An anti-UAV guidance information extraction method

Abstract

The application discloses a kind of anti-UAV guidance information extraction methods, comprising the following steps: S1, establish anti-UAV guidance system, set system state model, system observation model, the system state model includes acceleration submodel, Singer submodel and sinusoidal submodel;S2, by total strapdown seeker measurement target, obtain line-of-sight angle information, relative distance information, by airborne sensor obtains unmanned plane acceleration information, above-mentioned information is as the observation value of system observation model;S3, by filter to anti-UAV guidance system is filtered, respectively obtains acceleration submodel, Singer submodel and sinusoidal submodel filter result;S4, the filter result of submodel is fused and estimated, obtains final guidance information.The anti-UAV guidance information extraction method disclosed in the application obtains unmanned plane guidance information error, and accuracy is high.

Description

TECHNICAL FIELD

[0001] The present application relates to a method for extracting anti-UAV guidance information, and belongs to the field of sensor bias estimation. BACKGROUND

[0002] Due to the characteristics of flexibility, versatility and low price, unmanned aerial vehicles (UAVs) have been widely used in various fields. However, the misuse of UAVs has also posed a certain threat to public safety and economy. At present, the main countermeasures against non-cooperative invading UAVs are to use laser weapons and mechanical capture of UAVs. This usually requires some complex and expensive active sensors to measure the required guidance information, and the size and weight of these active sensors are difficult to be used for the growing small UAVs. In order to simplify the hardware system, light passive sensors such as total strapdown seeker have been widely concerned and applied in the field of anti-UAV due to their lower cost, simpler structure and higher reliability.

[0003] However, the total strapdown seeker fixed on the UAV body can only measure the body sight line angle information coupled with the body attitude, and cannot directly provide the inertial line of sight (LOS) rate information and target motion state information required for guidance. Moreover, due to the wide instantaneous field of view of the total strapdown seeker, measurement noise is inevitably introduced, which will be amplified when performing differential operation, resulting in reduced accuracy of the provided information.

[0004] In addition, the EKF algorithm and the single motion model of the target are widely used in engineering to extract the guidance information of anti-UAV, and this extraction method has many defects:

[0005] (1) When the dimension of the state quantity is high, the EKF algorithm needs to calculate the high-dimensional Jacobian matrix, the calculation complexity will increase greatly, the stability of the filter will decrease, and the estimation error will increase;

[0006] (2) The non-cooperative invading UAV is flexible and maneuverable, and its motion mode is difficult to predict. In the currently used filtering algorithm, the target is often assumed to move at a constant speed, and there is little research on maneuvering targets. In fact, the maneuvering of the target will make the constructed constant speed motion model inaccurate, thereby affecting the accuracy of the estimated inertial line of sight rate. Some guidance laws for maneuvering targets also need to accurately estimate the acceleration information of the target. Therefore, for non-cooperative invading UAVs, introducing a suitable target motion model can more effectively predict the target motion mode and reduce the estimation error of the extracted guidance information.

[0007] Therefore, it is necessary to propose an anti-UAV guidance information extraction method to solve the above problems. SUMMARY

[0008] In order to overcome the above problems, the present inventors have made an in-depth study and designed an anti-UAV guidance information extraction method, comprising the following steps:

[0009] S1, an anti-UAV guidance system is established, a system state model and a system observation model are set, the system state model comprises an acceleration sub-model, a Singer sub-model and a sine sub-model;

[0010] S2, the target is measured by a total strapdown seeker to obtain line-of-sight angle information and relative distance information, and UAV acceleration information is obtained by an airborne sensor, and the above information is taken as the observation value of the system observation model;

[0011] S3, the anti-UAV guidance system is filtered by a filter to obtain the filtering results corresponding to the acceleration sub-model, the Singer sub-model and the sine sub-model;

[0012] S4, the filtering results corresponding to the sub-models are fused and estimated to obtain the final guidance information.

[0013] According to an preferred embodiment of the present application, in S1, the state variable of the system state model is anti-UAV guidance information, which is expressed as:

[0014]

[0015] Wherein, q y represents the line-of-sight angle of the target relative to the pitch direction of the total strapdown seeker, represents the line-of-sight angle rate of the target relative to the pitch direction of the total strapdown seeker, q z represents the line-of-sight angle of the target relative to the yaw direction of the total strapdown seeker, represents the line-of-sight angle rate of the target relative to the yaw direction of the total strapdown seeker, r represents the relative distance between the target and the total strapdown seeker, represents the relative velocity between the target and the total strapdown seeker, a ux a uy a uz represents the acceleration of the aircraft in which the total strapdown seeker is located in three directions of the line-of-sight coordinate system, a tx a ty a tz represents the acceleration of the target in three directions of the line-of-sight coordinate system.

