Information fusion method and system for infrared imaging and ranging integrated fuze based on BAS improved particle filter

Abstract

The present invention proposes an infrared imaging and ranging integrated fuze information fusion method and system based on a BAS-improved particle filter, belonging to the technical field of air defense missile detonation precision control. To address the problems of complex nonlinear filtering algorithm processing and low accuracy of detonation control parameter estimation in the current prior art, the present invention facilitates a forward-detection missile-target intersection model by combining an infrared imaging detector with a laser rangefinder to obtain angle and distance information during the missile-target intersection process. This information is then fused and processed by an information fusion center, and the particle filter algorithm is optimized using the BAS, significantly reducing the number of particles used in the particle filter and the algorithm's time complexity. By filtering data sample particles based on the BAS-improved particle filter algorithm, the present invention can obtain the detonation delay time and angle information required for precise initial control, thereby obtaining more accurate detonation control parameters in a shorter time.

Description

Technical Field

[0001] The present invention relates to the technical field of air defense missile detonation precision control, and in particular to an infrared imaging and ranging integrated fuze information fusion method and system based on BAS improved particle filtering. Background Art

[0002] In modern warfare, the nation is placing increasing emphasis on air security, and air defense missile technology is rapidly developing. However, air defense missiles, which rely solely on formation detection and detonation control via seeker measurements, are no longer able to cope with the increasing speed and miniaturization of aerial threats and the complexities of the aerial battlefield.

[0003] Infrared imaging technology is widely used in air defense missiles because it not only operates at night but can also effectively identify the target's shape, which is crucial for precise detonation control. However, infrared imaging detectors can only obtain high-precision angle and angular velocity information, lacking target distance information and unable to determine the target's position in space. Air defense missiles using infrared imaging technology require a high-precision laser rangefinder that rotates coaxially with the infrared imaging detection device.

[0004] In the study of detonation control, more research has focused on using data obtained from information processing centers, while less research has been conducted on how to achieve the fusion of multi-sensor measurement data. Wen YF, Liu B. Unscented Kalman Filter Applied to Burst Delay Technology [J]. Journal of Detection & Control, 2009. This paper analyzes the hardware of the integrated fuze, achieves information synchronization, and uses the UKF algorithm to achieve the fusion of synchronized information. However, this method has high requirements for the hardware of the missile system and is unique. At the same time, it was found that the infrared imaging and ranging integrated fuze has the following characteristics: different start-up times, different sampling frequencies, and the sampling frequencies are not necessarily multiple. In order to solve this problem, the literature Hua WX, Liang JZ, Fei C, et al. Active-passive joint tracking algorithm with infrared sensors and laser [J]. infrared and laser engineering, 2001. adopted a sensor data fusion time alignment method based on the least squares method, and used EKF to filter the aligned data. Since the particle filter algorithm is not affected by the nonlinearity and non-Gaussian noise of the system function, it is widely used in artificial intelligence, computer vision, intelligent robots, detonation control and target tracking. However, the EKF method needs to process a large amount of data and consumes a lot of memory, which increases the time complexity of the algorithm. In addition, the traditional particle filter algorithm is prone to lose the effectiveness and diversity of samples, which will lead to the emergence of sample dilution problems. Summary of the Invention

[0005] In view of the problems existing in the above-mentioned prior art, the present invention aims to propose an infrared imaging and ranging integrated fuze information fusion method and system based on BAS improved particle filtering to solve the problems existing in the above-mentioned background technology.

[0006] The present invention addresses the problem that missile side fuzes have a short detection distance and are prone to occur after the detonation point. By switching from target tracking to local target feature point tracking during the missile-target intersection process, a missile-target intersection model is established. An integrated fuze combining an infrared imaging detector with a laser rangefinder is used to achieve forward detection, thereby extending the detection distance compared to side detection and providing sufficient time for the fuze. After obtaining detection data through the infrared imaging detector and the laser rangefinder, a linear interpolation method is used to achieve data alignment preprocessing of the two sensors. Subsequently, the present invention uses a single search algorithm with low computational complexity, namely a Beetle Antennae Search Algorithm (BAS), to optimize particles, thereby significantly reducing the number of particles used in a particle filter and lowering the time complexity of the algorithm. Data is filtered through an improved particle filter algorithm based on the BAS, thereby improving the accuracy of the detonation delay time.

