the coordinates of the satellite, and (x, y, z) is the coordinates of the positioning value. For the calculation method of the error covariance matrix P in the equation (12), refer to Non-Patent Document 2 below.
Atsushi Tsuchiya, Hiromichi Tsuji, "New Basics of GPS Surveying" Japan Surveying Association.

In the second embodiment, there are two methods for calculating the error covariance matrix. The first is a method of calculating an error covariance matrix using a positioning value, a receiving antenna position for receiving a signal used for measuring the positioning value, and an error in the arrival time when receiving the signal. Is. The second is an error covariance matrix using the positioning value, the receiving antenna position for receiving the signal used to measure the positioning value, and the error in the arrival time when receiving the preset signal. It is a method of calculation. Since the weighting processing unit 13 and the clustering processing unit 14 are the same as those in the first embodiment, the description thereof will be omitted.

As described above, in the positioning value classification device 100 according to the second embodiment, the positioning value is the positioning value calculated by GPS, and the error covariance matrix calculation unit 16 measures the positioning value and the positioning value. It is characterized in that the error covariance matrix is calculated using the position of the receiving antenna that receives the signal to be used and the arrival time error of the signal. With this configuration, the error covariance matrix can be accurately calculated, and by weighting the positioning value based on the calculated error covariance matrix, it becomes possible to appropriately perform the clustering process.

Further, in the positioning value classification device 100 according to the second embodiment, the error of the arrival time is set in the own device in advance. According to this configuration, since the error of the arrival time is set in the own device in advance, it is not necessary to calculate the error of the arrival time individually for the signal, and the error covariance matrix with a small amount of processing. Can be calculated.

Embodiment 3.
In the third embodiment, as an application example of the positioning value classification device 100 different from the first and second embodiments, a case where the positioning value classification device 100 classifies the triangulation positioning value will be described.

The configuration of the positioning value classification device 100 according to the third embodiment will be described. FIG. 9 shows a functional configuration diagram of the third embodiment. The positioning value classification device 100 according to the third embodiment includes a triangulation positioning value output unit 17, an error covariance matrix calculation unit 18, a weighting processing unit 13, and a clustering processing unit 14.

The hardware configuration diagram, the processing inside the processor, and the flowchart of the third embodiment are the same as those of the actual embodiment 1, and are shown in FIGS. 1, 2, and 4. However, the functional configuration diagram shown on the left side of the flowchart of FIG. 4 is changed to the configuration of FIG.

In FIG. 9, the triangulation positioning value output unit 17 in the positioning value classification device 100 outputs the triangulation positioning value stored in the memory 2.

（１７）

で求められる。ここで、

（１８）

である。なお、式（１７）の誤差共分散行列

算出方法は、以下の非特許文献３を参照されたい。
Ahmed Arafa，Sumant Dalmiya，Richard Klukas，and Jonathan F. Holzman，“Angle-of-arrival reception for optical wireless location technology，” Optics Express，Vol. 23，Issue 6，pp. 7755-7766，2015． The error covariance matrix calculation unit 18 calculates the error covariance matrix for each positioning point, as in the first embodiment. The error covariance matrix P is calculated as follows. Let the azimuth angles measured at two points in the three-dimensional space (xyz coordinate system) be θ ₁ and θ ₂ , respectively, and the elevation angles be φ ₁ and φ ₂ , respectively. The error covariance matrix P is

(17)

Is sought after. here,

(18)

Is. The error covariance matrix in Eq. (17)

For the calculation method, refer to Non-Patent Document 3 below.
Ahmed Arafa, Sumant Dalmiya, Richard Klukas, and Jonathan F. Holzman, “Angle-of-arrival reception for optical wireless location technology,” Optics Express, Vol. 23, Issue 6, pp. 7755-7766, 2015.

In the third embodiment, there are two methods for calculating the error covariance matrix. The first is a method of calculating an error covariance matrix using the positioning value, the position of the receiving antenna that receives the signal used for measuring the positioning value, and the arrival direction error of the signal. The second method is to calculate the error covariance matrix using the positioning value, the position of the receiving antenna that receives the signal used for measuring the positioning value, and the preset arrival direction error of the signal. ..

Since the weighting processing unit 13 and the clustering processing unit 14 are the same as those in the first embodiment, the description thereof will be omitted.

As described above, in the positioning value classification device 100 according to the third embodiment, the plurality of positioning values output from the radar positioning value output unit 19 which is the output unit are the positioning values calculated by the triangular survey, and the calculation unit. The error covariance matrix calculation unit 20 uses the positioning value, the position of the receiving antenna that receives the signal used for measuring the positioning value, and the arrival direction error of the signal to obtain the error covariance matrix. It is characterized by calculating. It is characterized by. With this configuration, the error covariance matrix can be accurately calculated, and by weighting the positioning value based on the calculated error covariance matrix, it becomes possible to appropriately perform the clustering process.

Further, in the positioning value classification device 100 according to the third embodiment, the arrival direction error is set in the own device in advance. According to this configuration, since the arrival direction error is set in the own device in advance, it is not necessary to calculate the error of the arrival time individually for the signal, and the error covariance matrix can be obtained with a small amount of processing. Can be calculated.