[0016] According to an preferred embodiment of the present application, the system state model is expressed as:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] where τ is the acceleration time constant, ω ux ,ω uy ,ω uz denote the uncertainty of the acceleration of the all-inertial seeker in three directions of the line-of-sight coordinate system, ω tx ,ω ty ,ω tz denote the uncertainty of the acceleration of the target in three directions of the line-of-sight coordinate system, and t denotes time.

[0023] Formula (three) is an acceleration sub-model, formula (four) is a Singer sub-model, and formula (five) is a sine sub-model.

[0024] According to a preferred embodiment of the present application, the measurement of the system observation model is:

[0025]

[0026] The system observation model is expressed as:

[0027]

[0028] wherein, wherein, is the line-of-sight angle of the pitch direction measured by the all-inertial seeker, is the line-of-sight angle of the yaw direction measured by the all-inertial seeker, denotes the relative distance between the target and the seeker measured by the all-inertial seeker, denotes the acceleration of the all-inertial seeker in three directions of the line-of-sight coordinate system measured by the airborne sensor;

[0029] denotes the relative distance between the target and the seeker measured by the all-inertial seeker, denotes the acceleration of the all-inertial seeker in three directions of the line-of-sight coordinate system measured by the airborne sensor, v1 is the angle measurement noise of the all-inertial seeker in the pitch direction, v2 is the angle measurement noise of the all-inertial seeker in the yaw direction, v3 is the distance measurement noise of the all-inertial seeker, and v4, v5, and v6 are the acceleration measurement noises of the aircraft in which the all-inertial seeker is located in three directions of the coordinate system;

[0030] R ij (i = 1, 2, 3; j = 1, 2, 3) are corresponding elements of the matrix L(γ, θ, ψ),

[0031]

[0032] Wherein, variable γ is the roll angle of the aircraft where the total strapdown seeker is located, θ is the pitch angle of the aircraft where the total strapdown seeker is located, and ψ is the yaw angle of the aircraft where the total strapdown seeker is located.

[0033] According to a preferred embodiment of the present application, in S3, the filter is a cubature Kalman filter.

[0034] According to a preferred embodiment of the present application, S4 comprises the following sub-steps:

[0035] S41, setting a transition probability matrix between sub-models to obtain a sub-model mixing probability;

[0036] S42, according to the sub-model mixing probability, mixing the output result of any sub-model with the output results of other sub-models, taking the mixed result and the observation at the next moment as the input of the filter of the sub-model, and obtaining the estimation result at the next moment through the filter;

[0037] S43, updating the probability of different sub-models;

[0038] S44, obtaining the comprehensive state estimation result at the next moment by weighting the estimation results obtained by the sub-model filters according to the obtained sub-model probabilities;

[0039] S45, repeating S42-S44 to obtain the comprehensive state estimation result at the latest moment, and taking the estimation result as the anti-UAV guidance information.

[0040] According to a preferred embodiment of the present application, in S41, the transition probability matrix between sub-models is expressed as:

[0041]

[0042] The sub-model mixing probability is expressed as:

[0043]

[0044] Wherein, represents the element of the i-th row and the j-th column of the mixing probability matrix at the k-1 moment, π ij represents the element of the i-th row and the j-th column of the transition probability matrix π, represents the sub-model probability corresponding to the i-th filter.

[0045] According to a preferred embodiment of the present application, in S42, the mixed result of the output result of any sub-model with the output results of other sub-models is expressed as:

[0046]

[0047]

[0048] wherein, represents the mixed estimation value of the jth sub-model at k-1 time, represents the estimation value of the ith sub-model at k-1 time.

[0049] represents the mixed estimation error covariance matrix of the jth sub-model at k-1 time, represents the estimation error covariance matrix of the ith sub-model at k-1 time, superscript T represents the transpose of a matrix.

