阅读说明：本技术 一种基于主被动联合声纳阵列的目标路径跟踪方法 (Target path tracking method based on active and passive combined sonar array ) 是由 王世瑞 姜和俊 荣英佼 孙广发 李梦蕾 于 2021-05-24 设计创作，主要内容包括：本发明公开了一种基于主被动联合声纳阵列的目标路径跟踪方法,涉及海洋技术领域,该方法利用主动声纳和被动声纳联合对被跟踪目标进行发现、监视和跟踪,综合规划协调阵列中各个声纳的工作状态以减少相互间干扰,在进行目标路径跟踪时,以阵列中各个声纳的量测数据为基础,通过基于先验概率的卡尔曼融合估计滤波器解算目标运动状态、处理由浅海环境干扰、传感器量测信息不同步以及系统未建模动态引起的非线性特性,并引入自适应因子处理滤波器中量测噪声方程矩阵的不确定性和时变特性,可以实现对运动的被跟踪目标的准确跟踪。(The invention discloses a target path tracking method based on an active and passive combined sonar array, which relates to the technical field of oceans and is characterized in that an active sonar and a passive sonar are combined to find, monitor and track a tracked target, the working states of all sonars in the array are comprehensively planned and coordinated to reduce mutual interference, when the target path is tracked, the measured data of all sonars in the array is taken as the basis, a Kalman fusion estimation filter based on prior probability is used for resolving a target motion state, processing nonlinear characteristics caused by shallow sea environment interference, sensor measurement information asynchronization and system unmodeled dynamic, and introducing uncertainty and time-varying characteristics of a measurement noise equation matrix in an adaptive factor processing filter, so that the accurate tracking of the moving tracked target can be realized.)

1. A target path tracking method based on an active and passive combined sonar array is characterized by comprising the following steps:

in each detection period, controlling the working state of each sonar in an active and passive combined sonar array to be kept unchanged, wherein the sonars in the active and passive combined sonar array comprise a plurality of active sonars and passive sonars, and each tracked target corresponds to at least one sonar;

for each tracked target, acquiring N of the tracked target through the corresponding sonar in the active and passive combined sonar array at the ith detection moment_iEach measured variable is recorded asWherein the k-th measured variableRepresenting the distance between the tracked target and the kth sonar, wherein k is more than or equal to 1;

according to the (k-1) th state variableAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueWhereinAndis a preset initial value;

according to the k-th predicted valueThe k-1 th state variableAnd the k-1 th root mean square matrixDetermining a kth coupling variance matrix between the state variable and the measured variable

Using said kth measured variableThe kth predicted valueAnd the kth coupling variance matrixUpdated to obtain the kth state variableAnd its corresponding k-th root mean square matrix

If k is<N_iLet k be k +1 and execute the said according to the k-1 th state variable againAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueA step (2);

if k is equal to N_iThen use the obtained N_iThe state variables carry out path tracking on the tracked target;

and when the current detection period is finished, adjusting the working state of each sonar in the active and passive combined sonar array based on the beam interference relationship of each sonar in the active and passive combined sonar array, and tracking the target path again in the next detection period.

2. The method of claim 1, further comprising:

filter determination of joint Gaussian probability density obedience of state variables and measured variables based on prior probabilityIs distributed and has:

the kth predicted valueAnd its variance matrixComprises the following steps:

kth coupling variance matrix between state variables and measured variablesComprises the following steps:

determining the kth predicted value after linear transformation by adopting Gaussian noise characteristicsAnd the kth coupling variance matrix

Wherein H^k(X_i)＝||X_i||₂And X_iThe state variable is represented by a number of variables, is a variance matrix of zero-mean white gaussian noise,is a preset initial value.

3. The method according to claim 1 or 2,

the k-th predicted valueThe kth coupling variance matrix

Wherein N is_xIs a coefficient of 2N_xIs [0, k]The largest even number in the interval, r is a coefficient,

4. method according to claim 1 or 2, wherein the updating results in a kth state variableAnd its corresponding k-th root mean square matrixThe method comprises the following steps of:

wherein the function QR () is used to perform a QR decomposition on the matrix,variance matrix that is zero mean gaussian white noiseThe root-mean-square matrix of (c),to representA corresponding gain matrix.

