Underwater vehicle flow field estimation method based on distributed consensus algorithm

Abstract

The invention provides an underwater vehicle flow field estimation method based on a distributed consensus algorithm. The method is suitable for an underwater environment which cannot be covered by GPS signals. The method does not need to directly measure the environmental flow velocity, and assumes that the aircraft can measure the relative position of a neighbor and the absolute position during floating. By converting these measurements into constraints of relative and absolute motion integral errors, the flow field estimation problem is converted into a non-linear inverse problem, and the set of non-linear equations is solved in a distributed manner. And each aircraft shares local flow field estimation with neighbors and updates an estimation value based on local constraints, so that efficient estimation of the flow field is realized. According to the method, effective work can still be kept when the communication network of the aircraft changes.

Description

Technical Field

[0001] The present invention belongs to the field of multi-vessel collaborative flow field estimation, and mainly relates to an underwater vehicle flow field estimation method based on a distributed consensus algorithm. Background Art

[0002] Ocean flow field information plays an important role in the fields of marine climate prediction, fishery management, pollutant diffusion estimation, and underwater robot navigation. Obtaining accurate flow field information can not only improve the navigation performance of underwater vehicles, but also effectively reduce the energy consumption of vehicles. At present, most of the ocean flow field measurements commonly used in practical applications are fixed-point long-term continuous observations with professional ADCP (Acoustic Doppler Current Profiler) and ADV (Acoustic Doppler Current Meter), or measurements are made using ocean observation radars. The operation is complicated and has high requirements for the collection equipment and environment. It is difficult to obtain real-time ocean flow field information during daily navigation tests or operations of underwater unmanned vehicles. It is particularly difficult to make each vehicle measure real-time flow field information during the collaborative operation of multiple underwater unmanned vehicles. Therefore, it is necessary to use flow field estimation methods to guide subsequent navigation operations. However, due to the complexity of the underwater environment, especially in areas where GPS signals cannot cover, accurate estimation of flow fields has always been a difficult problem.

[0003] In recent years, with the widespread use of autonomous underwater vehicles, the use of multiple vehicles to coordinate flow field estimation has become a hot research topic. By measuring the relative positions between vehicles and combining limited floating position data, the vehicles can estimate the flow field of the target sea area. However, many existing flow field estimation methods rely on direct flow velocity measurements, but in underwater environments, flow velocity measurement equipment is not always feasible or available. Therefore, when direct flow velocity measurements are not possible, how to efficiently estimate the flow field through indirect data becomes a challenge.

[0004] Although some existing centralized flow field estimation methods (such as motion tomography methods) can solve some problems, they usually require high computing resources and cannot adapt to large-scale or communication-constrained multi-vessel systems. In addition, the existing motion tomography methods only reconstruct the flow field through the actual floating position of the vehicle. The number of measurements is usually much smaller than the dimension of the flow field parameters to be estimated, resulting in the solution of this problem being an underdetermined problem. In this case, research methods on how to solve the nonlinear inverse problem for distributed flow field estimation have gradually received attention. However, the existing distributed methods still face problems such as unstable communication topology, low computational efficiency, and insufficient reliability in large-scale applications. Therefore, developing a distributed method that can efficiently perform flow field estimation under unstable communication conditions is of great significance for improving the navigation and environmental perception capabilities of underwater vehicles. Summary of the invention

[0005] The purpose of the present invention is to provide a method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm, aiming to solve the problem of nonlinear distributed flow field estimation caused by the lack of GPS signals in underwater environments. The method is suitable for underwater environments that cannot be covered by GPS signals, and the method does not require direct measurement of the ambient flow velocity. It is assumed that the vehicle can measure the relative position of its neighbors and the absolute position when it floats to the surface. By converting the measurement results into relative and absolute motion integral error constraints, the cooperative flow field estimation problem of large-scale cluster nodes with unstable communication topology is converted into a nonlinear inverse problem, which solves a set of determined nonlinear equations in a distributed manner. Next, a distributed consensus algorithm is proposed to solve these equations, in which each vehicle first shares its local flow field estimate with its neighbors, and then updates the estimate according to the local constraints to achieve distributed flow field estimation.