[0050] According to a preferred embodiment of the present application, in S43, the probability of the jth sub-model is updated as:

[0051]

[0052] wherein, is the likelihood probability of the jth sub-model at k time, represented as:

[0053]

[0054]

[0055]

[0056] wherein, represents the filter innovation of the jth sub-model at k time, represents the covariance matrix of the filter innovation, z k represents the observation at k time, represents the observation matrix at k time, represents the predicted estimation value of the jth sub-model at k time, represents the predicted error covariance matrix at k time, superscript T represents the transpose of a matrix, R k represents the noise variance matrix.

[0057] According to a preferred embodiment of the present application, in S44, the comprehensive state estimation result at k time is represented as:

[0058]

[0059]

[0060] wherein, represents the comprehensive state estimation value at k time, represents the state estimation value output by the jth sub-model at k time,

[0061] represents the integrated state estimation error covariance at time k,

[0062] represents the error covariance matrix of the i-th sub-model at time k.

[0063] The present application has the beneficial effects including:

[0064] (1) According to the anti-UAV guidance information extraction method provided by the present application, the obtained UAV guidance information error is lower and the accuracy is higher;

[0065] (3) According to the anti-UAV guidance information extraction method provided by the present application, the calculation amount is low, and the on-board computer can meet the operation requirement without additional investment cost. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 The figure shows the flow chart of the anti-UAV guidance information extraction method according to a preferred embodiment of the present application;

[0067] Figure 2 The line-of-sight angle error comparison of the pitch direction between Example 1 and Comparative Example 1 is shown;

[0068] Figure 3 The line-of-sight angle error comparison of the yaw direction between Example 1 and Comparative Example 1 is shown;

[0069] Figure 4 The line-of-sight angle rate error comparison of the pitch direction between Example 1 and Comparative Example 1 is shown,

[0070] Figure 5 The line-of-sight angle rate error comparison of the yaw direction between Example 1 and Comparative Example 1 is shown;

[0071] Figure 6 The average acceleration error comparison between Example 1 and Comparative Example 1 is shown. DETAILED DESCRIPTION

[0072] The present application will be further described in detail by the accompanying drawings and examples. Through these descriptions, the features and advantages of the present application will become more apparent.

[0073] The word "exemplary" here means "serving as an example, an implementation, or illustration". Any implementation described as "exemplary" is not necessarily to be construed as preferred or advantageous over other implementations. The drawings are not necessarily drawn to scale unless otherwise indicated.

[0074] According to the anti-UAV guidance information extraction method provided by the present application, the following steps are included:

[0075] S1, establish an anti-UAV guidance system, set a system state model, a system observation model, the system state model includes an acceleration sub-model, a Singer sub-model and a sinusoidal sub-model;

[0076] S2, measure the target through a total strapdown seeker, obtain line-of-sight angle information and relative distance information, obtain UAV acceleration information through an airborne sensor, and take the above information as observation values of the system observation model;

[0077] S3, filter the anti-UAV guidance system through a filter to obtain filtering results corresponding to the acceleration sub-model, the Singer sub-model and the sinusoidal sub-model;

[0078] S4, state fusion estimation is performed on the filtering results of the sub-models to obtain final guidance information.

[0079] In S1, the state variable of the system state model is anti-UAV guidance information, which is represented as:

[0080]

[0081] wherein q y represents the line-of-sight angle of the target relative to the pitch direction of the total strapdown seeker represents the line-of-sight angle rate of the target relative to the pitch direction of the total strapdown seeker, q z represents the line-of-sight angle of the target relative to the yaw direction of the total strapdown seeker, represents the line-of-sight angle rate of the target relative to the yaw direction of the total strapdown seeker, r represents the relative distance between the target and the total strapdown seeker, represents the relative velocity between the target and the total strapdown seeker, a ux , a uy , a uz represents the acceleration of the aircraft on which the total strapdown seeker is located in three directions of the line-of-sight coordinate system, a tx , a ty , a tz represents the acceleration of the target in three directions of the line-of-sight coordinate system

[0082] The system state model is represented as:

[0083]

[0084]

[0085]

[0086]

[0087]

[0088] where τ is the acceleration time constant, ω ux ,ω uy ,ω uz denote the uncertainty of the acceleration of the total strapdown seeker in three directions of the line-of-sight coordinate system, ω tx ,ω ty ,ω tz denote the uncertainty of the acceleration of the target in three directions of the line-of-sight coordinate system, and t denotes time.