[0007] In order to achieve the above technical objectives through the above technical ideas, the present invention adopts the following technical solutions:

[0008] According to one aspect of the present invention, a method for integrating infrared imaging and ranging fuze information fusion based on BAS-improved particle filtering is provided, comprising the following steps:

[0009] S1: Establish the missile-target intersection model in the O-XYZ missile body coordinate system;

[0010] S2: Measure the angle between the missile and the target through an infrared imaging detector, and measure the distance between the missile and the target through a laser rangefinder;

[0011] S3: The information fusion center integrates the angle information and the distance information to obtain the relative speed and relative distance between the missile and the target, thereby establishing a detonation delay time model;

[0012] S4: Use the particle filter algorithm improved based on BAS to filter the data after information fusion processing, so that it moves to a position with high fitness, and estimate the detonation delay time in real time to obtain high-precision detonation control parameters.

[0013] Furthermore, the specific process of establishing the missile-target intersection model in the above S1 is as follows:

[0014] Phase 1: Target Tracking

[0015] The missile M is at the origin and has a speed of V M In the same direction as OX, the target T moves at a speed of V T, and attack the missile M with an angle θ with the missile M. The two make uniform linear motion in the missile body coordinate system. At this stage, the target detonation delay time t measured by the infrared imaging detector from the initial detection position to the coordinate plane YOZ of the missile body coordinate system is obtained. go_1 ;

[0016] The second stage: from target tracking to local feature point tracking of the target

[0017] The local feature point of the target is G, and the speed is V T , the relative speed between the missile and the target is V R The projection point of point G on the O-YZ plane is represented by g, the angle between Og and the OZ axis is the detection angle β, the angle between OG and OX is the azimuth angle γ, and the intersection of the relative motion trajectory of the missile and the target and the plane O-YZ is represented by P, thereby establishing the missile-target rendezvous model.

[0018] Furthermore, at a relative speed V R The detonation delay time t of the local feature point G of the target moving to point P go_1 The calculation expression is as follows:

[0019]

[0020]

[0021] Among them, q is a parameter introduced to simplify formula (1), are expressed as the first-order derivatives of q and β, respectively.

[0022] Furthermore, it is characterized in that, in the above step S3, the process of information fusion processing is as follows:

[0023] The linear interpolation method is used to perform time alignment preprocessing on the angle information measured by the infrared imaging detector and the distance information measured by the laser rangefinder, including the following steps:

[0024] (I) Time alignment of the start time: According to the start time signal preset by the laser rangefinder, the infrared imaging detector starts a new periodic measurement to achieve the calibration of the start time;

[0025] (II) Time alignment of the infrared imaging detector and the laser rangefinder measurement data with disproportionate sampling periods:

[0026] ①Process the measurement data of the high sampling frequency sensor to obtain the motion curve of the measurement data;

[0027] ② Obtain the motion curve of the measurement data in the reference period through linear interpolation.

[0028] Furthermore, through information fusion processing, the relative speed and relative distance between the missile and the target are obtained as follows:

[0029] Relative distance between missile and target The expression is:

[0030]

[0031] Relative speed between missile and target The expression is:

[0032]

[0033] Where T is the sampling period of the laser rangefinder, i = 1, 2, ..., n, n is the number of sampling points obtained by the laser rangefinder just before the seeker loses control, R Xi ,R Yi ,R Zi and V RXi ,V RYi ,V RZi is the relative distance and velocity component between the missile and the target on the three coordinate axes of the missile body coordinate system at point i;

[0034] The target detonation delay time t from the position where the seeker loses control to the coordinate plane YOZ of the missile body coordinate system is obtained by measuring the infrared imaging detector and the laser rangefinder when the infrared imaging detector and the laser rangefinder are working at the moment before the seeker loses control. go_2 The expression is:

[0035]

[0036] The expression of the detonation delay time model is obtained as follows:

[0037]

[0038] Where τ is the detonation delay time, ρ is the miss distance, V0 is the scattering velocity of warhead fragments, and Δτ is the detonation delay time error adjusted according to the intersection information.