[0006] In order to solve the problems in the prior art, the present invention provides a method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm, comprising the following steps:

[0007] Step 1: Establish the flow field model and the motion model of each aircraft;

[0008] Step 2: The spacecraft measures the relative position and the absolute position when floating, converts the measurement results into relative and absolute motion integral error constraints, and converts the flow field estimation problem into a nonlinear inverse problem;

[0009] Step 3: Using a distributed consensus algorithm, without relying on direct flow velocity measurements, solve the nonlinear inverse problem described in step 2 to obtain a local flow field estimate; each vehicle first shares its local flow field estimate with its neighbors, and then updates the estimate based on local constraints. Through an iterative process, each vehicle continuously updates its flow field estimate until the preset convergence condition is reached.

[0010] Compared with the prior art, the present invention has the following beneficial effects:

[0011] (1) Traditional flow field estimation methods usually require direct measurement of flow velocity, which is often difficult to achieve in underwater environments. The present invention uses the relative position information between the vehicles and the absolute position information when floating, and can estimate the flow field without directly measuring the flow velocity, thereby reducing the implementation difficulty and cost.

[0012] (2) The present invention proposes a distributed algorithm that allows each underwater vehicle to independently perform flow field estimation without central control. This method allows each vehicle to perform calculations and information exchange in parallel, improving the efficiency of flow field estimation; reducing dependence on the central processing unit, and improving the reliability and robustness of the system.

[0013] (3) The method of the present invention can adapt to changes in the aircraft communication network and maintain the accuracy and stability of flow field estimation even when the network topology changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art or ordinary technicians, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 It is a schematic diagram of the principle of the method of the present invention.

[0016] Figure 2 Schematic diagram of cooperative flow field estimation in a discrete grid domain in an embodiment of the present invention.

[0017] Figure 3 This is a flow field simulation in an embodiment of the present invention.

[0018] Figure 4 This is the relative estimation error curve for Case 1 to Case 4.

[0019] Figure 5 Schematic diagram of eight aircraft moving in the simulated flow field.

[0020] Figure 6 There are three network structures consisting of eight aircraft.

[0021] Figure 7 It is the relative estimation error curve when the aircraft communication network changes. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0023] The purpose of the present invention is to provide a method for underwater vehicle flow field estimation based on a distributed consensus algorithm, aiming to solve the problem of nonlinear distributed flow field estimation caused by the lack of GPS signals in underwater environments. It is assumed that the vehicle can measure the relative position of its neighbors and the absolute position when it floats to the surface. By converting the measurement results into relative and absolute motion integral error constraints, the cooperative flow field estimation problem of large-scale cluster nodes with unstable communication topology is converted into an inverse problem, which solves a set of deterministic nonlinear equations in a distributed manner. Then, a distributed consensus algorithm is proposed to solve these equations, in which each vehicle first shares its local flow field estimate with its neighbors, and then updates the estimate according to the local constraints to achieve distributed flow field estimation.

[0024] The present invention proposes a method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm. The principle is as follows: Figure 1 As shown, the specific steps include:

[0025] Step 1: Establish the flow field model and the motion model of each aircraft

[0026] The discrete grid flow field is used to approximate the actual continuous flow field. The specific steps include: The horizontal area of ​​is divided into P = A × B grid cells. Let D(a,b) denote the (a,b)th cell, where a∈{1,2,...,A}, b∈{1,2,...,B}. Define index ι:=(a-1)B+b, then D ι ≡D(a,b), ι∈{1,2,...,P}. Grid unit D ι The flow rate in is considered constant and is denoted by in and Respectively represent f ι The x and y components of the flow field are defined by stacking the flow velocities of all grid cells. in and set up Represents the actual flow field discretized into a grid. The purpose of flow field estimation is to estimate the unknown vector f ★ .

[0027] There are n spacecraft, denoted as Move in D. Let t∈[0,∞) denote time, and the spacecraft v i The observation time interval is Aircraft i The actual position is expressed as Aircraft i and v j The actual relative position between ij (t): = rj (t)-r i (t), such as Figure 2 When the vehicle is navigating underwater, r ij (t) can be measured by an acoustic positioning system such as an ultra-short baseline (USBL). Since GPS signals are not available underwater, it is not possible to directly obtain the vehicle’s v i However, when t = 0 and t = T i When the spacecraft v i Float to the surface and measure actual position via GPS signal i (t). Aircraft v i The continuous time motion model of , considering the influence of the actual flow field, is expressed as:

[0028]

[0029] where f ★ (r i (t)) is the position r i The actual flow rate at is the speed of the vehicle relative to the water, expressed as:

[0030]

[0031] where θ i (t) is the spacecraft v i The yaw angle can be measured by the attitude sensor installed on the aircraft; It is a spacecraft v i The velocity against the water is determined by correlation with the propeller speed. Given an estimated flow field The spacecraft v can be calculated i The estimated trajectory of i (t), which satisfies:

[0032]

[0033] in Indicates estimated location The estimated flow velocity at . Accordingly, the estimated relative position is defined

[0034] Step 2: The spacecraft measures the relative position and the absolute position during buoyancy, converts the measurement results into relative and absolute motion integral error constraints, and converts the flow field estimation problem into a nonlinear inverse problem. The specific steps include the following:

[0035] Step 2.1: By calculating the actual flow field f ★ (r) and estimated flow field The difference between the actual trajectory of the spacecraft and the estimated trajectory is determined, thereby introducing absolute and relative motion integration errors. i The observation time interval Absolute motion integral error on The formula is as follows:

[0036]

[0037] Among them, r i (T i ) is the spacecraft v i At time T i The actual location of is the estimated position. On, the spacecraft i and The relative motion integral error between Defined as:

[0038]

[0039] in, Represents the aircraft v i The set of spacecraft whose relative positions are measured, possibly empty.

[0040] Step 2.2: Use the absolute and relative motion integrated errors to construct nonlinear error constraints related to the flow field in the form of a distributed framework (absolute motion integrated error constraints and relative motion integrated error constraints). i The absolute motion integral error constraint is set by setting d i =0, the formula is:

[0041]

[0042] Among them, v i (t) is the speed of the craft through water, is the estimated flow field.

[0043] Construct the relative motion integral error constraint. For the aircraft v i In time Surveying Vehicle The actual relative position of s=1,2,...,S ij By setting The relative motion integral error constraint can be derived as follows:

[0044]

[0045] In order to solve the problem of the lack of initial position and water speed of neighboring aircraft, the above equation is further rewritten as:

[0046]

[0047] in, For the aircraft v j The local calculation result is:

[0048]

[0049] and For the aircraft v j The result obtained through local calculation is expressed as:

[0050]

[0051] Through inter-vessel communication, the spacecraft v i and v j Can share their calculation results and Thus, the relative motion integral error constraint is established.

[0052] Step 2.3: Construct the distributed cooperative flow field estimation problem. In the distributed framework, the spacecraft v i The flow field is estimated by solving the following nonlinear equation:

[0053]

[0054] in

[0055]

[0056] Where * = x or y, And s∈{1,2,...,S ij}. Given an initial estimate of the flow field Design an iterative algorithm, the spacecraft v i According to the algorithm, a flow field estimation sequence is generated, namely So that when k→∞,

[0057] Step 3: Using a distributed consensus algorithm, without relying on direct flow velocity measurements, solve the nonlinear inverse problem described in step 2 to obtain a local flow field estimate; each vehicle first shares its local flow field estimate with its neighbors, and then updates the estimate based on local constraints. Through an iterative process, each vehicle continuously updates its flow field estimate until the preset convergence condition is reached.

[0058] Distributed cooperative flow field estimation algorithm 1 In the kth iteration, the spacecraft v iFirst, from its neighbors Receives the current estimate, and computes:

[0059]

[0060] in Yes j The local estimate at the end of the k-1th iteration and the beginning of the kth iteration. Then v i By sequentially Project onto each constraint in the nonlinear equation to perform the update:

[0061]

[0062] or

[0063]

[0064] In this way, v i Calculate in sequence Where n=1,2,...,K i .set up Then the kth iteration is completed. In the update equation, the relaxation parameter gradient Can be obtained by v i Use local information to calculate. For the gradient have:

[0065]

[0066]

[0067] The aircraft used in the method of the present invention are all conventional models of equipment; the method of the present invention can be implemented by those skilled in the art through programming.

[0068] The effectiveness of the present invention is verified below in conjunction with a specific implementation.

[0069] In a 1200×1200m 2 A vortex flow field is simulated in the region of Figure 3 (a)), where P = 9 × 9 and the maximum flow rate is 0.1 m / s. 18 aircraft (N = 18) are deployed in this area ( Figure 3 (b)), their velocity v i (t) is fixed at 0.45 m / s and the direction of motion remains parallel to the x-axis or y-axis during the motion. i and v j There is a relative position measurement between Table 1 shows the communication neighbor set C of each aircraft.i and the relative position measurement set M i , each aircraft and Listed in Table 1. The stopping condition is selected as k max =3000, relaxation parameter