[0089] Further, the statistical characteristics of ω tx ,ω ty ,ω tz may be expressed as:

[0090] E[ω(t)]=0

[0091] E[ω(t1)ω(t2)]=Qδ(t1-t2)

[0092] where E, δ, and Q respectively denote an expectation operator, a Dirac function, and a power spectral density, t1 and t2 denote any different time points, and t1>t2.

[0093] Further, in the system state model, formula (I) is obtained through a target kinematics model, formula (II) is obtained through the uncertainty of acceleration measurement of the total strapdown seeker, formula (III) is an acceleration submodel, formula (IV) is a Singer submodel, and formula (V) is a sinusoidal submodel. According to the present application, by using the constant acceleration model, the Singer model, and the sinusoidal model as submodels, the problem that the motion state is difficult to predict due to the flexible maneuver of the unmanned aerial vehicle is solved, so that the system state model with the three submodels introduced can more effectively predict the target motion mode, and the estimation error of the extracted guidance information is reduced.

[0094] Further, in S1, the target kinematics model is expressed as:

[0095]

[0096] According to the present application, in the system observation model, the measurement quantity is set as:

[0097]

[0098] The system observation model is expressed as:

[0099]

[0100] where, is the line-of-sight angle of the pitch direction measured by the total strapdown seeker, is the line-of-sight angle of the yaw direction measured by the total strapdown seeker, represents the relative distance between the target and the seeker measured by the total strapdown seeker, represents the acceleration of the total strapdown seeker in three directions of the line-of-sight coordinate system measured by the on-board sensor;

[0101] v1 is the angle measurement noise of the total strapdown seeker in the pitch direction, v2 is the angle measurement noise of the total strapdown seeker in the yaw direction, v3 is the distance measurement noise of the total strapdown seeker, and v4, v5 and v6 are the acceleration measurement noises of the aircraft in which the total strapdown seeker is located in three directions of the coordinate system;

[0102] R ij (i = 1, 2, 3; j = 1, 2, 3) are corresponding elements of the matrix L(γ, θ, ψ),

[0103]

[0104] wherein the variable γ is the roll angle of the aircraft in which the total strapdown seeker is located, θ is the pitch angle of the aircraft in which the total strapdown seeker is located, and ψ is the yaw angle of the aircraft in which the total strapdown seeker is located.

[0105] In S3, the filter is a cubature Kalman filter. Through the cubature Kalman filter, the nonlinear filtering problem can be converted into a multi-dimensional integration problem of the product of a nonlinear function and a Gaussian probability density.

[0106] Through the filter, the estimated value and the estimated error covariance of the next moment of the sub-model can be obtained.

[0107] Specifically, after the system state model and the system observation model are arranged in the following discrete form, estimation is performed through the cubature Kalman filter:

[0108]

[0109] wherein x k , z k represent the state quantity and the observation quantity of the system at the k moment respectively, f(·) and h(·) represent the state nonlinear function and the measurement nonlinear function respectively;

[0110] ω k-1 represents the process noise at the k-1 moment, is a zero-mean Gaussian white noise, and the variance is Q k-1 ;

[0111] v k represents the observation noise at the k moment, is a zero-mean Gaussian white noise, and the variance is R k .

[0112] According to the present application, S4 comprises the following sub-steps:

[0113] S41, set a transition probability matrix between sub-models, and obtain a sub-model mixing probability;

[0114] S42, according to the sub-model mixing probability, mix the output result of any sub-model with the output results of other sub-models, take the mixed result and an observation at a next moment as inputs of a filter of the sub-model, and obtain an estimation result at the next moment through the filter;

[0115] S43, update the probabilities of different sub-models;

[0116] S44, obtain a comprehensive state estimation result at the next moment according to the estimation result obtained by weighting the sub-model filters according to the obtained sub-model probabilities;

[0117] S45, repeat S42-S44 to obtain a comprehensive state estimation result at a new moment, and take the estimation result as anti-UAV guidance information.

[0118] In S41, the transition probability matrix π between sub-models is represented as:

[0119]

[0120] The above values are obtained by the inventors according to experience,

[0121] The sum of the transition probabilities of each sub-model can meet the requirement that the sum is 1, and the main diagonal element value of the matrix is larger, which conforms to the general rule of target state transition.

