A method and apparatus for predicting the state of underwater dynamic targets based on TS fuzzy mapping

Abstract

This invention discloses a method and apparatus for predicting the state of underwater dynamic targets based on T-S fuzzy logic. The method includes: first, acquiring the input variables of the underwater dynamic target at time t-1 for T-S fuzzy inference; then, using the fused underwater dynamic target motion model represented by equation (6) at time t-1 to perform extended Kalman filtering to predict the motion state of the underwater dynamic target at time t; then, taking t = t+1 until the tracking ends. Where X... t X t‑1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. mix B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t The relationship matrix. This invention provides more accurate prediction of the state of underwater dynamic targets and can meet the requirements of underwater dynamic target tracking systems.

Description

Technical Field

[0001] This invention relates to the field of underwater dynamic target tracking technology, and in particular to a method and apparatus for predicting the state of underwater dynamic targets in three-dimensional space based on TS fuzzing. Background Technology

[0002] As a high-tech underwater platform, AUV (Autonomous Underwater Vehicle) is widely used in civilian and military fields such as marine resource exploration, underwater dynamic target detection, and underwater dynamic target reconnaissance. It has become an important marine equipment that countries are vying to develop.

[0003] Among these technologies, AUV dynamic underwater target tracking is one of the core technologies for achieving underwater dynamic target detection and reconnaissance. AUVs use onboard sensors such as sonar and cameras to detect underwater dynamic targets, process the data using algorithms to obtain the motion state information of the underwater dynamic targets, and then execute corresponding control commands such as trajectory tracking and attitude adjustment based on the motion state of the underwater dynamic targets. Accurate prediction of the motion state of underwater dynamic targets is the foundation of dynamic tracking control technology, and its core is to predict the position and velocity information of underwater dynamic targets based on sensor measurement data.

[0004] To accurately predict the motion state of underwater dynamic targets, current technologies mainly employ an interactive multi-model combined with filtering method.

[0005] For example, patent document CN107193009A discloses an interactive multi-model algorithm. This algorithm uses model probability as an evaluation index for each model to acquire filtering innovation and corresponding covariance, designs fuzzy inference for model transition probabilities, and proposes a fuzzy adaptive interactive multi-model tracking algorithm. However, the technical solution in this patent document cannot achieve underwater dynamic target motion prediction in three-dimensional space, which has significant limitations in practical engineering applications.

[0006] For example, patent document CN N110378411A discloses a support vector machine-assisted underwater dynamic target tracking method based on interactive multi-model (IMM). This method trains an IMM classification model by collecting the motion trajectories of underwater dynamic targets as historical data. This classification model is then used to determine the real-time motion model of the underwater dynamic target, and finally, the IMM is combined for filtering and estimation. This method utilizes the support vector machine to enable the IMM algorithm to react quickly to changes in the motion model of underwater dynamic targets. However, this method requires a large amount of historical data to train the support vector machine, and currently, large-scale underwater dynamic target motion datasets are lacking. Summary of the Invention

[0007] The purpose of this invention is to provide a method and apparatus for predicting the state of underwater dynamic targets based on TS fuzzy mapping, which can accurately predict the motion state of underwater dynamic targets in three-dimensional space without requiring a large amount of historical data.

[0008] To achieve the above objectives, this invention provides a method for predicting the state of underwater dynamic targets based on TS fuzzy mapping, comprising:

[0009] First, the input variables for TS fuzzy inference of the underwater dynamic target at time t-1 are collected. Then, the fused underwater dynamic target motion model represented by Equation (6) is used to perform extended Kalman filtering at time t-1 to predict the motion state of the underwater dynamic target at time t. Then, t = t+1 is taken until the tracking ends.

[0010]

[0011] In the formula, X t X t-1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. mix B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t The relation matrix.

[0012] Furthermore, the method for obtaining the fused underwater dynamic target motion model, as represented by equation (6), specifically includes:

[0013] Step 1: Based on the principle of pure azimuth underwater dynamic target tracking of AUV, select the observation equation, and then select several motion models based on the motion characteristics of underwater dynamic targets.

