CN111307143B - Bionic navigation algorithm for multi-target evolution search based on geomagnetic gradient assistance - Google Patents

Abstract

Firstly, acquiring geomagnetic parameter information of a position and a destination of a carrier at the current moment; and at the initial navigation moment, the carrier walks along the east direction and the north direction respectively to obtain geomagnetic parameter gradient information, and then the course angle is predicted according to the principle that geomagnetic parameters converge at the same place at the same time. In order to reduce the invalid searching process, in the evolutionary algorithm, the population sample space is restricted according to the predicted course angle, and the searching efficiency is improved. And secondly, the method is inspired by a parallel approach method in missile tracking, the evaluation criterion of the sample is improved, the sample is evaluated more accurately, and the navigation search path is optimized. According to the method, multiple parameters of the end point geomagnetic field are used as target values, efficient and rapid path search is carried out under the condition that no prior geomagnetic map exists, and geomagnetic autonomous navigation of the autonomous underwater vehicle during long-term navigation is achieved.

Description

Bionic navigation algorithm for multi-target evolution search based on geomagnetic gradient assistance

Technical Field

The invention relates to a bionic navigation algorithm based on geomagnetic gradient assisted multi-target evolutionary search, which is suitable for realizing geomagnetic autonomous navigation of an autonomous underwater vehicle during long-term navigation by taking multiple parameters of a terminal geomagnetic field as target values under the condition of no prior geomagnetic field.

Background

The autonomous underwater vehicle integrates advanced technologies such as underwater acoustic communication, intelligent control, energy storage, multi-sensor measurement and information fusion, has the advantages of good autonomy, strong flexibility, small size, light weight, wide range of motion, good concealment and the like, can be applied to petroleum resource investigation, submarine pipeline investigation, marine environment investigation, underwater equipment maintenance and the like, and is one of important tools for future marine detection and development. Unlike navigation systems for terrestrial or airborne vehicles, the rapid attenuation of radio waves in an underwater environment makes radio navigation systems, represented by the global navigation satellite system, no longer suitable for autonomous underwater vehicles. Currently, the commonly used underwater navigation technologies mainly include inertial navigation, underwater acoustic positioning navigation and geophysical navigation. The inertial navigation system has the characteristics of autonomy, continuity, concealment and the like, is often used as a main navigation system of the AUV, but the error of the inertial navigation system is accumulated along with time, and is not suitable for long-endurance long-distance navigation. The acoustic navigation can be divided into 3 types of ultra-short baselines, short baselines and long baselines, wherein the underwater acoustic array is complicated in arrangement and recovery work, the action ranges and the precision of the ultra-short baselines and the short baselines are limited, and the active navigation is adopted, so that the concealment is poor. The geophysical navigation system is a technology for navigating by utilizing the physical characteristics of the earth, mainly comprises 3 types of terrain matching, geomagnetic matching and gravity matching, has the characteristics of strong autonomy, good concealment, no limitation of regions and time and the like, but the navigation precision mainly depends on the precision of a prior map, and the acquisition of the prior map becomes a restriction condition of geophysical field navigation.

Recent studies have shown that many living beings on the earth can be located and navigated based on the information of the earth's magnetic field. For example, a pigeon flying in a strange place can be homed after flying for hundreds of kilometers; the Pacific salmon can navigate from open ocean to correct coastal areas by using geomagnetic clues when spawning and migrating; adult green turtles may also help them return to the egg laying site, etc. by detecting geomagnetic information. The relevant experimental verification and research analysis of the behavior of various animals by using geomagnetic navigation show that: the geomagnetic field is a reliable navigation information source for long-distance movement of animals. Clearly, storing the complete geomagnetic map in their brains is less likely, which provides a biological basis for navigation without relying on a priori geomagnetic maps. Meanwhile, the unique correspondence between the earth magnetic field vector and each point in the near-earth space provides a sufficient theoretical basis for geomagnetic navigation. Therefore, under the condition of no prior geomagnetic field, the multi-parameter of the end-point geomagnetic field is taken as a target value, and the solution of multiple targets is combined with navigation motion to construct the bionic geomagnetic navigation method. From the perspective of biomagnetic trend sensitivity, researchers provide a geomagnetic bionic navigation method which does not depend on evolutionary search of prior geomagnetic data, but the existing method has the problems of long navigation time consumption, low efficiency and strong path.

