Multi-target classification method in distributed acoustic positioning network
阅读说明：本技术 分布式声学定位网络中多目标分类方法 (Multi-target classification method in distributed acoustic positioning network ) 是由 兰华林 王泽华 师俊杰 吕云飞 梅继丹 靳建嘉 于 2019-10-22 设计创作，主要内容包括：本发明提供一种分布式声学定位网络中多目标分类方法,解决了现有目标分类方法并不能有效的提高定位效率和定位准确度问题,属于声学被动定位领域多目标定位技术。本发明包括：S1、将3个以上节点布放在空间不同的位置,每个节点包括一个声传感器阵列,用于接收目标辐射的信号；根据该信号对目标进行方位估计,得到每个节点不同方位对应的信号能量和频谱特征；对每个节点的信号能量的空间方位特性进行排序,得到每个节点N个信号能量大的方位和频谱特征组成的目标；以第一个节点的各目标和对应的目标频谱特征作为参考,确定其余各节点的N个目标与第一个节点的各目标的对应关系；将相同目标的方位信息结合各节点的位置信息,解算出各目标的位置。(The invention provides a multi-target classification method in a distributed acoustic positioning network, solves the problem that the existing target classification method cannot effectively improve the positioning efficiency and the positioning accuracy, and belongs to the multi-target positioning technology in the field of acoustic passive positioning. The invention comprises the following steps: s1, more than 3 nodes are arranged at different positions in space, and each node comprises an acoustic sensor array used for receiving signals radiated by a target; carrying out azimuth estimation on the target according to the signal to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node; sequencing the spatial orientation characteristics of the signal energy of each node to obtain N targets consisting of orientation with large signal energy and frequency spectrum characteristics of each node; determining the corresponding relation between N targets of the rest nodes and each target of the first node by taking each target of the first node and the corresponding target spectrum characteristic as reference; and combining the azimuth information of the same target with the position information of each node to calculate the position of each target.)
1. The multi-target classification method in the distributed acoustic positioning network is characterized by comprising the following steps:
s1, arranging M nodes at different spatial positions, wherein each node comprises an acoustic sensor array for receiving signals radiated by a target, M is a positive integer greater than or equal to 3, and the target and the acoustic sensor arrays of the nodes are positioned in the same horizontal plane;
s2, carrying out azimuth estimation on the target according to the signals received from each node to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node;
s3, sequencing the spatial orientation characteristics of the signal energy of each node to obtain N targets consisting of orientation with large signal energy and frequency spectrum characteristics of each node;
s4, taking each target of the first node and the corresponding target spectrum characteristics as reference, and determining the corresponding relation between the N targets of the rest nodes and each target of the first node, wherein the corresponding relation indicates that the two targets are the same target;
and S5, combining the azimuth information of the same target with the position information of each node to calculate the position of each target.
2. The method for multi-target classification in a distributed acoustic positioning network as claimed in claim 1, wherein the S4 includes:
and calculating correlation coefficients of the frequency spectrum characteristics of the N targets of the ith node and the frequency spectrum characteristics of the jth target of the first node, wherein the target corresponding to the maximum value of the N correlation numbers and the jth target of the first node are the same target, the maximum value needs to be greater than a set threshold value, i is 2, … M, j is 1, and 2 … N.
3. The method for multi-target classification in a distributed acoustic positioning network as claimed in claim 2, wherein the S2 includes:
and carrying out Fourier transformation on the signals received from each node, carrying out azimuth estimation on the target in each frequency band, and carrying out histogram estimation on the azimuth estimation result to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node.
4. The method for multi-target classification in a distributed acoustic positioning network according to claim 2, wherein the sensor array of each node is a quaternary cross array, a circular ring array, a cylindrical array or a vector sensor array.
Technical Field
The invention relates to a multi-target classification method in a distributed acoustic positioning network, and belongs to the multi-target positioning technology in the field of acoustic passive positioning.
Background
Object classification is a problem that multi-object passive detection and localization cannot be avoided. The characteristics of the target are the basis of target classification, and how to find and confirm the characteristics of the target is the key of multi-target classification. The more effective acoustic target features generally have a definite physical meaning: the time domain, frequency domain and spatial domain features of the signal (horizontal position and depth features of the target). The method of classifying the object is a machine learning method. A problem with distributed acoustic positioning networks is the acoustic characteristics of the target.