[0122] The sub-model mixing probability is represented as:

[0123]

[0124] Wherein, represents an element in the i-th row and the j-th column of the mixing probability matrix at the k-1 moment, π ij represents an element in the i-th row and the j-th column of the transition probability matrix π, represents a sub-model probability corresponding to the i-th filter.

[0125] In S42, the result after mixing the output result of any sub-model with the output results of other sub-models is represented as:

[0126]

[0127] Wherein, represents a mixed estimation value of the j-th sub-model at the k-1 moment, represents an estimation value of the i-th sub-model at the k-1 moment.

[0128] represents a mixed estimation error covariance matrix of the j-th sub-model at the k-1 moment, denotes the estimation error covariance matrix of the i-th sub-model at time k-1, superscript T denotes the transpose of a matrix.

[0129] The estimation error covariance matrix P and the observation z k as the input of the j-th filter, and the output result of the filter at time k is obtained

[0130] In S43, the probability of the j-th sub-model is updated as:

[0131]

[0132] wherein, is the likelihood probability of the j-th sub-model at time k, and is expressed as:

[0133]

[0134] wherein, denotes the filter innovation of the j-th sub-model at time k, denotes the covariance matrix of the filter innovation, z k denotes the observation at time k, denotes the observation matrix at time k, denotes the predicted estimation value of the j-th sub-model at time k, denotes the predicted error covariance matrix at time k, superscript T denotes the transpose of a matrix, R k denotes the noise variance matrix.

[0135] In S44, the comprehensive state estimation result at time k is expressed as:

[0136]

[0137] wherein, denotes the comprehensive state estimation value at time k, denotes the state estimation value output by the j-th sub-model at time k,

[0138] denotes the comprehensive state estimation error covariance

[0139] denotes the error covariance matrix of the i-th sub-model at time k.

[0140] Embodiment

[0141] Embodiment 1

[0142] Simulation experiments are performed, and the anti-UAV guidance information extraction is performed in the following manner:

[0143] S1, establish an anti-UAV guidance system, set a system state model and a system observation model, the system state model comprising an acceleration sub-model, a Singer sub-model and a sinusoidal sub-model;

[0144] S2, measure a target through a total strapdown seeker to obtain line-of-sight angle information and relative distance information, and obtain UAV acceleration information through an airborne sensor, and take the information as observation values of the system observation model;

[0145] S3, filter the anti-UAV guidance system through a filter to obtain filtering results corresponding to the acceleration sub-model, the Singer sub-model and the sinusoidal sub-model;

[0146] S4, fuse and estimate the filtering results corresponding to the sub-models to obtain final guidance information.

[0147] In S1, a state variable of the system state model is anti-UAV guidance information, and is expressed as:

[0148]

[0149] The system state model is expressed as:

[0150]

[0151]

[0152]

[0153]

[0154]

[0155] In the system observation model, the measurement is:

[0156]

[0157] The system observation model is expressed as:

[0158]

[0159] R ij (i = 1, 2, 3; j = 1, 2, 3) are corresponding elements of the matrix L (γ, θ, ψ),

[0160]

[0161] In S3, the filter is a cubature Kalman filter.

[0162] S4 comprises the following sub-steps:

[0163] S41, a transition probability matrix between sub-models is set, and a sub-model mixing probability is obtained;

[0164] S42, according to the sub-model mixing probability, for any sub-model, an output result of the sub-model is mixed with output results of other sub-models, the mixed result and an observation at a next moment are taken as inputs of a filter of the sub-model, and an estimation result at the next moment is obtained through the filter;

[0165] S43, probabilities of different sub-models are updated;

[0166] S44, an estimation result obtained by weighting the sub-model filter according to the obtained sub-model probability is taken as a comprehensive state estimation result at a next moment;

[0167] S45, S42-S44 are repeated, a comprehensive state estimation result at a latest moment is obtained, and the estimation result is taken as anti-UAV guidance information.