[0014] Step 2, establish the TS fuzzy inference unit, which specifically includes:

[0015] Step 21: Determine the input variables used for TS fuzzy inference;

[0016] Step 22: Set a fuzzy set for each input variable;

[0017] Step 23: Set several sub-fuzzy sets for each fuzzy set;

[0018] Step 24: Set fuzzy rules based on the subfuzzy set and motion model;

[0019] Step 25: Using the TS fuzzy inference method provided by equations (2), (4) and (5), obtain the probability δ of each motion model in the i-th fuzzy rule.i (z(t)):

[0020]

[0021]

[0022]

[0023] In the formula, A i B i C i Let z(t) represent the state transition matrix, noise gain matrix, and output matrix corresponding to the known i-th fuzzy rule, respectively, and z(t) = [z1(t)z2(t)…z j (t)…z l [(t)] represents the antecedent variable of the fuzzy rule, z j (t) represents the j-th antecedent condition, l represents the total number of antecedent conditions for each fuzzy rule, and μ ij (z j (t) represents the preceding variable z. j Membership degree of (t);

[0024] Step 26, calculate the probabilities δ of each motion model obtained in step 25. i (z(t)) is used as the weight of each motion model, and the weights are summed to obtain the fused state transition matrix A. mix and input matrix B mix Thus, the fused underwater dynamic target motion model described by equation (6) is obtained.

[0025] Furthermore, the parameters in the fused underwater dynamic target motion model represented by equation (6) are obtained through equation (3):

[0026]

[0027] Furthermore, the input variables include the relative acceleration and turning angular velocity of the underwater dynamic target. The motion models selected in step 1 are uniform motion model (CV), uniform acceleration motion model (CA), and three-dimensional cooperative turning model (3D-CT), respectively. In step 22, the fuzzy set of relative acceleration is set as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 , is the first fuzzy set S 1 Second fuzzy set S 2 Each subset is set up with 3 fuzzy subsets: negative large fuzzy subset NL, medium fuzzy subset NEU, and positive large fuzzy subset PL. Step 24 obtains the 9 fuzzy rules shown in the table below based on the subfuzzy subsets and the motion model:

[0028]

[0029] The present invention also provides an underwater dynamic target state prediction device based on TS fuzzy mapping, which includes:

[0030] The input acquisition module is used to acquire the input variables of the underwater dynamic target at time t-1 for TS fuzzy inference;

[0031] The underwater dynamic target tracking module uses the fused underwater dynamic target motion model represented by equation (6) to perform extended Kalman filtering at time t-1 to predict the motion state of the underwater dynamic target at time t, and then takes t=t+1 until the tracking ends:

[0032]

[0033] In the formula, X t X t-1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. mix B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t The relation matrix.

[0034] Furthermore, the fused underwater dynamic target motion model represented by equation (6) is obtained through a TS fuzzy inference unit, which specifically includes:

[0035] The input variable determination module is used to determine the input variables used for TS fuzzy inference;

[0036] The fuzzy set setting module is used to set a fuzzy set for each input variable;

[0037] The sub-fuzzy set setting module is used to set several sub-fuzzy sets for each fuzzy set;

[0038] The fuzzy rule setting module is used to set fuzzy rules based on the subfuzzy set and the motion model;

[0039] The motion model probability calculation module is used to obtain the probability δ of each motion model in the i-th fuzzy rule by employing the TS fuzzy inference method provided by equations (2), (4), and (5). i (z(t)):

[0040]

[0041]

[0042]

[0043] In the formula, A i B i C i Let z(t) represent the state transition matrix, noise gain matrix, and output matrix corresponding to the known i-th fuzzy rule, respectively, and z(t) = [z1(t)z2(t)…z j (t)…z l [(t)] represents the antecedent variable of the fuzzy rule, z j (t) represents the j-th antecedent condition, l represents the total number of antecedent conditions for each fuzzy rule, and μ ij (z j (t) represents the preceding variable z. j The membership degree of (t).

[0044] Furthermore, the parameters in the fused underwater dynamic target motion model represented by equation (6) are obtained through equation (3):

[0045]

[0046] Furthermore, the input variables include the relative acceleration and turning angular velocity of the underwater dynamic target, and the motion models are uniform motion model (CV), uniform acceleration motion model (CA), and three-dimensional cooperative turning model (3D-CT), respectively. The fuzzy set setting module sets the fuzzy set of relative acceleration as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 The sub-fuzzy set setting module is the first fuzzy set S. 1 Second fuzzy set S 2 Each subset is set with three fuzzy subsets: negative large fuzzy subset NL, medium fuzzy subset NEU, and positive large fuzzy subset PL. The fuzzy rule setting module obtains the nine fuzzy rules shown in the table below based on the subfuzzy subsets and the motion model:

[0047]