[0039] Furthermore, in the above step S4, the sampling process in the particle filter algorithm is improved based on BAS, including the following steps:

[0040] (I) Establish the fitness function of the i-th particle

[0041] Assumptions is the center of mass position of the i-th particle at time t, corresponding to the position The fitness function is Defined as:

[0042]

[0043] Among them, z t is the latest observation data, R is the measurement noise variance, is the predicted value of the sampled particle;

[0044] (II) Optimize particle distribution and move sample particles to positions with high fitness;

[0045] (III) when the optimal value of the sample particles reaches a preset threshold, the optimization stops;

[0046] Furthermore, the steps of the particle filter algorithm improved based on BAS are as follows:

[0047] (I) Parameter initialization: Select sampling points {X i ,i=1,2,…,N}, where the state probability, state noise and observation noise are always uncorrelated and are expressed as

[0048] The parameter initialization results are:

[0049]

[0050] (II) Importance Sampling:

[0051] According to formula (13), we can get the particle and sample particle Weight Finally, we get the weighted particle swarm Get the initial state estimate and the actual observed value z t ;

[0052] (III) BAS Search

[0053] ①Predict the state of the sampled particles;

[0054] ② Initialize the initial value of the beetle whisker search algorithm, calculate the fitness values ​​of the left and right beetle whiskers according to formula (12), select the beetle whisker direction with high fitness, and move forward with a step size s;

[0055] ③ Within the search range, if the difference between the current position of the longicorn and the position with the highest fitness reaches the set threshold, the search is stopped; otherwise, the search is stopped after the set number of iterations is reached;

[0056] ④ Perform this operation on all N sampled particles and normalize the weights of the N sampled particles

[0057] (IV) Status Update

[0058] After resampling the sampled particles, the filtered state estimate is output Determine whether the state filtering value reaches the set threshold; if so, stop filtering; otherwise, return to step (II).

[0059] According to another aspect of the present disclosure, there is provided an infrared imaging and ranging integrated fuze information fusion system based on BAS improved particle filtering, comprising:

[0060] Modeling system, used for model building;

[0061] Infrared imaging detection system, used to obtain angle information between the missile and the target;

[0062] The laser ranging system is used to obtain the distance information between the missile and the target; the information fusion system is used to fuse the data of the infrared imaging detection system and the laser ranging system.

[0063] According to another aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores computer instructions, and the processor is configured to execute the computer instructions stored in the memory to implement any one of the methods described above.

[0064] According to another aspect of the present disclosure, a computer-readable storage medium storing computer instructions is provided, where the computer instructions are used to enable a computer to execute any one of the methods described above.

[0065] By adopting the above technical solution, the present invention has the following beneficial technical effects compared with the prior art:

[0066] 1. Starting from the perspective that the detection distance of missile side fuzes is short and they are easy to appear after the explosion point, the present invention establishes a research object of the missile and target intersection model based on the idea of ​​forward detection. The distance and angle information of the relative motion between the missile and the target are obtained by combining an infrared imaging detector with a laser rangefinder. The data is further pre-processed by time alignment through an information fusion center. Then, the pre-processed data samples are optimized based on the particle filter algorithm. The BAS is used to optimize the spatial distribution state of the particles in the particle filter algorithm, so that they are concentrated at a position close to the actual state, thereby improving the filter estimation accuracy and obtaining more accurate detonation control parameters.

[0067] 2. The method of the present invention can obtain more accurate detonation delay time and angle information, thereby improving the detonation control accuracy and the efficiency of detonation-war coordination; compared with the existing algorithms commonly used in the field of detonation control, the particle filter algorithm improved based on BAS of the present invention improves the accuracy of detonation delay time by 61.82%, the accuracy of azimuth angle by 66.28%, and the accuracy of detection angle by 36.15%, and reduces the time complexity of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments.

[0069] Figure 1 A comparison between forward detection and lateral detection of the integrated fuze of the present invention;

[0070] Figure 2 This is a model diagram of the projectile-target intersection of the present invention;

[0071] Figure 3 It is a flow chart of information fusion of the present invention;

[0072] Figure 4 is a graph showing an error in the detonation delay time of the detonation control parameters during the filtering process of the present invention;

[0073] Figure 5 is a graph showing an azimuth angle γ error of the detonation control parameter during the filtering process of the present invention;

[0074] Figure 6 is a curve diagram of the detection angle β error of the detonation control parameter in the filtering process of the present invention;