[0168] In S41, the transition probability matrix between sub-models is represented as:

[0169]

[0170] The sub-model mixing probability is represented as:

[0171]

[0172] In S42, the mixed result of any sub-model output result and output results of other sub-models is represented as:

[0173]

[0174] In S43, the probability of the jth sub-model is updated as:

[0175]

[0176] wherein, is a likelihood probability of the jth sub-model at the kth moment, and is represented as:

[0177]

[0178] In S44, the comprehensive state estimation result at the kth moment is represented as:

[0179]

[0180] In the simulation experiment, a sampling period of the sensor is set as T s =0.01s, and the measurement noise is a high Gaussian white noise with a 0 mean value, v1~N(0,(0.1×deg) 2), v2 ~ N(0, (0.1 x deg) 2 ), v3 ~ N(0, (1 x m) 2 ), v 4,5,6 ~ N(0, (0.1 x m / s 2 ) 2 ) Process noise is set as ω ux,uy,uz ~ N(0, (0.05 x m / s 2 ) 2 ), ω tx,ty,tz ~ N(0, (0.25 x m / s 2 ) 2 ) The initial error proportion of system state quantity is ρ = 0.2, and the initial covariance The initial probability of the model is set as The target UAV makes a sinusoidal maneuver in the air, and the acceleration expression thereof in the inertial system is as follows:

[0181]

[0182] Comparative Example 1

[0183] The same simulation experiment as in Example 1 is performed, except that step S4 is not performed, and in S3, the extended Kalman filter is directly used to filter the anti-UAV guidance system, and the obtained filtering result is taken as the final guidance information.

[0184] Experimental Example

[0185] The results of Example 1 and Comparative Example 1 are compared, as shown in Figures 2 to 6 and Table 1.

[0186] Among them, Figure 2 shows the line-of-sight angle error comparison between Example 1 and Comparative Example 1 in the pitch direction,

[0187] Figure 3 shows the line-of-sight angle error comparison between Example 1 and Comparative Example 1 in the yaw direction;

[0188] Figure 4 shows the line-of-sight angular rate error comparison between Example 1 and Comparative Example 1 in the pitch direction,

[0189] Figure 5 shows the line-of-sight angular rate error comparison between Example 1 and Comparative Example 1 in the yaw direction;

[0190] Figure 6 shows the average acceleration error comparison between Example 1 and Comparative Example 1.

[0191] Table 1 shows the average absolute error of the results in Example 1 and Comparative Example 1.

[0192] Table 1

[0193]

[0194] From Figures 2 to 6 As can be seen from Table 1, the estimation error of the state quantity obtained in Example 1 is lower, especially the average error of target acceleration, which is reduced from 1.28 m / s 2 in Comparative Example 1 to 0.43 m / s 2 , indicating that the method in Example 1 can more effectively realize the extraction of anti-UAV guidance information.

[0195] In the description of the present application, it should be noted that the terms "upper", "lower", "inner", "outer", "front", "back" and the like indicate the orientation or positional relationship in the working state of the present application, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second", "third", "fourth" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0196] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0197] The above describes the present application in combination with the preferred embodiments, but these embodiments are only exemplary and serve only to illustrate. On this basis, various substitutions and improvements can be made to the present application, which all fall within the scope of protection of the present application.

Claims

1. A method for anti-UAV guidance information extraction, characterized in that, The method comprises the following steps: S1, establishing an anti-UAV guidance system, setting a system state model and a system observation model, the system state model comprising an acceleration sub-model, a Singer sub-model and a sinusoidal sub-model; S2, measuring a target through a total strapdown seeker to obtain line-of-sight angle information and relative distance information, and obtaining UAV acceleration information through an airborne sensor, and taking the information as observation values of the system observation model; S3, filtering the anti-UAV guidance system through a filter to obtain filtering results corresponding to the acceleration sub-model, the Singer sub-model and the sinusoidal sub-model; S4, fusing and estimating the filtering results corresponding to the sub-models to obtain final guidance information.

2. The anti-UAV guidance information extraction method according to claim 1, wherein in S1, a state variable of the system state model is anti-UAV guidance information, and is expressed as:

3. The anti-UAV guidance information extraction method according to claim 1, wherein the system state model is expressed as: where q y represents the line-of-sight angle of the target relative to the pitch direction of the total strapdown seeker, represents the line-of-sight angle rate of the target relative to the pitch direction of the total strapdown seeker, q z represents the line-of-sight angle of the target relative to the yaw direction of the total strapdown seeker, represents the line-of-sight angle rate of the target relative to the yaw direction of the total strapdown seeker, r represents the relative distance between the target and the total strapdown seeker, represents the relative velocity between the target and the total strapdown seeker, a ux , a uy , a uz represents the acceleration of the aircraft on which the total strapdown seeker is located in three directions of the line-of-sight coordinate system, a tx , a ty , a tz represents the acceleration of the target in three directions of the line-of-sight coordinate system. Formula (3) is the acceleration sub-model, formula (4) is the Singer sub-model, and formula (5) is the sinusoidal sub-model.