[0048] This invention first establishes the discrete nonlinear state and observation equations of the underwater dynamic target tracking system based on the pure azimuth underwater dynamic target tracking principle of AUVs. Then, according to the motion characteristics of the underwater dynamic target, a suitable set of motion models is selected. Next, a TS fuzzy logic inference system is established, using the relative acceleration and turning angular velocity of the underwater dynamic target as input. A fused underwater dynamic target motion model is obtained by weighting the motion models in the set. This allows for the fusion of multiple underwater dynamic target motion models to obtain a model that better reflects the current motion state of the underwater dynamic target. Finally, the fused motion model is used to predict the state of the underwater dynamic target through extended Kalman filtering. Therefore, this invention, in the underwater dynamic target tracking system, determines the matching degree between each motion model and the underwater dynamic target motion state through fuzzy inference and fuses them to obtain a new motion model. This results in more accurate underwater dynamic target state prediction and meets the requirements of the underwater dynamic target tracking system. Attached Figure Description

[0049] Figure 1 A flowchart for underwater dynamic target state prediction based on TS fuzzing provided in an embodiment of the present invention. Detailed Implementation

[0050] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0051] like Figure 1 As shown, the underwater dynamic target state prediction method based on TS fuzzy mapping provided in this embodiment of the invention includes:

[0052] First, the input variables of the underwater dynamic target for TS fuzzy inference are collected. Then, the fused underwater dynamic target motion model represented by Equation (6) is used to perform extended Kalman filtering at time t-1 to predict the motion state of the underwater dynamic target at time t. Then, t = t+1 is taken until the tracking ends.

[0053]

[0054] In the formula, X t X t-1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. mix B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t The relational matrix is ​​the identity matrix in this embodiment.

[0055] In one embodiment, the method for obtaining the fused underwater dynamic target motion model represented by equation (6) specifically includes:

[0056] Step 1: Based on the principle of pure azimuth underwater dynamic target tracking of AUV, select the observation equation, and then select several motion models to form a motion model set M based on the motion characteristics of underwater dynamic targets, such as maneuverability and the ability to move in straight lines and turn in three-dimensional space.

[0057] In one embodiment, several motion models for underwater dynamic targets are selected based on their motion characteristics: uniform velocity motion (CV), uniform acceleration motion (CA), and three-dimensional cooperative turning (3D-CT). Of course, those skilled in the art can add, reduce, or replace motion models based on the motion characteristics of the underwater dynamic target or actual research needs. Examples include: Current Statistical (CS) model; Singer model; Constant Turning Rate and Velocity (CTRV) model, etc.

[0058] The state transition matrix and noise gain matrix of the motion model selected in this embodiment will be explained in detail below:

[0059] A CV B CV Let A represent the state transition matrix and noise gain matrix of the uniform motion model (CV), respectively. CA B CA Let A represent the state transition matrix and noise gain matrix of the uniformly accelerated motion model (CA), respectively. CT B CT These represent the state transition matrix and noise gain matrix of the 3D cooperative turning model (3D-CT), respectively:

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] In the formula, T represents the preset sampling time, w represents the preset turning angular velocity, and I 3×3 Represents a 3x3 identity matrix, 0 3×3 This represents a zero matrix with 3 rows and 3 columns.

[0067] The observation equation can be, but is not limited to, the sonar-based measurement model described by equation (1). Through the measurement model, the state vector X of the underwater dynamic target to be predicted at time t can be obtained. t and observation Z t The relationship between them:

[0068]

[0069] In the formula, X t =[x t y t z t v xt v yt v zt [x] represents the state vector of the underwater dynamic target to be predicted at time t. t y t z t These represent the x, y, and z coordinates of the underwater dynamic target in geodetic coordinates, respectively. xt v yt v zt These represent the velocity values ​​of the underwater dynamic target along the x, y, and z axes, respectively. Represents the observable, ψ t The azimuth angle is the angle between the projection of the line connecting the underwater dynamic target and the AUV (e.g., the line connecting their centers) onto the xoy plane and the x-axis. t The pitch angle is the angle between the line connecting the underwater dynamic target and the AUV (e.g., the line connecting their centers) and the xoy plane. t The distance between the underwater dynamic target and the AUV. h(X) is the rate of change of distance. t ) is the measurement equation, v t To measure noise, which is randomly generated by the measuring equipment, it is generally assumed to be white noise with a certain variance.

[0070] Step 2: Establish TS fuzzy inference units.