5. The method of claim 2, wherein the variance matrix of zero-mean white gaussian noiseCharacteristic value of (a) and the kth measured variableIs correlated, the method further comprising:

computing Is the mean of all the measured variables before the ith detection period;

detecting whether or not to satisfyWherein the content of the first and second substances,is a preset parameter;

if it isThen utilize andassociated kth adaptive scaling factorCompensationTo obtainAnd correct it

6. The method of claim 5,

the kth adaptive scaling factor Is thatRoot mean square matrix of (d).

7. The method of claim 1, wherein said adjusting the operating state of each sonar in the active-passive combined sonar array in combination with the beam interference relationship of each sonar in the active-passive combined sonar array comprises:

allocating the tracked target to a target beam of a target active sonar and opening the target active sonar to switch to the target beam, wherein the target active sonar is an active sonar closest to the tracked target, each active sonar corresponds to at least two directional beams with different detection angle ranges, and the target beam is a directional beam of which the detection angle range of the active sonar is matched with the azimuth information of the tracked target; and closing the wave beam with wave beam interference with the target wave beam of the target active sonar according to the wave beam interference relation.

8. The method of claim 7 wherein the active-passive combined sonar array is used to detect at least two different tracked targets, and then, when adjusting the operating state of each sonar:

detecting newly appeared tracked targets through the passive sonar, wherein the newly appeared tracked targets are not distributed to any active sonar, sequentially redistributing all the newly appeared tracked targets and all the tracked targets distributed to the corresponding target active sonar according to the processing priority, and correspondingly adjusting the working state of each sonar.

9. The method of claim 8, wherein the processing priority of the tracked object corresponds to an object attribute of the tracked object, the method further comprising: when a tracked target is detected for the first time through a passive sonar, determining the target attribute of the tracked target according to the frequency domain information of the tracked target.

10. The method according to claim 8, wherein the processing priority of the tracked object corresponds to the moving speed of the tracked object, and the faster the moving speed of the tracked object, the higher the corresponding processing priority.

Technical Field

The invention relates to the technical field of oceans, in particular to a target path tracking method based on an active and passive combined sonar array.

Background

The shallow water moving targets refer to various artificial aircrafts capable of autonomously moving in the water depth of less than 10 meters, and the targets are small in size, low in moving speed and difficult to find by shore-based detection facilities, so that accurate identification and tracking cannot be carried out.

Disclosure of Invention

The invention provides a target path tracking method based on an active and passive combined sonar array aiming at the problems and technical requirements, and the technical scheme of the invention is as follows:

a target path tracking method based on an active and passive combined sonar array comprises the following steps:

in each detection period, the working state of each sonar in the active and passive combined sonar array is controlled to be kept unchanged, the sonars in the active and passive combined sonar array comprise a plurality of active sonars and passive sonars, and each tracked target corresponds to at least one sonar;

for each tracked target, acquiring N of the tracked target by corresponding sonar in the active and passive combined sonar array at the ith detection moment_iEach measured variable is recorded asWherein the k-th measured variableThe distance between the tracked target and the kth sonar is represented, and k is larger than or equal to 1;

according to the (k-1) th state variableAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueWhereinAndis a preset initial value;

according to the k-th predicted valueState variable of k-1And the k-1 th root mean square matrixDetermining a kth coupling variance matrix between the state variable and the measured variable

Using the kth measured variableThe kth predicted valueAnd the kth coupled variance matrixUpdated to obtain the kth state variableAnd its corresponding k-th root mean square matrix

If k is<N_iLet k be k +1 and execute again according to the k-1 th state variableAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueA step (2);

if k is equal to N_iThen use the obtained N_iThe state variables carry out path tracking on the tracked target;

and when the current detection period is finished, adjusting the working state of each sonar in the active and passive combined sonar array based on the beam interference relationship of each sonar in the active and passive combined sonar array, and tracking the target path again in the next detection period.