Embodiment 4.
In the fourth embodiment, as an application example of the positioning value classification device 100 different from the first to third embodiments, a case where the positioning value classification device 100 classifies the radar positioning value will be described.

FIG. 10 shows a functional configuration diagram of the fourth embodiment. The positioning value classification device 100 according to the fourth embodiment includes a radar positioning value output unit 19, an error covariance matrix calculation unit 20, a weighting processing unit 13, and a clustering processing unit 14.

The hardware configuration diagram, the processing inside the processor, and the flowchart of the fourth embodiment are the same as those of the actual embodiment 1, and are shown in FIGS. 1, 2, and 4. However, the functional configuration diagram shown on the left side of the flowchart of FIG. 4 is changed to the configuration of FIG.

（１９）

で求められる。ここで、

（２０）

（２１）

（２２）

（２３）

である。実施の形態４では、誤差共分散行列の算出方法は２つある。１つ目は、測位値と受信アンテナ位置と到達時間誤差を用いて誤差共分散行列を計算する方法である。２つ目は、測位値と受信アンテナ位置とあらかじめ設定された到達時間誤差を用いて誤差共分散行列を計算する方法である。In FIG. 10, the radar positioning value output unit 19 outputs the radar positioning value stored in the memory 2. The error covariance matrix calculation unit 20 calculates the error covariance matrix for each positioning point, as in the first embodiment. The error covariance matrix P is calculated as follows. Let θ be the azimuth angle measured in the three-dimensional space (xyz coordinate system), Φ be the elevation angle, and R be the distance. At this time, the error covariance matrix P is

(19)

Is sought after. here,

(20)

(21)

(22)

(23)

Is. In the fourth embodiment, there are two methods for calculating the error covariance matrix. The first is a method of calculating an error covariance matrix using the positioning value, the receiving antenna position, and the arrival time error. The second method is to calculate the error covariance matrix using the positioning value, the receiving antenna position, and the preset arrival time error.

Since the weighting processing unit 13 and the clustering processing unit 14 are the same as those in the first embodiment, the description thereof will be omitted.

As described above, in the positioning value classification device 100 according to the fourth embodiment, the positioning value is the positioning value calculated by the radar method, and the error covariance matrix calculation unit 20 which is the calculation unit has the positioning value and the positioning value. It is characterized in that the error covariance matrix is calculated using the position of the receiving antenna that receives the signal used for measuring the positioning value and the arrival time error of the signal. With this configuration, the error covariance matrix can be accurately calculated, and by weighting the positioning value based on the calculated error covariance matrix, it becomes possible to appropriately perform the clustering process.

Further, in the positioning value classification device 100 according to the fourth embodiment, the error of the arrival time is set in the own device in advance. According to this configuration, since the error of the arrival time is set in the own device in advance, it is not necessary to calculate the error of the arrival time individually for the signal, and the error covariance matrix with a small amount of processing. Can be calculated.

1: Processor, 2: Memory, 3: Display, 11: TDOA / FDOA positioning value output unit, 12: Error covariance matrix calculation unit, 13: Weighting processing unit, 14: Clustering processing unit, 15: GPS positioning value output Unit, 17: Triangulation positioning value output unit, 16, 18, 20: Error covariance matrix calculation unit, 19: Radar positioning value output unit, 100: Positioning value classification device

Claims (7)

A TDOA / FDOA positioning value output unit that outputs a plurality of positioning values for a plurality of radio sources calculated from TDOA and FDOA using a plurality of satellites, expressed in a vector format, and stored in a memory .
An error covariance matrix calculation unit that calculates an error covariance matrix for each of the positioning values ,
A weighting processing unit that weights each of the positioning values by using the inverse matrix of the error covariance matrix as a weight .
A clustering processing unit that classifies the positioning value into one or a plurality of clusters by using the processing result of the weighting processing unit is provided .
The error covariance matrix is
The TDOA error and the FDOA error obtained from the positioning value, the position of the receiving antenna that receives the signal used for measuring the positioning value, and the peak of the cross-correlation function between the signals received by the receiving antenna. Calculated using, or
It is calculated by using the positioning value, the position of the receiving antenna, and the preset error between the TDOA and the FDOA.
Positioning value classification device. The clustering processing unit treats each of the plurality of positioning values output from the TDOA / FDOA positioning value output unit as a cluster, and repeats the process of integrating two clusters having a short distance into one cluster to form the positioning value. Is classified into one or more clusters ,
The positioning value classification device according to claim 1 . The positioning value classification device according to claim 2 , wherein the distance is a distance based on the Ward method, the center of gravity method, or the shortest distance method. The positioning value classification device according to claim 2 , wherein the distance is calculated by using the sum of squares of the distances between the center of gravity of the positioning value included in the cluster and the individual positioning values included in the cluster. The positioning value according to claim 2 , wherein the distance is a distance between the center of gravity of the positioning value included in one of the two clusters and the center of gravity of the positioning value included in the other cluster. Sorting device. The positioning value classification device according to claim 2 , wherein the distance is a distance between a positioning value included in one of the two clusters and a positioning value included in the other cluster. One of claims 2 to 6 , wherein the clustering processing unit integrates the two clusters into one cluster when the distance between the two clusters meets a preset condition. The positioning value classification device according to item 1.

Images