[0071] In one embodiment, step 2 specifically includes:

[0072] Step 21: Determine the input variables used for TS fuzzy inference.

[0073] For example, in this embodiment, the input variables used for TS fuzzy inference include the relative acceleration and turning angular velocity of the underwater dynamic target. It should be noted that those skilled in the art can also adjust the type of input variables, such as the relative velocity of the underwater dynamic target, according to the selected set of motion models or actual research needs. Further, the relative acceleration range is set to -0.2 to 0.2 m / s². 2The turning angular velocity range is set to -1 to 1° / s.

[0074] Step 22: Set a fuzzy set for each input variable.

[0075] For example, in this embodiment, the fuzzy set for relative acceleration is set as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 .

[0076] Step 23: Set several sub-fuzzy sets for each fuzzy set.

[0077] For example, in this embodiment, the input variables are relative acceleration and turning angular velocity, and the fuzzy set of relative acceleration is set as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 Then, let S be the first fuzzy set. 1 Second fuzzy set S 2 Each subset has three fuzzy subsets: the smaller fuzzy subset NL, the medium fuzzy subset NEU, and the larger fuzzy subset PL.

[0078] Step 24: Set fuzzy rules based on the subfuzzy set and motion model.

[0079] For example, in the above embodiment, for the first fuzzy set S 1 Second fuzzy set S 2 The three fuzzy subsets set for each step and the motion model provided in the embodiment section of step 1 are used to obtain the nine fuzzy rules shown in the table below.

[0080]

[0081]

[0082] Each fuzzy subset has a corresponding membership function. The larger the membership value of the membership function, the higher the degree to which the quantity belongs to its fuzzy subset, and vice versa. Preferably, the membership function is a Gaussian function.

[0083] Step 25: Using the TS fuzzy inference method provided by the following formulas (2), (4), and (5), obtain the probabilities of each motion model selected in step 1.

[0084]

[0085] In the formula, A i B i C iLet X represent the state transition matrix, noise gain matrix, and output matrix corresponding to the known i-th fuzzy rule, respectively. t-1 Y represents the state vector of the underwater dynamic target at time t-1, u represents the process noise vector of the input variables, and Y represents the state vector of the underwater dynamic target at time t-1. t This represents the output of the motion model at time t.

[0086] For example: The i-th fuzzy rule is set as follows: If the first fuzzy set S 1 The fuzzy subset S in i1 The second fuzzy set S 2 The fuzzy subset S in i2 Therefore, the i-th fuzzy rule in the above embodiment is set as follows: if the fuzzy subset S of the relative acceleration... i1 The fuzzy subset S of the turning angular velocity i2 , then i=1, 2,...,9, u=[u a u w ] is the process noise vector, u a and u w These represent the relative acceleration and turning angular velocity process noise, respectively, expressed using a zero-mean Gaussian normal distribution.

[0087] By using single-point fuzzification, product inference, and weighted average defuzzification methods, the global fuzzy model represented by equation (3) can be obtained:

[0088]

[0089] In the formula, z(t)=[z1(t)z2(t)…z j (t)…z l [(t)] represents the antecedent variable of the fuzzy rule, z j (t) represents the j-th antecedent condition, and l represents the total number of antecedent conditions for each fuzzy rule. For example, l = 2 means that each fuzzy rule has two antecedent conditions, such as the antecedent condition in this embodiment: the fuzzy subset S in relative acceleration. i1 The fuzzy subset S in the turning angular velocity i2 A mix B mix Let δ represent the state transition matrix and input matrix after the fusion of each motion model, respectively. i (z(t)) represents the probability of each motion model in the consequent of the i-th fuzzy rule, calculated using equation (4), and satisfies the condition shown in equation (5):

[0090]

[0091]

[0092] In the formula, μ ij(z j (t) represents the preceding variable z. j The membership degree of (t).

[0093] Step 26, calculate the probabilities δ of each motion model obtained in step 25. i (z(t)) is used as the weight of each motion model in the motion model set M determined in step 1. The weights are then summed to obtain the fused state transition matrix A. mix and input matrix B mix This leads to a fused underwater dynamic target motion model, expressed as equation (6) above. Obviously, this weighted fused underwater dynamic target motion model better matches the current motion state of the underwater dynamic target, which is obviously beneficial for accurately predicting the motion state of underwater dynamic targets in three-dimensional space.