The further technical scheme is that the method also comprises the following steps:

filter determination of joint Gaussian probability density obedience of state variables and measured variables based on prior probabilityIs distributed and has:

the kth predicted valueAnd its variance matrixComprises the following steps:

kth coupling variance matrix between state variables and measured variablesComprises the following steps:

determining the kth predicted value after linear transformation by adopting Gaussian noise characteristicsAnd the kth coupled variance matrix

Wherein H^k(X_i)＝||X_i||₂And X_iThe state variable is represented by a number of variables, is a variance matrix of zero-mean white gaussian noise,is a preset initial value.

The further technical proposal is that the kth predicted valueKth coupled variance matrix

Wherein N is_xIs a coefficient of 2N_xIs [0, k]The largest even number in the interval, r is a coefficient,

the further technical proposal is that the kth state variable is obtained by updatingAnd its corresponding k-th root mean square matrixThe method comprises the following steps of:

wherein the function QR () is used to perform a QR decomposition on the matrix,variance matrix that is zero mean gaussian white noiseThe root-mean-square matrix of (c),to representA corresponding gain matrix.

The further technical scheme is that the variance matrix of the zero mean Gaussian white noiseCharacteristic value of (a) and a k-th measured variableIs correlated, the method further comprising:

computing Is the mean of all the measured variables before the ith detection period;

detecting whether or not to satisfyWherein the content of the first and second substances,is a preset parameter;

if it isThen utilize andassociated kth adaptive scaling factorCompensationTo obtainAnd correct it

The further technical scheme is that the kth self-adaptive scaling factor Is thatRoot mean square matrix of (d).

The further technical scheme is that the working state of each sonar in the active and passive combined sonar array is adjusted by combining the wave beam interference relationship of each sonar in the active and passive combined sonar array, and the method comprises the following steps:

allocating a tracked target to a target beam of a target active sonar, and opening the target active sonar to switch to the target beam, wherein the target active sonar is an active sonar which is closest to the tracked target, each active sonar corresponds to at least two directed beams with different detection angle ranges, and the target beam is a directed beam of which the detection angle range of the active sonar is matched with the azimuth information of the tracked target; and closing the wave beam with wave beam interference with the target wave beam of the target active sonar according to the wave beam interference relation.

The technical scheme is that the active and passive combined sonar array is used for detecting at least two different tracked targets, and when the working state of each sonar is adjusted:

detecting newly appeared tracked targets through the passive sonar, wherein the newly appeared tracked targets are not distributed to any active sonar, sequentially redistributing all the newly appeared tracked targets and all the tracked targets distributed to the corresponding target active sonars according to the processing priority, and correspondingly adjusting the working state of each sonar.

The further technical scheme is that the processing priority of the tracked target corresponds to the target attribute of the tracked target, and then the method further comprises the following steps: and when the tracked target is detected for the first time through the passive sonar, determining the target attribute of the tracked target according to the frequency domain information of the tracked target.

The further technical scheme is that the processing priority of the tracked target corresponds to the movement speed of the tracked target, and the faster the movement speed of the tracked target is, the higher the corresponding processing priority is.

The beneficial technical effects of the invention are as follows:

the method utilizes the combination of active sonar and passive sonar to discover, monitor and track a tracked target, in the system, the active sonar and the passive sonar are arranged at different spatial positions in a shallow water area according to respective working condition requirements and serve as measuring nodes in the system to provide measuring data, the method comprehensively plans and coordinates the startup and shutdown working time slots of all sensors in the array, and the signal interference among the sonars can be fully considered. The method is based on the measurement data of sonar in the array, solves the motion state of a target through a Kalman fusion estimation filter based on prior probability, processes the nonlinear characteristics caused by shallow sea environment interference, sensor measurement information asynchronization and system unmodeled dynamic, and introduces uncertainty and time-varying characteristics of a measurement noise equation matrix in an adaptive factor processing filter, thereby realizing accurate tracking of the moving tracked target.

Drawings

Fig. 1 is a schematic structural diagram of an active-passive joint sonar array of the present application.

Fig. 2 is a schematic flow chart of a target path tracking method according to the present application for performing fusion processing on data at a detection time.