[0075] Figure 7 The estimation accuracy of the detonation delay time for different particle numbers based on the PF and BAS-PF algorithms of the present invention;

[0076] Figure 8 The estimation accuracy of the azimuth angle γ for different particle numbers based on the PF and BAS-PF algorithms of the present invention;

[0077] Figure 9 It is the estimation accuracy of the azimuth angle β based on different particle numbers under the PF and BAS-PF algorithms of the present invention. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the specific implementation methods of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0079] Reference Figure 1-9 The present invention provides an infrared imaging and ranging integrated fuze information fusion method and system based on BAS improved particle filtering, comprising the following steps:

[0080] S1: Establish the missile-target intersection model in the O-XYZ missile body coordinate system;

[0081] Before establishing the missile-target intersection model, the present invention adopts an integrated fuze to realize forward detection to solve the problem that the missile side fuze has a short detection distance and is easy to appear after the explosion point. Compared with the side detection, the forward detection has a longer detection distance and can provide sufficient time for the fuze. The comparison between forward detection and side detection is shown in the figure. Figure 1 shown.

[0082] Phase 1: Target tracking. When the number of pixels in the target image and the number of pixels in the infrared imaging detector reach a set ratio (the ratio is set according to the specific infrared imaging system), the target image will fill the field of view of the infrared imaging detector, and the seeker will turn to track the local feature points of the target.

[0083] The missile M is at the origin and has a speed of V M In the same direction as OX, the target T moves at a speed of V T , and attack the missile M with an angle θ with the missile M. The two make uniform linear motion in the missile body coordinate system. At this stage, the target detonation delay time t measured by the infrared imaging detector from the initial detection position to the coordinate plane YOZ of the missile body coordinate system is obtained. go_1 , this time is used to determine the power-on time of the laser rangefinder;

[0084] The second stage: from target tracking to local feature point tracking of the target

[0085] like Figure 2 As shown, the local feature point of the target is G and the speed is V T , the relative speed between the missile and the target is V R , the projection point of point G on the O-YZ plane is represented by g, the angle between Og and the OZ axis is the detection angle β, the angle between OG and OX is the azimuth angle γ, the intersection point of the relative motion trajectory between the missile and the target and the plane O-YZ is represented by P, and the following missile-target rendezvous model is established;

[0086] At relative speed V R The detonation delay time t of the local feature point G of the target moving to point P go_1 The calculation expression is as follows:

[0087]

[0088]

[0089] Among them, q is a parameter introduced to simplify formula (1), are expressed as the first-order derivatives of q and β, respectively.

[0090] S2: Measure the angle between the missile and the target through infrared imaging, and measure the distance between the missile and the target through a laser rangefinder;

[0091] The infrared imaging detector is based on the detonation delay time t go_1 Controls the start time of the laser rangefinder, which measures the distance between the missile and the target at the target tracking point.

[0092] S3: The information fusion center integrates the angle information and the distance information to obtain the relative speed and relative distance between the missile and the target, thereby establishing a detonation delay time model;

[0093] In the target tracking process of multiple sensors, the present invention fuses the information of multiple sensors through the information fusion center, utilizes the richness, fault tolerance and complementarity of the measurement information of multiple sensors, improves data redundancy, and thus improves the accuracy of the detonation control parameters;

[0094] The process of information fusion processing is as follows Figure 3 The specific process is as follows:

[0095] In traditional multi-sensor target tracking, the data obtained by multiple sensors need to be "spatially aligned" and "temporally aligned". In this invention, since the measurement coordinate system of the infrared imaging detector and the laser rangefinder is the projectile coordinate system, there is no need to spatially align the data of the two sensors. Only time alignment preprocessing is required for the data.

[0096] In the present invention, a linear interpolation method is used to perform time alignment preprocessing on the angle information measured by the infrared imaging detector and the distance information measured by the laser rangefinder, including the following steps:

[0097] (I) Time alignment of the start time: According to the start time signal preset by the laser rangefinder, the infrared imaging detector starts a new periodic measurement to achieve the calibration of the start time;

[0098] (II) Time alignment of the infrared imaging detector and the laser rangefinder measurement data with disproportionate sampling periods:

[0099] ①Process the measurement data of the high sampling frequency sensor to obtain the motion curve of the measurement data;

[0100] ② Obtain the motion curve of the measurement data in the reference period through linear interpolation.