Disclosure of Invention

To solve the above existing problems. The invention provides a bionic navigation algorithm for multi-target evolutionary search based on geomagnetic gradient assistance. Under the condition of no prior geomagnetic map, aiming at the problems of strong randomness, low search efficiency and the like in an evolution search strategy, a bionic navigation algorithm of multi-target evolution search based on geomagnetic gradient assistance is provided, so that the search efficiency is improved, and a navigation path is more reliable and accurate. To achieve this object:

the invention provides a bionic navigation algorithm for multi-target evolution search based on geomagnetic gradient assistance, which comprises the following specific steps of:

step 1: acquiring geomagnetic parameters of a position where a carrier is located at the current moment and a target position;

step 2: calculating gradient information of geomagnetic parameters, predicting a course angle according to a simultaneous and same-ground convergence principle of multiple geomagnetic parameters, and further constraining a sample space of an evolutionary algorithm;

and step 3: judging whether the destination is reached: constructing an objective function according to the current position and geomagnetic parameters of the destination, judging whether the carrier reaches the destination or not by observing the current objective function, and stopping searching to finish navigation if the carrier reaches the destination; otherwise, jumping to the step 4;

and 4, step 4: selecting the motion direction of the next step according to an evolutionary algorithm, performing posterior evaluation on the executed sample, and updating the sample space in real time; and repeating the steps until the navigation is completed.

As a further improvement of the invention, the geomagnetic parameter elements comprise three components of the magnetic field, namely a north component, an east component and a vertical component, the total intensity of the magnetic field, a horizontal component of the magnetic field, a declination and a dip angle, which are all or part of the three components.

As a further improvement of the present invention, step 4 is to guide the carrier to the destination as soon as possible, and multiple parameters should be kept to converge simultaneously and simultaneously during the path search process as much as possible, that is, the following conditions are satisfied:

wherein, B_i,k,B_j,kTwo different geomagnetic parameters at time k, B_i,TB_j,TTwo geomagnetic parameters at the target position, in the above formula, the geomagnetic parameter B at the time of k +1_i,k+1B_j,k+1The geomagnetic parameter and the geomagnetic gradient information at the previous time can be expressed as:

the course angle gamma is obtained by simultaneously solving the two formulas_kComprises the following steps:

and carrying out sample space constraint on the evolutionary search algorithm based on the predicted course angle.

As a further improvement of the invention, the magnitude and unit among geomagnetic parameters are considered, and after the loss function constructed by the current position of the carrier and the target position is normalized, the method comprises the following steps:

wherein, B_i,0、B_i,T、B_i,kThe i-th geomagnetic parameter information respectively including the initial position, the destination, and the position of the carrier at the time k, when the carrier moves to the destination, the loss function value is theoretically 0, and therefore F (B) is satisfied when the loss function value of the current position magnetic parameter is small_k) ≦ ε, the navigator may be considered to reach the destination, where ε is a minimal amount near 0, set according to navigation accuracy.

As a further improvement of the present invention, step 3 is a search as close to the shortest path as possible, a sample evaluation function is improved, and in order to converge each parameter of the geomagnetism as soon as possible, an input heading angle is required to satisfy:

(B_aim-B_k)//(B_k+1-B_k)

thus, it is possible to align the vector (B)_aim-B_k) And (B)_k+1-B_k) Angle therebetween

Observations were made to evaluate the quality of the samples:

as further improvement of the invention, a sample evaluation function is improved, and a corresponding population updating rule is formulated based on the improved evaluation criterion as follows:

wherein, P_pAnd (4) for the reproduction proportion of the sample, when the sample does not meet the reproduction condition, replacing the current sample with the randomly generated sample to realize the real-time update of the population.