Fig. 3 is a schematic flow chart of the correction by the adaptive scaling factor.

Detailed Description

The following further describes the embodiments of the present invention with reference to the drawings.

The application discloses a target path tracking method based on an active and passive combined sonar array, which utilizes an active and passive combined sonar array formed by a plurality of sonars to track a target path, wherein the active and passive combined sonar array comprises a plurality of active sonars and passive sonars which are arranged to form a preset array form, as shown in figure 1. Wherein, each passive sonar corresponds to an omnidirectional wave beam and can measure the target within 360-degree range. Each active sonar corresponds to at least two directional beams with different detection angle ranges, each directional beam can only detect the target in the corresponding detection angle range, at most one directional beam corresponding to one active sonar works at the same moment, the switching time between the directional beams is short and can be ignored, and the switching can be regarded as being completed instantly.

In the active and passive combined sonar array, the active sonar starts to emit sound signals to output measurement information after a tracked target enters a detection range of the active sonar, and whether the tracked target enters the detection range of the active sonar or not is judged by the passive sonar, and the active sonar and the passive sonar coordinate to realize target path tracking. However, the active sonar needs to emit an acoustic signal to realize the detection of the target when working, the signal is an interference source when the passive sonar works, and meanwhile, different active sonars can also form an interference source with each other, so that the on-off state of each sonar needs to be planned to enable the sonar to be coordinated with each other in the working process of the active and passive combined sonar array, and the coordination process of the application is as follows:

when the active and passive combined sonar array starts to work at an initial moment, all passive sonars are in a power-on state, all active sonars are in a power-off state, when the passive sonars find a tracked target, distance information and azimuth information of the tracked target can be acquired, then the tracked target is distributed to a target beam of a target active sonar and the target active sonar is switched to the target beam according to the arrangement structure of the active and passive combined sonar array, the target active sonar is an active sonar which is closest to the tracked target in distance, and the target beam is a directed beam of which the detection angle range of the active sonar is matched with the azimuth information of the tracked target. And then, the wave beam which has wave beam interference with the target wave beam of the target active sonar is closed according to the wave beam interference relation, and the wave beam which has wave beam interference with the target wave beam of the target active sonar can be an omnidirectional wave beam of the passive sonar or one directional wave beam of the active sonar. Once a tracked target is assigned to the directional beam of the corresponding active sonar, the passive sonar no longer processes information related to the tracked target.

The beam interference relationship between each beam in the active and passive combined sonar array can be predetermined according to the arrangement structure, if two beams can cause interference with each other when working simultaneously, the two beams have beam interference, otherwise, the two beams do not have beam interference and can work simultaneously, and according to the introduction, the beam interference exists between each directional beam corresponding to the same active sonar.

In the working process of the active and passive combined sonar array, there is often more than one tracked target, and when there are a plurality of tracked targets, it is necessary to consider which tracked targets are preferentially detected under the condition of detection resource conflict, so that the processing priority of each tracked target needs to be determined. The processing priority of the tracked target corresponds to the target attribute of the tracked target, when the tracked target is detected for the first time through the passive sonar, the target attribute of the tracked target is determined according to the frequency domain information of the tracked target, the target attribute commonly comprises ships, frogmans and the like, and the processing priorities corresponding to different target attributes are usually configured in advance. Optionally, the processing priority of the tracked target further corresponds to the movement speed of the tracked target, and the faster the movement speed of the tracked target is, the higher the corresponding processing priority is, because the tracked target with the slower movement speed may replace the measurement data with the extrapolated data, and the tracked target with the faster movement speed needs more measurement data to achieve accurate tracking. In actual processing, the processing priority of the tracked target may be determined by any one of the target attribute and the movement speed, and preferably, in an embodiment of the present application, the target attribute is used as a division basis of the priority of the first level, the movement speed is used as a division basis of the priority of the second level, and the processing priorities of the tracked targets are sorted by the two parameters.