[0094] In one embodiment, the first step of extended Kalman filtering using the fused underwater dynamic target motion model, represented by equation (6), predicts the prior target state at time t. The motion model is expressed as equation (7):

[0095] X t =A mix *X t-1 +B mix *u (7)

[0096] Step 3: Apply extended Kalman filtering to the fused underwater dynamic target motion model represented by equation (6) at time t-1 to estimate the posterior estimates of the underwater dynamic target's state variables at time t. Then take t = t + 1, and repeat steps 2 and 3 until the tracking ends.

[0097] The underwater dynamic target state prediction device based on TS fuzzy mapping provided in this embodiment of the invention includes an input acquisition module and an underwater dynamic target tracking module, wherein:

[0098] The input acquisition module is used to acquire the input variables of underwater dynamic targets at time t-1 for TS fuzzy inference.

[0099] The underwater dynamic target tracking module is used to perform extended Kalman filtering at time t-1 using the fused underwater dynamic target motion model represented by the above formula (6), to predict the motion state of the underwater dynamic target at time t, and then take t=t+1 until the tracking ends.

[0100] In one embodiment, the fused underwater dynamic target motion model represented by equation (6) is obtained through a TS fuzzy inference unit. The TS fuzzy inference unit specifically includes an input variable determination module, a fuzzy set setting module, a subfuzzy set setting module, a fuzzy rule setting module, and a motion model probability calculation module, wherein:

[0101] The input variable determination module is used to determine the input variables used for TS fuzzy inference.

[0102] The fuzzy set setting module is used to set a fuzzy set for each input variable.

[0103] The sub-fuzzy set setting module is used to set several sub-fuzzy sets for each fuzzy set.

[0104] The fuzzy rule setting module is used to set fuzzy rules based on the subfuzzy set and the motion model.

[0105] The motion model probability calculation module is used to obtain the probability δ of each motion model in the i-th fuzzy rule by employing the TS fuzzy inference method provided by equations (2), (4), and (5) above. i (z(t)).

[0106] In one embodiment, the input variables include the relative acceleration and turning angular velocity of the underwater dynamic target, and the motion models are uniform motion model (CV), uniform acceleration motion model (CA), and three-dimensional cooperative turning model (3D-CT), respectively. The fuzzy set setting module sets the fuzzy set of the relative acceleration as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 The sub-fuzzy set setting module is the first fuzzy set S. 1 Second fuzzy set S 2 Each is set with 3 fuzzy subsets: negative large fuzzy subset NL, medium fuzzy subset NEU, and positive large fuzzy subset PL. The fuzzy rule setting module obtains the 9 fuzzy rules shown in the table above based on the subfuzzy subsets and the motion model.

[0107] This invention addresses the problem of pure orientation dynamic underwater target tracking by AUVs. It uses fuzzy logic reasoning to fuse multiple models and obtain a suitable model to predict the underwater dynamic target's state in real time based on the fused underwater dynamic target's state, which helps improve prediction accuracy.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Those skilled in the art should understand that modifications can be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the state of underwater dynamic targets based on TS fuzzy mapping, characterized in that, include: First, the input variables for TS fuzzy inference of the underwater dynamic target at time t-1 are collected. Then, the fused underwater dynamic target motion model represented by Equation (6) is used to perform extended Kalman filtering at time t-1 to predict the motion state of the underwater dynamic target at time t. Then, t = t+1 is taken until the tracking ends. In the formula, X t X t-1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. min B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t Relationship matrix; The method for obtaining the fused underwater dynamic target motion model, as expressed in equation (6), specifically includes: Step 1: Based on the principle of pure azimuth underwater dynamic target tracking of AUV, select the observation equation, and then select several motion models based on the motion characteristics of underwater dynamic targets. Step 2, establish the TS fuzzy inference unit, which specifically includes: Step 21: Determine the input variables used for TS fuzzy inference; Step 22: Set a fuzzy set for each input variable; Step 23: Set several sub-fuzzy sets for each fuzzy set; Step 24: Set fuzzy rules based on the subfuzzy set and motion model; Step 25: Using the TS fuzzy inference method provided by equations (2), (4) and (5), obtain the probability δ of each motion model in the i-th fuzzy rule. i (z(t)): δ i (z(t))>0, In the formula, A i B i C i Let z(t) represent the state transition matrix, noise gain matrix, and output matrix corresponding to the known i-th fuzzy rule, respectively, and z(t) = [z1(t) z2(t)…z j (t)…z l [(t)] represents the antecedent variable of the fuzzy rule, z j (t) represents the j-th antecedent condition, l represents the total number of antecedent conditions for each fuzzy rule, and μ ij (z j (t) represents the preceding variable z. j Membership degree of (t); Step 26, calculate the probabilities δ of each motion model obtained in step 25. i (z(t)) is used as the weight of each motion model, and the weights are summed to obtain the fused state transition matrix A. mix and input matrix B mix Thus, the fused underwater dynamic target motion model described by equation (6) is obtained.