[0101] Furthermore, through information fusion processing, the relative speed and relative distance between the missile and the target are obtained as follows:

[0102] Relative distance between missile and target The expression is:

[0103]

[0104] Relative speed between missile and target The expression is:

[0105]

[0106] Where T is the sampling period of the laser rangefinder, i = 1, 2, …, n, n is the number of sampling points obtained by the laser rangefinder just before the seeker loses control, and are the components of the relative distance and velocity between the missile and the target on the three coordinate axes of the missile body coordinate system at point i.

[0107] From the above, we can obtain the target detonation delay time t from the position where the seeker loses control to the coordinate plane YOZ of the missile body coordinate system, which is measured by the infrared imaging detector and the laser rangefinder when they are working at the moment before the seeker loses control. go_2 , used to adjust the fuze detonation delay time, the detonation delay time t go_2 The expression is:

[0108]

[0109] According to formulas (3)-(5), the expression of the detonation delay time model is obtained as follows:

[0110]

[0111] Where τ is the detonation delay time, ρ is the miss distance, V0 is the scattering velocity of warhead fragments, and Δτ is the detonation delay time error adjusted according to the intersection information.

[0112] S4: Use the particle filter algorithm improved based on BAS to filter the data after information fusion processing, so that it moves to a position with high fitness, and estimate the detonation delay time in real time to obtain high-precision detonation control parameters.

[0113] Considering that the relative speed between the missile and the target changes very little at the end of the missile-target intersection, it can be assumed that the target tracking aiming point moves in a straight line at a constant speed relative to the air defense missile. Therefore, the constant speed straight line model can be used to describe the relative motion state of the target aiming point and the air defense missile at the end of the trajectory. The state vector of the target is selected as The state equation is as follows:

[0114] X(k+1)=ΦX(k)+ΓW(k) (7)

[0115] After time alignment, the data obtained by the two sensors are expanded to obtain the observation equation of the system;

[0116]

[0117] in, is the transfer matrix, is the noise distribution matrix, W(k) has a mean of 0 and a covariance of q w I 6*6 Gaussian white noise, V(k) has a mean of 0 and a covariance of q v I 3*3 is Gaussian white noise, T is the reference period, R(k), γ(k) and β(k) are the aligned distance information and angle information respectively.

[0118] When dealing with nonlinear system problems, traditional nonlinear filtering algorithms convert nonlinear systems into linear systems or directly process nonlinear systems, resulting in a relatively complex data processing process. Particle filtering, on the other hand, is a filtering method based on random simulation and combines the Monte Carlo principle with Bayesian filtering. This method is unaffected by the nonlinearity of system functions and non-Gaussian noise, and can be applied to real-time state estimation of detonation delay time.

[0119] However, the traditional particle filter algorithm is prone to lose the effectiveness and diversity of samples during the filtering process, resulting in dilution of sample particles, and increasing the number of particles will increase the time complexity of the algorithm, which is not desirable in high real-time fields (such as the field of initiation control). Therefore, the present invention adopts a single search algorithm with low computational complexity, namely the beetle whisker search algorithm to improve the sampling process in the particle filter algorithm, so that the sampling points falling in the prior distribution area are gradually transferred to the maximum likelihood area, thereby obtaining a particle filter algorithm improved based on the beetle whisker search algorithm.

[0120] The Beetle Whiskers Search (BAS) algorithm, proposed in 2017, is an intelligent swarm optimization algorithm inspired by the foraging of beetles in nature. The principle of this algorithm is that when beetles in nature are looking for food, they need to use the intensity of the food odor detected on their left and right sides to determine the next direction to continue searching for food, and continue to move forward according to the intensity of the odor until they find the point with the strongest odor in the entire environment.

[0121] Assumptions is the center of mass position of the i-th particle at time t, corresponding to the position The fitness function is Its maximum value corresponds to the location of the odor source; according to the above principle, the longhorn beetle whisker search model is expressed as follows:

[0122] (I) Define the direction of the beetle's whiskers:

[0123]

[0124] Here, rand(·) is a random function and k represents the spatial dimension. The coordinates of the longicorn beetle's whiskers are then obtained.

[0125]

[0126] in, represents the coordinates of the right beard of the longicorn. represents the coordinates of the left beard of the longhorn beetle. The distance between the left and right beards is d rl .