Then, the working state of each sonar is adjusted based on the processing priority of each tracked target, specifically: and sequentially redistributing all tracked targets according to the processing priority, wherein all tracked targets comprise newly appeared tracked targets and already appeared tracked targets detected by passive sonar, the newly appeared tracked targets are not distributed to any active sonar, and the already appeared tracked targets are distributed to corresponding target active sonars. When each tracked target is redistributed, the specific method is similar to the method for the single tracked target, and details are not repeated in this application.

And readjusting the working state of each sonar in the active and passive combined sonar array according to the detection result every preset time, wherein the working state of each sonar in the active and passive combined sonar array is kept unchanged in the time period between every two adjustments.

As described above, after each tracked target enters the range of the active-passive combined sonar array, the tracked target is allocated to a beam of at least one sonar, that is, each tracked target corresponds to at least one sonar, and each sonar in the active-passive combined sonar array can acquire relevant information of the corresponding tracked target, which is recorded as a measurement variable, including frequency domain information, distance information, azimuth information, and the like, as mentioned above. However, the time corresponding to the measurement variable of each sonar does not have a uniform timestamp, and the complex characteristics of the shallow sea acoustic propagation channel cause that active and passive sonar measurement noise is difficult to be described by deterministic statistical parameters, but presents high nonlinearity and strong time-varying characteristics, so that the asynchronous measurement variables need to be further fused and calculated to realize path tracking.

The equation of motion of the tracked object is defined as the following form X_i＝F_iX_i-1+W_i，X_i＝[x_p,i,x_v,i,y_p,i,y_v,i]^TIs a vector, i.e. a state variable, x describing the motion state of the tracked target at the ith detection moment_p,iIndicating the position of the tracked object with reference to the x-axis of the coordinate system, x_v,iIndicating the speed, y, of the tracked object with reference to the x-axis of the coordinate system_p,iIndicating the position of the tracked object with reference to the y-axis of the coordinate system, y_v,iRepresenting the speed of the tracked object with reference to the y-axis of the coordinate system, where the x-axis of the coordinate system usually selects the geographical east direction of the location and the y-axis of the coordinate system selects the geographical north direction. F_iIs a state transition matrix, W, determined by a kinematic equation_iIs zero mean white Gaussian noise, and its variance matrix is defined as Q_i. At the ith detection moment, N of the tracked target is obtained through the corresponding sonar in the active and passive combined sonar array_iEach measured variable is recorded asWherein the k-th measured variableRepresents the distance between the tracked target and the kth sonar, and k is more than or equal to 1. Within one detection period, there may be a plurality of detection instants.

The measurement model is written asWherein H (X)_i,X^k)＝||X_i-X^k||₂，X^kAnd a coordinate value representing the kth sonar.V_i ^kIs zero mean white Gaussian noise, and its variance matrix is defined asSince the measurement accuracy is related to the distance, the method has the advantages of simple structure, low cost, and high accuracyCharacteristic value of (a) and the kth measured variableIs related in size, i.e.Is a function of distance and can be written as

The method updates the measurement variable and the state variable by using the filter based on the prior probability, and is realized as follows:

defining N acquired at the ith detection time_iEach measured variable is recorded asAll the measured variables from the first detection time to the ith detection time are recorded as Z₁:_iThe prior probability density function at the i-1 th detection time is denoted as p (X)_i-1|Z_1:i-1) The probability density function of the state transition from the i-1 st detection time to the i-th detection time is denoted as p (X)_i|X_i-1) Of a state variable X_iWith the kth measured variableA likelihood function ofUnder the above definition, the filtering process is divided into three parts:

(1) initialization: setting the probability score of the initial timeCloth function p (X)₀)。

(2) And (3) prediction: p (X)_i|Z_1:i-1)＝∫p(X_i|X_i-1)p(X_i-1|Z_1:i-1)dX_i-1。

(3) Measurement updating:

(4) updating the measurement space:

in the above algorithm, in order to improve the efficiency of the probability density function calculation, a gaussian noise characteristic is used for linear transformation, specifically:

(1) the state variable prediction calculation formula of the prediction link is as follows:

P_i|i-1＝F_iP_i-1|i-1F_i ^T+Q_i；

(2) the joint probability density calculation formula of the state variable and the measurement variable in the measurement updating link is as follows:

wherein the content of the first and second substances,

(3) the joint probability density calculation formula of the state variable and the measurement variable in the measurement space updating link is as follows:

wherein the content of the first and second substances,

the joint gaussian probability density obeys the following distribution:wherein:

the kth predicted valueAnd its variance matrixCalculated by the following two formulas respectively:

kth coupling variance matrix between state variables and measured variablesComprises the following steps:

wherein H^k(X_i)＝||X_i||₂And X_iThe state variable is represented by a number of variables, is a variance matrix of zero-mean white gaussian noise,is a preset initial value.