2. The underwater dynamic target state prediction method based on TS fuzzy mapping as described in claim 1, characterized in that, The parameters in the fused underwater dynamic target motion model, represented by equation (6), are obtained through equation (3):

3. The underwater dynamic target state prediction method based on TS fuzzy mapping as described in claim 2, characterized in that, The input variables include the relative acceleration and turning angular velocity of the underwater dynamic target. The motion models selected in step 1 are uniform motion model (CV), uniform acceleration motion model (CA), and three-dimensional cooperative turning model (3D-CT). In step 22, the fuzzy set of the relative acceleration is set as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 , is the first fuzzy set S 1 Second fuzzy set S 2 Each subset is set up with 3 fuzzy subsets: negative large fuzzy subset NL, medium fuzzy subset NEU, and positive large fuzzy subset PL. Step 24 obtains the 9 fuzzy rules shown in the table below based on the subfuzzy subsets and the motion model:

4. A device for predicting the state of underwater dynamic targets based on TS fuzzy mapping, characterized in that, include: The input acquisition module is used to acquire the input variables of the underwater dynamic target at time t-1 for TS fuzzy inference; The underwater dynamic target tracking module uses the fused underwater dynamic target motion model represented by equation (6) to perform extended Kalman filtering at time t-1 to predict the motion state of the underwater dynamic target at time t, and then takes t=t+1 until the tracking ends: In the formula, X t X t-1 Let Y and Y represent the state vectors of the underwater dynamic target at times t and t-1, respectively, and u represent the process noise vector of the input variable. t A represents the output vector of the motion model at time t. mix B mix These are the state transition matrix and the input matrix, respectively, and C represents the pre-set output vector Y. t With state vector X t Relationship matrix; The fused underwater dynamic target motion model represented by Equation (6) is obtained through the TS fuzzy inference unit, which specifically includes: The input variable determination module is used to determine the input variables used for TS fuzzy inference; The fuzzy set setting module is used to set a fuzzy set for each input variable; The sub-fuzzy set setting module is used to set several sub-fuzzy sets for each fuzzy set; The fuzzy rule setting module is used to set fuzzy rules based on the subfuzzy set and the motion model; The motion model probability calculation module is used to obtain the probability δ of each motion model in the i-th fuzzy rule by employing the TS fuzzy inference method provided by equations (2), (4), and (5). i (z(t)): δ i (z(t))>0, In the formula, A i B i C i Let z(t) represent the state transition matrix, noise gain matrix, and output matrix corresponding to the known i-th fuzzy rule, respectively, and z(t) = [z1(t) z2(t)…z j (t)…z l [(t)] represents the antecedent variable of the fuzzy rule, z j (t) represents the j-th antecedent condition, l represents the total number of antecedent conditions for each fuzzy rule, and μ ij (z j (t) represents the preceding variable z. j The membership degree of (t).

5. The underwater dynamic target state prediction device based on TS fuzzy mapping as described in claim 4, characterized in that, The parameters in the fused underwater dynamic target motion model, represented by equation (6), are obtained through equation (3):

6. The underwater dynamic target state prediction device based on TS fuzzy mapping as described in claim 5, characterized in that, The input variables include the relative acceleration and turning angular velocity of the underwater dynamic target. The motion models are uniform motion model (CV), uniform acceleration motion model (CA), and three-dimensional cooperative turning model (3D-CT), respectively. The fuzzy set setting module sets the fuzzy set of relative acceleration as the first fuzzy set S. 1 Set the fuzzy set corresponding to the turning angular velocity as the second fuzzy set S. 2 The sub-fuzzy set setting module is the first fuzzy set S. 1 Second fuzzy set S 2 Each subset is set with three fuzzy subsets: negative large fuzzy subset NL, medium fuzzy subset NEU, and positive large fuzzy subset PL. The fuzzy rule setting module obtains the nine fuzzy rules shown in the table below based on the subfuzzy subsets and the motion model:

Images