4. The anti-UAV guidance information extraction method according to claim 1, wherein in the system observation model, a measurement is: where τ is the acceleration time constant, ω ux ,ω uy ,ω uz denote the uncertainty of the acceleration of the total strapdown seeker in three directions of the line-of-sight coordinate system, ω tx ,ω ty ,ω tz denote the uncertainty of the acceleration of the target in three directions of the line-of-sight coordinate system, and t denotes time. The system observation model is expressed as: v1 is a pitch direction angle measurement noise of the total strapdown seeker, v2 is a yaw direction angle measurement noise of the total strapdown seeker, v3 is a distance measurement noise of the total strapdown seeker, and v4, v5 and v6 are acceleration measurement noises of an aircraft on which the total strapdown seeker is located in three directions of a coordinate system.

5. The anti-UAV guidance information extraction method according to claim 1, wherein in S3, the filter is a cubature Kalman filter.

6. The anti-UAV guidance information extraction method according to claim 1, wherein S4 comprises the following sub-steps: wherein is the line-of-sight angle in the pitch direction measured by the total strapdown seeker, is the line-of-sight angle in the yaw direction measured by the total strapdown seeker, denotes the relative distance between the target and the seeker measured by the total strapdown seeker, denotes the accelerations of the total strapdown seeker in the three directions of the line-of-sight coordinate system measured by the on-board sensors; S41, setting a transition probability matrix between sub-models to obtain a sub-model mixing probability; R ij (i = 1,2,3; j = 1,2,3) are the corresponding elements of the matrix L(γ,θ,ψ), Wherein, variable γ is the roll angle of the aircraft where the total strapdown seeker is located, is the pitch angle of the aircraft where the total strapdown seeker is located, and ψ is the yaw angle of the aircraft where the total strapdown seeker is located. S42, according to the sub-model mixing probability, mixing an output result of any sub-model with output results of other sub-models, taking the mixed result and a next time observation as inputs of a filter of the sub-model, and obtaining an estimation result at the next time through the filter; S43, updating probabilities of different sub-models; S44, obtaining a comprehensive state estimation result at the next time by weighting estimation results obtained by sub-model filters according to the obtained sub-model probabilities; S45, repeating S42-S44 to obtain a comprehensive state estimation result at the latest time, and taking the estimation result as anti-UAV guidance information.

7. The anti-UAV guidance information extraction method according to claim 6, wherein in S41, the transition probability matrix between sub-models is expressed as: The sub-model mixing probability is expressed as:

8. The anti-UAV guidance information extraction method according to claim 6, wherein in S42, a result after mixing an output result of any sub-model with output results of other sub-models is expressed as:

9. The anti-UAV guidance information extraction method according to claim 6, wherein ​ ​ ​ ​ wherein, denotes the element of the mixture probability matrix in the i-th row and j-th column at time k-1, π ij denotes the element of the transition probability matrix π in the i-th row and j-th column, denotes the sub-model probability corresponding to the i-th filter. ​ ​ wherein, denotes the mixed estimate of the jth sub-model at time k-1, denotes the estimate of the ith sub-model at time k-1, denotes the mixed estimation error covariance matrix of the jth sub-model at time k - 1, denotes the estimation error covariance matrix of the ith sub-model at time k - 1, with superscript T denotes the transpose of a matrix. ​ In S43, the probability of the jth sub-model is updated as: wherein, is the likelihood probability of the jth sub-model at time k, denoted as: wherein, denotes the filter innovation of the jth sub-model at time k, denotes the covariance matrix of the filter innovation, z k denotes the observation at time k, denotes the observation matrix at time k, denotes the predicted estimate of the jth sub-model at time k, denotes the predicted error covariance matrix at time k, superscript T denotes the matrix transpose, R k denotes the noise variance matrix.

10. The anti-UAV guidance information extraction method of claim 6, wherein, In S44, the comprehensive state estimation result at the kth moment is represented as: wherein denotes the aggregated state estimate at time k, denotes the state estimate output by the jth sub-model at time k, represents the overall state estimation error covariance at time k, represents the error covariance matrix of the i-th sub-model at time k.

Images