[0127] (II) The search behavior of the longicorn can be expressed as:

[0128]

[0129] Among them, δ t represents the step size factor, the initial value is the same as the search area, and sign(·) represents the sign function.

[0130] The steps of the BAS algorithm are as follows:

[0131] (1) Initialize the step size factor, iteration termination condition, initial position information, number of iterations, etc.

[0132] (2) Calculate the positions of the left and right whiskers of the longicorn according to the formula;

[0133] (3) Calculate the fitness function value of the longhorn beetle at the current position and store it, update the step size according to the formula and update the position of the longhorn beetle;

[0134] (4) Determine whether the set iteration termination condition has been reached or whether the set maximum number of iterations has been exceeded. If the termination condition has been reached, the global optimal solution will be output; otherwise, jump to step (2).

[0135] Based on the above-mentioned longicorn beetle whisker search algorithm, the present invention optimizes the particle distribution according to the BAS algorithm, transfers the sample particles in the particle filter to the high likelihood area, and moves the particles to the position with high fitness. When the optimal value of the sample particles reaches a preset threshold, the optimization stops; through the above-mentioned optimization, the particles can be concentrated in a position close to the real state, thereby solving the problem of sample particle dilution.

[0136] In the particle filter algorithm improved based on BAS, the fitness function of the longhorn beetle at its location is Defined as:

[0137]

[0138] Among them, z t is the latest observation data, R is the measurement noise variance, is the predicted value of the sampled particle; the fitness function value of N particles is obtained according to the position of each sample particle and formula (12)

[0139] The specific steps of the particle filter algorithm improved based on BAS are as follows:

[0140] (1) Parameter initialization: Select sampling points {X i , i=1,2,…,N}, where the state probability, state noise and observation noise are always uncorrelated and are expressed as E[W(k)V T (j)] = 0;

[0141] The parameter initialization results are:

[0142]

[0143] (2) Importance sampling:

[0144] According to formula (13), we can get the particle and sample particle Weight Finally, we get the weighted particle swarm Get the initial state estimate and the actual observed value z t ;

[0145] (3) Longicorn beetle search

[0146] ①Predict the state of the sampled particles;

[0147] ② Initialize the initial value of the beetle whisker search algorithm, calculate the fitness values ​​of the left and right beetle whiskers according to formula (12), select the beetle whisker direction with high fitness, and move forward with a step size s;

[0148] ③ Within the search range, if the difference between the current position of the longicorn and the position with the highest fitness reaches the set threshold, the search is stopped; otherwise, the search is stopped after the set number of iterations is reached;

[0149] ④ Perform this operation on all N sampled particles and normalize the weights of the N sampled particles

[0150] (4) Status Update

[0151] After resampling the sampled particles, the filtered state estimate is output Determine whether the state filter value reaches the set threshold; if so, stop filtering; otherwise, return to step (2).

[0152] For the infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering proposed in the present invention, in the MATLAB simulation environment, under typical simulation conditions, the filtering effects of the method of the present invention (BAS-PF) are compared with those of the particle filter algorithm (PF) and the EKF algorithm commonly used in the detonation control algorithm.

[0153] The simulation conditions are as follows: the ranging error of the laser ranging system is Gaussian white noise with an average of 0 and a standard deviation of 1m; the angle measurement error of the infrared imaging detection system is Gaussian white noise with an average of 0 and a standard deviation of 1rad; the state noise is 1m; the simulation step size is 100; the number of particles in the PF and BAS-PF algorithms is 100; the relative distance between the projectile and the target is 300m; and the relative speed between the projectile and the target is 1500m / s.

[0154] Table 1. Comparison of three filtering algorithms

[0155]

[0156] Through simulation experiments, experimental comparisons of the detonation delay time error curve, azimuth error curve and detection angle error curve of the detonation control parameters under the three filtering algorithms are obtained. Figure 4-6 As shown in Table 1 above, the experimental data shows the root mean square error (RMSE) of the detonation control parameters under the three algorithms, which can intuitively reflect the filtering accuracy of different algorithms.

[0157] from Figure 4-6 It can be seen from the above that the final results of the three algorithms for particle filtering are all stable. Figure 4 It can be seen that the errors of the EKF algorithm and the BAS-PF algorithm are small. Compared with the PF and BAS-PF algorithms, the error fluctuation of the EKF is larger.