Compared with the prior art, the invention has the advantages that: according to the geomagnetic bionic navigation algorithm provided by the application, the optimal advancing direction is continuously adjusted according to a search strategy without relying on a priori geomagnetic map, and navigation path search is completed quickly and efficiently. Compared with the existing time sequence evolution search method, the randomness in the advancing process is greatly reduced, the heading angle predicted by geomagnetic gradient information is used for restricting the sample space, and a plurality of invalid searches are avoided, so that the navigation path is more stable, and the realization of engineering application is facilitated; and secondly, modifying an evaluation criterion in the evolutionary algorithm, and evaluating the sample more accurately, so that the navigation path search is close to the shortest path.

Drawings

FIG. 1 is a flow chart of a navigation method of the present application;

FIG. 2 is a schematic diagram of the evaluation strategy of the present application;

FIG. 3 is a schematic comparison diagram of a navigation path without geomagnetic anomaly in an embodiment of the present application;

fig. 4 is a schematic diagram illustrating convergence of geomagnetic parameters of a path before improvement without geomagnetic anomaly according to an embodiment of the present application;

fig. 5 is a schematic diagram illustrating convergence of geomagnetic parameters of a path without geomagnetic anomaly improvement according to an embodiment of the present application;

FIG. 6 is a schematic comparison diagram of a navigation path with geomagnetic anomaly according to an embodiment of the present application;

fig. 7 is a schematic diagram illustrating convergence of geomagnetic parameters of a path before improvement of geomagnetic anomaly according to an embodiment of the present application;

fig. 8 is a schematic diagram illustrating convergence of geomagnetic parameters of a path with improved geomagnetic anomaly according to an embodiment of the present application.

Detailed Description

The invention is described in further detail below with reference to the following detailed description and accompanying drawings:

the invention provides a bionic navigation algorithm for multi-target evolutionary search based on geomagnetic gradient assistance. Under the condition of no prior geomagnetic map, aiming at the problems of strong randomness, low search efficiency and the like in an evolution search strategy, a bionic navigation algorithm of multi-target evolution search based on geomagnetic gradient assistance is provided, so that the search efficiency is improved, and a navigation path is more reliable and accurate.

The geomagnetic field includes a plurality of geomagnetic characteristic quantities, and from a bionic angle, the biological motion behavior has a characteristic of being sensitive to a geomagnetic field change trend, so that a process of geomagnetic bionic navigation can be regarded as a process of search convergence of a plurality of characteristic parameters of the geomagnetic field from an initial position to each characteristic parameter of a target position, and the implementation method shown in fig. 1 specifically includes the following steps:

1) carrier motion model establishment

In the bionic navigation process of the underwater vehicle based on geomagnetic parameters, the underwater vehicle can be regarded as a particle, and the motion equation can be expressed as follows:

wherein (x)_k,y_k) The position of the underwater vehicle at the moment k is shown, and u is the input of the system and is related to the heading angle theta and the speed v of the underwater vehicle. Assuming that the vehicle is moving at a constant speed within Δ T, V can be represented by a constant V, and the above equation reduces to:

where L denotes a motion step, and L ═ Δ T × V.

2) Course angle prediction

Accurate geomagnetic gradient information is difficult to obtain, and the gradient information of geomagnetic parameters is approximated by simple decomposition of geomagnetic parameter changes between two continuous sampling points of east and north. At the beginning of navigation, setting the heading angles of the carrier in the first step and the second step as 0 degree and 90 degrees respectively, measuring and recording the geomagnetic parameters of each position by using a magnetic sensor, and calculating geomagnetic gradient information by the following formula:

wherein, B_i,k,B_j,kTwo different geomagnetic parameters at the time k, L is the motion step length of the carrier,

respectively, the geomagnetic parameters at the k time_i,kAnd B_j,kGeomagnetic gradients in the x-axis and y-axis directions.