[0070] Since there is no prior knowledge of the flow field, an initial estimate is set for each vehicle. The final estimated flow field and vehicle trajectory are shown in Figure 3 (c) and 3(d). It can be seen that the actual and estimated trajectories almost overlap, indicating the accuracy of the flow field estimation. The accuracy of the flow field estimation will inevitably degrade in areas where no aircraft passes. In practice, the flow velocity in these areas can be obtained by interpolation. In order to better illustrate the accuracy of the proposed distributed algorithm, the grid cells where no aircraft passes are not considered when calculating the estimation error. For comparison, the same flow field is estimated using a tree-based distributed method under the same simulation settings. Table 2 compares the estimation errors of the method proposed in the present invention and the tree-based distributed method. Table 2 shows the estimation errors of the two methods, where RMS represents root mean square error, VAR represents variance, and STD represents standard deviation.

[0071] The relative error is further defined as ||ff ★ || / ‖f ★ ‖×100%, the final relative errors of the proposed method and the tree-based distributed method are 3.113% and 2.893%, respectively. It can be seen that the performance of the proposed method is comparable to that of the tree-based distributed method. However, the proposed method does not require the establishment of a tree-shaped communication network between aircraft, which reduces the requirements for communication conditions between aircraft. In order to further explore the proposed distributed algorithm, four comparative simulations (cases 1 to 4) were performed. In these cases, the number of communication neighbors of each aircraft was gradually increased (Table 3 shows the number of communication neighbors of each aircraft in case 1 to case 4 ), that is, increase In the four cases, other simulation settings are kept the same as the above simulation. It is worth noting that in each case So that the total constraint equation used for flow field estimation remains the same in the four cases. Set k max =500, Figure 4 The relative estimation error curves for the four cases are shown. The results are consistent with the physical intuition that having more neighbors can speed up the transmission of the estimated flow field in the multi-vessel system, thereby speeding up of convergence.

[0072] Table 1

[0073]

[0074]

[0075] Table 2

[0076]

[0077] Table 3

[0078]

[0079] The effectiveness of the proposed algorithm under changing aircraft networks is continued to be studied.

[0080] Figure 5 (a) shows the flow field to be estimated, where P = 4 × 4, the x-component of the velocity increases from 0 to 0.05 m / s, and the y-component decreases from 0.05 to 0 m / s. Eight vehicles are deployed in the area ( Figure 5 (b)), their velocity through water is constant at 0.35 m / s, Setting k max =100, Initial guess

[0081] Figure 7 The relative estimation error curve is shown when the spacecraft communication network changes. Consider that the spacecraft communication network will change with the number of iterations, such as Figure 6 Shown (including Figure 6 (a) Figure 6 (b) Figure 6 (c) and Figure 6 (d)). Specifically, Figure 6 (a) and Figure 6 The network in (b) does not even satisfy the connectivity. Based on the above simulation settings, three simulation experiments are performed to estimate Figure 5 (a) The flow field shown: 1) Network 1 and Network 2 are used alternately, that is, network mod (k, 2) + 1 is used in the kth iteration; 2) Network 1, Network 1 and Network 3 are used alternately, that is, Network 1 is used twice and Network 3 is used once in every three iterations; 3) Network 1 and Network 3 are used alternately. Figure 7 The relative estimation error curves of the above three experiments are shown. It can be seen that the proposed algorithm can still successfully estimate the flow field even when the network changes. In particular, when the communication network of the multi-vessel system alternates between network 1 and network 2, the vehicles cannot establish a connected network at any time, resulting in the inability to form a tree-structured network, but the distributed method proposed in this paper can still work normally.

[0082] The innovation of the present invention is:

[0083] 1. The focus of the present invention is to propose a distributed consensus flow field estimation algorithm, which can effectively alleviate the problems of the centralized framework that cannot adapt to the simultaneous operation of large-scale cluster nodes or limited underwater communication, break through the limit on the number of underwater unmanned vehicle clusters, and improve the navigation and environmental perception capabilities of multiple underwater vehicles.

[0084] 2. The present invention solves the underdetermined problem of nonlinear solution that has always existed. It assumes that the vehicle can measure the relative position of its neighbors and the absolute position when it floats to the surface. By converting the measurement results into relative and absolute motion integral error constraints, the problem of cooperative flow field estimation for large-scale cluster nodes with unstable communication topology is converted into an inverse problem, which solves a set of deterministic nonlinear equations in a distributed manner.

[0085] 3. The method of the present invention can adapt to changes in the aircraft communication network and maintain the accuracy and stability of flow field estimation even when the network topology changes.

[0086] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, which all belong to the protection scope of the present invention.