[0158] exist Figure 5 and Figure 6 In the simulation, the errors of EKF and PF fluctuate greatly, while those of BAS-PF fluctuate less and converge faster. When the number of simulation steps is about 50, the errors of the three algorithms vary to varying degrees, but then the errors of the algorithms tend to be stable. This is because the inverse trigonometric function is sensitive to angles. When the angle approaches ±90°, a sudden change occurs, causing the error to increase sharply. However, the algorithm can still estimate the detonation control parameters well.

[0159] From the experiment, it is found that when the number of particles is 100, the simulation times of the three algorithms based on EKF, PF and BAS-PF are 2.2459s, 0.8859s and 0.5386s respectively. It can be seen that the improved particle filter algorithm (BAS-PF) of the present invention effectively saves the time required for particle filtering; since reducing the number of particles in the algorithm can reduce the time complexity of the algorithm, the present invention further uses the PF and BAS-PF algorithms as the basis, and simulates the simulation time accuracy by selecting different particle numbers, such as Figure 7-9 As shown in the figure, N50 and N100 respectively indicate that the number of particles used by the algorithm is 50 and 100. Figure 6-8 In the figure, the number of particles of the PF algorithm is 100, and the number of particles of the BAS-PF algorithm is 50 and 100 respectively. It can be found that when the number of particles of the PF algorithm is increased, the accuracy of the PF algorithm is improved. Moreover, when the number of particles of the BAS-PF algorithm of the present invention is 50, the accuracy of the estimated detonation control parameters of the PF algorithm when the number of particles is 100 can be obtained. This shows that the algorithm of the present invention effectively reduces the complexity of the algorithm time.

[0160] From the RMSE values ​​of various parameters in Table 1, it can be found that the improved BAS-PF algorithm can achieve higher detonation control time and angle accuracy.

[0161] By comparing and analyzing several common nonlinear filtering algorithms used in integrated fuze information fusion methods, this paper proposes an integrated fuze information fusion method based on a BAS-modified particle filtering algorithm to obtain more accurate detonation delay time and angle information. This method improves detonation control precision and achieves higher-precision fuze-warfare coordination efficiency. Compared with existing algorithms, this method improves the accuracy of detonation delay time by 61.82%, the accuracy of azimuth angle by 66.28%, and the accuracy of detection angle by 36.15%, while reducing the time required by 1.7073 seconds.

[0162] Therefore, the improved particle filter algorithm based on BAS proposed in the present invention can obtain more accurate detonation control parameters in a shorter time, verifying the feasibility and reliability of the proposed method.

[0163] The above are only preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. For those skilled in the art, the technical solutions and concepts of this application are susceptible to various equivalent substitutions and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this application shall be included within the scope of protection of the present invention.

Claims

1. The infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering is characterized by: The following steps are involved: S1: Establish the missile-target intersection model in the O-XYZ missile body coordinate system; S2: Measure the angle between the missile and the target through an infrared imaging detector, and measure the distance between the missile and the target through a laser rangefinder; S3: The information fusion center integrates the angle information and the distance information to obtain the relative speed and relative distance between the missile and the target, thereby establishing a detonation delay time model; S4: Use the particle filter algorithm improved by BAS to filter the data after information fusion processing, so that it moves to the position with high fitness, and estimate the detonation delay time in real time to obtain high-precision detonation control parameters; In the above step S4, the sampling process in the particle filter algorithm is improved based on BAS, including the following steps: (I) Establish the fitness function of the i-th particle : Assumptions is the center of mass position of the i-th particle at time t, corresponding to the position The fitness function is Defined as: （12） in, The latest observation data, is the measurement noise variance, is the predicted value of the sampled particle; (II) Optimize particle distribution and move sample particles to positions with high fitness; (III) When the optimal value of the sample particle reaches a preset threshold, the optimization stops; The steps of the particle filter algorithm improved based on BAS are as follows: (I) Parameter initialization: from the state prior distribution Select sampling points , where the state probability, state noise, and observation noise are always uncorrelated, in all Next, it is expressed as , the parameter initialization result is: （13） (II) Importance sampling: According to formula (13), we get particles and sample particles Weight , and finally get the weighted particle swarm , get the initial state estimate and the actual observed value ; (III) BAS search: ①Predict the state of the sampled particles; ② Initialize the initial value of the beetle whisker search algorithm, calculate the fitness values ​​of the left and right beetle whiskers according to formula (12), select the beetle whisker direction with high fitness, and move forward by step size s; ③ Within the search range, if the difference between the current position of the longicorn and the position with the highest fitness reaches the set threshold, the search is stopped; otherwise, the search is stopped after the set number of iterations is reached; ④ Perform this operation on all N sampled particles and normalize the weights of the N sampled particles ; (IV) State update: After resampling the sampled particles, the filtered state estimate is output , determine whether the state filtering value reaches the set threshold; if it reaches, stop filtering, otherwise, return to step (II).