In multi-target passive positioning, a false target needs to be removed, so that the positioning efficiency and the positioning accuracy are improved, and regarding the removal of the false target, a false point elimination algorithm based on redundant information and a three-station cross positioning false point elimination algorithm research are provided in the prior art, and the purpose of removing the false point is achieved by selectively utilizing azimuth angle or time difference redundant information measured by multiple stations for association. The prior art also has a direction finding positioning false point eliminating algorithm combining time difference information, which combines arrival time difference information on the basis of cross positioning by using direction finding azimuth information and performs optimal matching according to the similarity of direction finding data between direction finding stations, can realize correct association of data without other prior information and eliminate false points. However, the target classification method using these false target removing algorithms cannot effectively improve the positioning efficiency and the positioning accuracy.
Disclosure of Invention
The invention provides a multi-target classification method in a distributed acoustic positioning network, aiming at the problem that the existing target classification method cannot effectively improve the positioning efficiency and the positioning accuracy.
The invention discloses a multi-target classification method in a distributed acoustic positioning network, which comprises the following steps:
s1, arranging M nodes at different spatial positions, wherein each node comprises an acoustic sensor array for receiving signals radiated by a target, M is a positive integer greater than or equal to 3, and the target and the acoustic sensor arrays of the nodes are positioned in the same horizontal plane;
s2, carrying out azimuth estimation on the target according to the signals received from each node to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node;
s3, sequencing the spatial orientation characteristics of the signal energy of each node to obtain N targets consisting of orientation with large signal energy and frequency spectrum characteristics of each node;
s4, taking each target of the first node and the corresponding target spectrum characteristics as reference, and determining the corresponding relation between the N targets of the rest nodes and each target of the first node, wherein the corresponding relation indicates that the two targets are the same target;
and S5, combining the azimuth information of the same target with the position information of each node to calculate the position of each target.
Preferably, the S4 includes:
and calculating correlation coefficients of the frequency spectrum characteristics of the N targets of the ith node and the frequency spectrum characteristics of the jth target of the first node, wherein the target corresponding to the maximum value of the N correlation numbers and the jth target of the first node are the same target, the maximum value needs to be greater than a set threshold value, i is 2, … M, j is 1, and 2 … N.
Preferably, the S2 includes:
and carrying out Fourier transformation on the signals received from each node, carrying out azimuth estimation on the target in each frequency band, and carrying out histogram estimation on the azimuth estimation result to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node.
Preferably, the sensor array of each node is a four-element cross array, a circular ring array, a cylindrical array or a vector sensor array.
The method has the advantages that in the distributed acoustic positioning network, a plurality of nodes are adopted to detect the target, the acoustic sensor array of each node detects the target, the processing is carried out according to the frequency spectrum characteristics of the detected acoustic signals, the same target detected by each node is found out according to the correlation coefficient of the frequency spectrum characteristics, the false target in the distributed multi-base-station passive positioning is eliminated, the positioning efficiency and the positioning accuracy are improved, and the position of each target is calculated according to the azimuth information of the same target and the position information of each node.
Drawings
FIG. 1 is a schematic diagram illustrating a multi-objective classification method in a distributed acoustic positioning network according to the present invention;
FIG. 2 is a schematic diagram of histogram-based orientation estimation performed by each node;
FIG. 3 is a schematic diagram of a 1 st target characteristic spectrum of the 1 st node;
FIG. 4 is a schematic diagram of the characteristic spectrum of each target of the
FIG. 5 is a graph of the calculated correlation coefficient r21j1(j) Schematic representation of (a).
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The invention is further described with reference to the following drawings and specific examples, which are not intended to be limiting.
The multi-target classification method in the distributed acoustic positioning network is realized under the following assumed conditions:
(1) the incident acoustic signal meets the far-field plane wave condition;
(2) the sensor array of each node may be any array having horizontal azimuth measurement capability for incident acoustic signals, including, but not limited to, a quaternary cross array, a circular array, a cylindrical array, and a vector sensor array. .
(3) Each node can receive the signal of the target.
(4) The target acoustic signal has different spectral characteristics.
(5) The target radiated acoustic signal meets the free field condition, and multi-path and reflected signals are not considered;
(6) the target is located in the same horizontal plane as the acoustic sensor array of each node.
(7) Strict time synchronization is maintained among the nodes.