Based on the algorithm principle, a Kalman fusion estimation filter which is subjected to linear transformation on the filter based on the prior probability by adopting Gaussian noise characteristics is obtained, and the measurement updating process at the ith detection moment is as follows:

(1) determining an initial value of a state variable at an ith detection timeAnd the initial value of its root mean square matrixIn addition to this, a variance matrix P is defined_i|i。

Wherein the content of the first and second substances,F_iis a state transition matrix determined by kinematic equations,is a preset value, and may be generally defined as a real number greater than 0. Variance matrix P_i|iWhich can be generally defined as a positive definite matrix.S_i-1|i-1Is P_i-1|i-1Root mean square matrix of, P_i-1|i-1Is the variance matrix at the i-1 st detection instant and is typically a positive definite matrix, S_Q,iIs zero mean white Gaussian noise W_iOf the variance matrix Q_iRoot mean square matrix of (d).

(2) According to the firstk-1 state variablesAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueIs calculated by the formulaWherein N is_xIs a coefficient of 2N_xIs [0, k]The largest even number in the interval, r is a coefficient,

(3) determining a kth coupling variance matrix between the state variable and the measured variable

The application predicts the value according to the kState variable of k-1And the k-1 th root mean square matrixTo calculateThe calculation formula is as follows:

(4) updated to obtain the kth state variableAnd its corresponding k-th root mean square matrix

Using mainly the k-th measured variable during updatingThe kth predicted valueAnd the kth coupled variance matrixThe calculation formula is as follows:

wherein the function QR () is used to perform a QR decomposition on the matrix,variance matrix that is zero mean gaussian white noiseThe root-mean-square matrix of (c),to representThe corresponding gain matrix is determined by the product of the measured value's Gaussian distribution and the estimated value's Gaussian distribution, and in the actual calculation process, it will be generally approximateThe value is taken as an identity matrix.

(5) If k is<N_iLet k be k +1 and execute again according to the k-1 th state variableAnd its corresponding k-1 root mean square matrixCalculating to obtain the kth predicted valueI.e. the determination of the (k + 1) th state variable is continued.

(6) If k is equal to N_iThen use the obtained N_iAnd the state variables carry out path tracking on the tracked target.

And processing and path tracking by utilizing the loop till the (i + 1) th detection time.

In the above method, as mentioned aboveCharacteristic value of (a) and the kth measured variableThe actual distance is difficult to obtain in the calculation process, if continuous estimation cannot be maintained, the estimator is easy to diverge, and for the continuity of the calculation, the adaptive characteristic and the robustness of the estimator are improved, the embodiment also utilizes the method and the deviceAssociated kth adaptive scaling factorCompensationSpecifically, the method comprises the following steps:

computing Is the average of all the measured variables before the i-th test period,see the disclosure above.

Detecting whether or not to satisfyWherein the content of the first and second substances,is a preset parameter.

If it isIs based onAndthe calculation was performed as described above.

If it isThen utilize andassociated kth adaptive scaling factorCompensationTo obtainAnd correct itTo obtainThen based onAndthe calculation is carried out, and the subsequent specific method is similar.

In the above process, the scaling factor is adaptedIs calculated by the formulaWherein Is thatRoot mean square matrix of (d).

What has been described above is only a preferred embodiment of the present application, and the present invention is not limited to the above embodiment. It is to be understood that other modifications and variations directly derivable or suggested by those skilled in the art without departing from the spirit and concept of the present invention are to be considered as included within the scope of the present invention.