2. The infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering according to claim 1 is characterized in that: The specific process of establishing the projectile-target intersection model in step S1 is as follows: Phase 1: target tracking; Missile M is at the origin and has a speed In line with OX direction, target T is at speed , and the angle between it and the missile M is Attack missile M, and both make uniform linear motion in the missile body coordinate system. At this stage, the target detonation delay time from the initial detection position to the coordinate plane YOZ of the missile body coordinate system measured by the infrared imaging detector is obtained. ; The second stage: switching from target tracking to local feature point tracking of the target; The local feature point of the target is G, and the speed is , the relative speed between the missile and the target is , the projection point of point G on the O-YZ plane is represented as g, and The angle between the axes is the detection angle , and The angle between them is the azimuth The intersection of the relative motion trajectory between the missile and the target and the plane O-YZ is represented as P, and the missile-target intersection model is established.

3. The infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering according to claim 2 is characterized in that: At relative speed Detonation delay time of the local feature point G of the target moving to point P The calculation expression is as follows: （1） （2） in, is a parameter introduced to simplify formula (1), Respectively expressed as and The first derivative of .

4. The infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering according to claim 1 is characterized in that: In the above step S3, the process of information fusion processing is as follows: The linear interpolation method is used to perform time alignment preprocessing on the angle information measured by the infrared imaging detector and the distance information measured by the laser rangefinder, including the following steps: (I) Time alignment of the start time: According to the start time signal preset by the laser rangefinder, the infrared imaging detector starts a new periodic measurement to achieve the calibration of the start time; (II) Temporal alignment of infrared imaging detector and laser rangefinder measurement data with disproportionate sampling periods: ①Process the measurement data of the high sampling frequency sensor to obtain the motion curve of the measurement data; ② Obtain the motion curve of the measurement data in the reference period through linear interpolation.

5. The infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering according to claim 4 is characterized in that: Through information fusion processing, the relative speed and relative distance between the missile and the target are obtained as follows: Relative distance between missile and target The expression is: （3） Relative speed between missile and target The expression is: （4） Where T is the sampling period of the laser rangefinder, i=1,2,…,n, n is the number of sampling points obtained by the laser rangefinder just before the seeker loses control, and is the relative distance and velocity component between the missile and the target on the three coordinate axes of the missile body coordinate system at point i; The target detonation delay time from the position where the seeker loses control to the coordinate plane YOZ of the missile body coordinate system is obtained by measuring the infrared imaging detector and the laser rangefinder when the infrared imaging detector and the laser rangefinder are working at the moment before the seeker loses control. The expression is: （5） The expression of the detonation delay time model is obtained as follows: （6） in, is the detonation delay time, is the off-target amount, is the scattering speed of warhead fragments, It is the detonation delay time error adjusted according to the intersection information.

6. The infrared imaging and ranging integrated fuze information fusion system based on BAS improved particle filtering is characterized by: The system is used to implement the infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering as described in any one of claims 1 to 5, and the system includes: Modeling system, used for model building; Infrared imaging detection system, used to obtain angle information between the missile and the target; Laser ranging system, used to obtain the distance information between the missile and the target; The information fusion system is used to fuse the data of the infrared imaging detection system and the laser ranging system.

7. An electronic device comprising a memory and a processor, characterized in that: Computer instructions are stored in the memory, and the processor is used to run the computer instructions stored in the memory to implement the steps of the infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute the steps of the infrared imaging and ranging integrated fuze information fusion method based on BAS improved particle filtering as described in any one of claims 1 to 5.

Images