The method of the present embodiment includes:
s1, arranging M nodes at different spatial positions, wherein each node comprises an acoustic sensor array used for receiving acoustic signals radiated by a target, the acoustic sensors convert the acoustic signals into electric signals, M is a positive integer greater than or equal to 3, and the target and the acoustic sensor array of each node are positioned in the same horizontal plane; the present embodiment employs a four-element cross array, as shown in fig. 1.
S2, carrying out azimuth estimation on the target according to the signals received from each node to obtain signal energy and spectrum characteristics corresponding to different azimuths of each node;
the method comprises the following steps of carrying out Fourier transform on signals received from each node, carrying out azimuth estimation on a target in each frequency band, carrying out histogram estimation on the azimuth estimation result, and obtaining signal energy and spectrum characteristics corresponding to different azimuths of each node.
S3, sequencing the spatial orientation characteristics of the signal energy of each node to obtain N targets consisting of orientation with large signal energy and frequency spectrum characteristics of each node;
performing histogram method orientation estimation on each node, as shown in fig. 2, comparing the histogram estimation result signal power r (a) with a threshold C, and determining that a target exists according to the orientation of the threshold, i (i is 1,2, … M; j is 1,2, … N) of the ith nodei) Corresponding to a frequency spectrum of Sij(f)。
S4, taking each target of the first node and the corresponding target spectrum characteristics as reference, and determining the corresponding relation between the N targets of the rest nodes and each target of the first node, wherein the corresponding relation indicates that the two targets are the same target;
the method for finding the same target in the step comprises the following steps: calculating a correlation coefficient between the frequency spectrum characteristics of the N targets of the ith node and the frequency spectrum characteristics of the jth target of the first node, wherein a target corresponding to a maximum value of the N correlation numbers is the same as the jth target of the first node, the maximum value needs to be greater than a set threshold value, i is 2, … M, j is 1, and 2 … N, specifically as follows:
firstly, fixing the target sequence number of the 1 st node, and then selecting the 2 nd node target:
the characteristic frequency spectrum S of each target of the 2 nd node2j(f) Respectively corresponding to the 1
characteristic spectrum S of each target of 2 nd node2j(f) Respectively corresponding to the 2 nd target characteristic spectrum S of the 1 st node12(f) Calculating a correlation coefficient r21j1(j) Calculating j0＝argmaxj|r21j2(j)|，j0And a threshold CrComparing, if exceeding the threshold, j0The corresponding target is the 2 nd node and the 2 nd target of the 1 st node, namely: are the same target;
calculating each target till the 1 st node, namely characteristic spectrum S of each target till the 2 nd node2j(f) Respectively with node 1N1Target characteristic spectrum
And then carrying out node 3 target selection:
the characteristic frequency spectrum S of each target of the 3 rd node3j(f) Respectively corresponding to the 1
characteristic spectrum S of each target of 3 rd node3j(f) Respectively corresponding to the 2 nd target characteristic spectrum S of the 1 st node12(f) Calculating a correlation coefficient r31j2(j) Calculating j0＝argmaxj|r31j2(j)|，j0And a threshold CrComparing, if exceeding the threshold, j0The corresponding target is the same target as the 2 nd target of the 3 rd node and the 1 st node;
calculating each target till the 1 st node, namely each target characteristic spectrum S till the 3 rd node3j(f) Respectively with node 1N1Target characteristic spectrum
Similarly, each node sequentially selects the targets until the Mth node target selection is performed:
the characteristic frequency spectrum S of each target of the Mth nodeMj(f) Respectively corresponding to the 1
characteristic spectrum S of each target of Mth nodeMj(f) Respectively corresponding to the 2 nd target characteristic spectrum S of the 1 st node12(f) Calculating a correlation coefficient rM1j2(j) Calculating j0＝argmaxj|rM1j2(j)|，j0And a threshold CrComparing, if exceeding the threshold, j0The corresponding target is the same target as the No. 2 target of the No. 1 node;
calculating each target until the 1 st node, namely until each target characteristic spectrum S of the M nodeMj(f) Respectively with node 1N1Target characteristic spectrum
And S5, combining the azimuth information of the same target with the position information of each node to calculate the position of each target.
Although the invention herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that features described in different dependent claims and herein may be combined in ways different from those described in the original claims. It is also to be understood that features described in connection with individual embodiments may be used in other described embodiments.
- 上一篇：一种医用注射器针头装配设备
- 下一篇：一种全声场定向的声源定位方法及装置