Secondly, in order to optimize the navigation path and improve the search efficiency, the geomagnetic multiparameter should satisfy the principle of simultaneous and local convergence as much as possible in the navigation path search process, namely:

wherein, B_i,k,B_j,k，B_i,T，B_j,TTwo different geomagnetic parameters of the time k and the target position respectively. In the above formula, the geomagnetic parameter B at the time k +1_i,k+1B_j,k+1The geomagnetic parameter and the geomagnetic gradient information at the previous time can be expressed as:

and solving the two formulas simultaneously to obtain a course angle as follows:

because the calculation of the geomagnetic gradient information has errors, the error of the predicted heading angle of the geomagnetic gradient information cannot be avoided, and therefore the navigation precision is reduced. And the geomagnetic gradient information prediction method has poor anti-interference performance and cannot overcome the interference caused by geomagnetic abnormal areas, so that the method does not completely depend on the predicted course angle to advance, but carries out the constraint of a sample space in an evolutionary search strategy based on the predicted course angle.

3) Sample space initialization

In order to reduce the randomness and invalid search in the search process, the predicted course angle is utilized to constrain the sample space. According to the carrier motion model, the carrier course angle is used as the only input, and the selection of the course angle determines a navigation search path, so that the carrier course angle is selected as a population sample.

And carrying out discrete sampling on the carrier course angle, initializing a sample space, and constraining the sample space according to the predicted course angle. Thus, the sample space is initialized as follows:

wherein, theta_iIs the carrier course angle, gamma_kTo predict course angle, D_θFor the sampling interval, N is the population sample size, β is the constraint threshold of the sample space, R_iIs [ (gamma) as_k-β)/D_θ,(γ_k+β)/D_θ]Any random integer of (a).

4) Judging whether the terminal point is reached

The geomagnetic parameter environment of the current location may be described as:

B＝{B₁,B₂,...B_n}

wherein, B₁,B₂,…B_nThe parameter elements of the geomagnetic field may be part or all of parameters of three components of the geomagnetic field (a north component, an east component, and a vertical component), a total intensity of the geomagnetic field, a declination, a horizontal component, and the like. Constructing an objective function of the ith geomagnetic parameter according to the geomagnetic information of the current position and the destination as follows:

f_i,k(B)＝(B_i,k-B_i,T)²

wherein B is_i,kAnd B_i,TAnd the ith geomagnetic parameter information of the carrier position and the destination at the time k respectively. And judging whether the carrier reaches the destination or not by observing the current objective function. And (3) considering magnitude and unit among geomagnetic parameters, and carrying out normalization processing on the target function to obtain:

wherein, B_i.0、B_i,TIth geomagnetism of initial position and destination of carrier respectivelyAmount information, when the carrier travels to the destination, the objective function value is theoretically 0, and therefore, when the objective function value of the current position magnetic parameter is small, F (B) is satisfied_k) ≦ ε, the navigator may be considered to reach the destination, where ε is a minimal amount near 0, set according to navigation accuracy. And if the above conditions are not met, jumping to the next step.

5) Sample selection

Under the condition of no historical information, a sample is randomly selected from the sample space in a preference-free mode to serve as a carrier course angle, and the probability of each sample being selected is the same. Thus, the probability that a certain course angle is selected depends on the fraction of the same number of samples. Time K, population sample θ_iThe selected probability is:

6) posterior evaluation of samples

Effective evaluation of the sample is very important for searching the navigation path, and is directly related to whether the search path is correct or not. In the existing evolutionary algorithm, monotone decreasing of a target function is taken as an evaluation criterion, and if the target function of the k +1 step is smaller than the kth step, a corresponding sample is considered to be better, and propagation operation is carried out. However, in this case, the navigation path that can be obtained travels in the direction of the maximum change of one of the geomagnetic parameters, and deviates from the shortest navigation path. The sample merit function is improved here for a search as close to the shortest path as possible. As shown in fig. 2, in order to converge the geomagnetic parameters as soon as possible, the input heading angle is required to satisfy:

(B_aim-B_k)//(B_k+1-B_k)

thus, it is possible to align the vector (B)_aim-B_k) And (B)_k+1-B_k) Angle therebetween

And observing to evaluate the quality of the sample.

7) Sample population update

Updating population samples based on the evaluation criteria:

wherein, P_pThe reproduction ratio of the sample. And when the sample does not meet the propagation conditions, replacing the current sample with a randomly generated sample to realize real-time update of the population. And repeating the steps until the navigation is completed.