Claims

1. A method for estimating flow field of underwater vehicles based on a distributed consensus algorithm, characterized in that: The following steps are involved: Step 1: Establish the flow field model and the motion model of each aircraft; Step 2: The spacecraft measures the relative position and the absolute position when floating, converts the measurement results into relative and absolute motion integral error constraints, and converts the flow field estimation problem into a nonlinear inverse problem; Step 3: Using a distributed consensus algorithm, without relying on direct flow velocity measurement, solve the nonlinear inverse problem described in step 2 to obtain a local flow field estimate; each vehicle first shares its local flow field estimate with its neighbors, and then updates the estimate based on local constraints. Through an iterative process, each vehicle continuously updates its flow field estimate until the preset convergence condition is reached.

2. The method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm according to claim 1, characterized in that: In step 1, a discrete grid flow field is used to approximate the actual continuous flow field, which specifically includes the following steps: The horizontal area is divided into P = A × B grid cells, each grid cell D ι The flow rate in is considered constant and is denoted by and Respectively represent f ι The x and y components of the flow field are defined by stacking the flow velocities of all grid cells together. in and set up Represents the actual flow field discretized into a grid. The purpose of flow field estimation is to estimate the unknown vector f ★ .

3. The method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm according to claim 2, characterized in that: In step 1, there are N spacecraft, denoted as v = {v1, ..., v N }, moving in D; let t∈[0,∞) represent time, spacecraft v i The observation time interval is Aircraft i The actual position is expressed as In the aircraft i After floating to the surface, you will get: aircraft v i and v j The actual relative position between ij (t): = r j (t)-r i (t), measured underwater by acoustic positioning system; aircraft v i The continuous time motion model of , considering the influence of the actual flow field, is expressed as: where f ★ (r i (t)) is the position r i The actual flow rate at is the velocity of the vehicle relative to the water; given an estimated flow field Computing Vehicle v i The estimated trajectory of i (t), which satisfies: in Indicates estimated location The estimated flow velocity at ; accordingly, define the estimated relative position 4. The method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm according to claim 3 is characterized in that: The step 2 specifically includes the following steps: Step 2.1: Calculate the actual flow field f ★ (r) and estimated flow field The difference between the actual trajectory and the estimated trajectory of the spacecraft is determined, thereby introducing the absolute and relative motion integral errors; defined in the spacecraft v i The observation time interval Absolute motion integral error on The formula is as follows: Among them, r i (T i ) is the spacecraft v i At time T i The actual location, is the estimated position; also in the time interval On, the spacecraft i and The relative motion integral error between Defined as: in, Represents the aircraft v i A collection of spacecraft that measure relative positions; Step 2.2: Use the absolute and relative motion integrated errors to construct nonlinear error constraints associated with the flow field in the form of a distributed framework. i The absolute motion integral error constraint is set by setting d i =0, the formula is: Among them, v i (t) is the speed of the craft through water, is the estimated flow field; Construct the relative motion integral error constraint, set For the aircraft v i In time Surveying Vehicle The actual relative position of s=1,2,...,S ij , by setting The relative motion integral error constraints are derived as follows: In order to solve the problem of the lack of initial position and water speed of neighboring aircraft, the above equation is further rewritten as: in, For the aircraft v j The local calculation result of is expressed as: and For the aircraft v j The result obtained through local calculation is expressed as: Through inter-vessel communication, the spacecraft v i and v j Share your calculation results and Thus, the relative motion integral error constraint is established; Step 2.3: Construct the distributed cooperative flow field estimation problem; in the distributed framework, the spacecraft v i The flow field is estimated by solving the following nonlinear equation: in, Where * = x or y, And s∈{1,2,...,S ij }; Given an initial estimate of the flow field Design an iterative algorithm, the spacecraft v i According to the algorithm, a flow field estimation sequence is generated, namely So that when k→∞, 5. The method for estimating the flow field of an underwater vehicle based on a distributed consensus algorithm according to claim 4 is characterized in that: The step 3 specifically comprises the following steps: In the kth iteration of the distributed cooperative flow field estimation algorithm, the spacecraft v i First, from its neighbors Receives the current estimate, and computes: in Yes j The local estimate at the end of the k-1th iteration and the beginning of the kth iteration, then v i By sequentially Project onto each constraint in the nonlinear equation to perform the update: or In this way, v i Calculate in sequence Where n=1,2,...,K i ;set up Then the kth iteration is completed; In the update equation, the relaxation parameter gradient By v i Calculate using local information; for gradient have:

Images