The following detailed description of embodiments of the invention is intended to be illustrative, and not to be construed as limiting the invention.

To verify the effectiveness of the present invention, simulation experiments were performed under MATLAB. An International geomagnetism model (International geomagnetism Reference Field) IGRF-12 is used for simulating an actual Geomagnetic Field environment. In the embodiment, a magnetic dip angle and a magnetic declination angle are selected as magnetic parameters for heading angle prediction, a geomagnetic north component, a geomagnetic east component and a geomagnetic vertical component are selected as parameter elements for navigation search, initial point coordinates are set to be 5 degrees of north latitude, 5 degrees of east longitude, and initial position geomagnetic parameters are set to be B_0,x＝31165nT,B_0,y＝-1798.6nT,B_0,z-9342.8 nT; setting destination coordinates as 10 degrees of north latitude and 10 degrees of east longitude and corresponding geomagnetic parameters as B_T,x＝33756nT,B_T,y＝-680.8nT,B_T,z-1939.5 nT; the population size N is 35; sample sampling interval D_θIs 1 degree; in order to improve the navigation accuracy, the carrier movement step length is 3 nautical miles in the initial stage, namely, the course angle is updated every 3 nautical miles, and when the objective function is less than 0.005, the carrier movement step length is 1.5 nautical miles; the sample constraint threshold is 40 °; and when the condition is met, the carrier is considered to reach the destination, and the navigation is finished.

In order to illustrate the effectiveness of the algorithm provided by the invention, the algorithm after the improvement is compared with the algorithm before the improvement under the conditions of no geomagnetic abnormal area and geomagnetic abnormal area.

Navigation is performed in a region without geomagnetic anomaly, search tracks of two algorithms are shown in fig. 3, and convergence processes of three geomagnetic navigation parameters are shown in fig. 4 and 5. As can be seen from the navigation tracks in fig. 3, the tracks of the two navigation algorithms have a large difference, the randomness in the initial stage of the evolutionary search strategy before improvement is strong, the optimal path is found through continuous trial and error, the variation of the carrier course angle is large, and the method is not suitable for engineering application. Meanwhile, due to inaccuracy of the evaluation criterion, simultaneous and same-place convergence of multiple parameters cannot be guaranteed, so that the navigation search path deviates from the optimal path, and time and resource waste is caused. The improved algorithm restrains the sample space, thereby greatly reducing the randomness of the carrier motion and accelerating the search efficiency. The improvement on the evaluation function enables the navigation path to be close to the optimal path, the search of the path can be adjusted in real time in the carrier advancing process, the anti-interference capability is strong, the navigation track is relatively straight, and the engineering application is facilitated. The convergence condition of the geomagnetic parameters can also show that the convergence curves of the two algorithms can gradually approach 0 along with the accumulation of time, and the vector is guided to continuously approach to the target position. However, the comparison shows that in the improved algorithm, the convergence speed of geomagnetic multiple parameters is high, the convergence curve is smooth, the corresponding navigation path is stable, the simultaneous and same-place convergence of the multiple parameters can be basically realized, and the effectiveness of the algorithm improvement is proved.

The geomagnetic abnormal field is caused by uneven distribution of magnetic rocks in the earth crust, so that the magnitude and the strength of the geomagnetic abnormal field are different in space. The strength of weak anomalous fields is less than 1nT, while the strength of some strong anomalous fields can reach even several times the main magnetic field. The geomagnetic abnormal area is inevitably encountered in the navigation process during long-term navigation. In a simulation experiment, the geomagnetic abnormal region is constructed by a normal geomagnetic field and a multi-peak function, the geomagnetic abnormal region is 6-8 degrees N of north latitude, and 5.5-7.5 degrees of east longitude. The search trajectories of the two algorithms are shown in fig. 6, and the convergence process of the three geomagnetic navigation parameters is shown in fig. 7 and 8. As can be seen from the navigation track in fig. 6, since the constraint relationship between the geomagnetic parameters in the geomagnetic abnormal region and the motion of the carrier changes, the algorithm before improvement falls into a chaotic state during searching, and cannot smoothly pass through the abnormal region, so that the anti-interference performance is poor. The improved path searched by the algorithm shows that the path does not fall into a local abnormal area although the path generates some fluctuation when the carrier passes through the abnormal area, and the carrier can quickly find the optimal course again after leaving the abnormal area, so as to guide the carrier to reach the destination. Similarly, as can be seen from the convergence curve of the three components of the geomagnetism, when the carrier reaches an abnormal region, the algorithm before improvement still searches for a path according to the constraint relation between the geomagnetic component and the path of the abnormal region, so that the carrier falls into a chaotic search state and cannot complete the navigation task; the improved algorithm and the constraint of the sample space enable the algorithm to break through the constraint relation between the geomagnetic component and the path, so that the abnormal area can be smoothly walked out, the re-search of the optimal path can be rapidly realized, and the anti-interference performance is strong.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.

Claims (1)

1. The bionic navigation algorithm for the multi-target evolutionary search based on the assistance of the geomagnetic gradient comprises the following specific steps and is characterized in that:

step 1: acquiring geomagnetic parameters of a position where a carrier is located at the current moment and a target position;

the elements of the geomagnetic parameters comprise three components of a magnetic field, namely a north component, an east component, a vertical component, the total intensity of the magnetic field, a horizontal component of the magnetic field, a declination angle and a dip angle;

step 2: calculating gradient information of geomagnetic parameters, predicting a course angle according to a simultaneous and same-ground convergence principle of multiple geomagnetic parameters, and further constraining a sample space of an evolutionary algorithm;

and step 3: judging whether the destination is reached: constructing an objective function according to the current position and geomagnetic parameters of the destination, judging whether the carrier reaches the destination or not by observing the current objective function, and stopping searching to finish navigation if the carrier reaches the destination; otherwise, jumping to the step 4;

step 3 is to guide the carrier to the destination as soon as possible, and multiple parameters should be kept to be converged simultaneously and simultaneously as much as possible in the path searching process, that is, the following conditions are met:

wherein, B_i,k,B_j,kTwo different geomagnetic parameters at time k, B_i,TB_j,TTwo geomagnetic parameters at the target position, in the above formula, the geomagnetic parameter B at the time of k +1_i,k+1B_j,k+1The geomagnetic parameter and the geomagnetic gradient information at the previous time can be expressed as:

the course angle gamma is obtained by simultaneously solving the two formulas_kComprises the following steps:

carrying out sample space constraint on the evolutionary search algorithm based on the predicted course angle;

considering the magnitude and unit between geomagnetic parameters, normalizing the loss function constructed by the current position of the carrier and the target position is as follows:

wherein, B_i,0、B_i,T、B_i,kI-th geomagnetic parameter information including initial position, destination and k-th carrier position of the carrier, loss function value when the carrier moves to the destinationTheoretically 0, therefore, when the value of the loss function of the magnetic parameter at the current position is small, F (B) is satisfied_k) If the navigation precision is less than or equal to epsilon, the navigator can be considered to reach the destination, wherein epsilon is a very small quantity close to 0, and the navigation precision is set;

and 4, step 4: selecting the motion direction of the next step according to an evolutionary algorithm, performing posterior evaluation on the executed sample, and updating the sample space in real time; repeating the steps until the navigation is completed;

and 4, searching the shortest path as close as possible, improving the sample evaluation function, and requiring the input heading angle to meet the following requirements in order to enable all geomagnetic parameters to be converged as soon as possible:

(B_aim-B_k)//(B_k+1-B_k)

thus, it is possible to align the vector (B)_aim-B_k) And (B)_k+1-B_k) Angle therebetween

Observations were made to evaluate the quality of the samples:

the sample evaluation function is improved, and a corresponding population updating rule is formulated based on the improved evaluation criterion as follows:

wherein, P_pAnd (4) for the reproduction proportion of the sample, when the sample does not meet the reproduction condition, replacing the current sample with the randomly generated sample to realize the real-time update of the population.

Images