Deep-sea-oriented fault underwater vehicle positioning and search-and-rescue system and method

Abstract

The invention discloses a deep-sea-oriented fault underwater vehicle positioning, searching and rescuing system and method. The system comprises a seabed environment construction module, an underwater vehicle position prediction module and an underwater vehicle searching module. The seabed environment construction module is used for constructing a seabed environment model of a sea area where the target underwater vehicle is located; the underwater vehicle position prediction module is used for constructing a position prediction model to predict the fault position of the target underwater vehicle in the seabed environment model when the target underwater vehicle breaks down; and the underwater vehicle searching module is used for searching and rescuing the target underwater vehicle based on the fault position. Aiming at the problems of position prediction and search and rescue of the deep-sea underwater vehicle, an underwater vehicle position prediction model based on federated Kalman filtering and a Markov chain Monte Carlo method is established, and searching and positioning of the underwater vehicle under the three conditions of normal navigation, power out-of-control and power failure are achieved.

Description

Technical Field

[0001] The present invention relates to the field of deep-sea submersible positioning, and in particular to a deep-sea faulty submersible positioning and search and rescue system and method. Background Art

[0002] At present, the global ocean covers an area of ​​about 360 million square kilometers, equivalent to 71% of the earth's surface area. As one of the four major strategic spaces for human survival, the ocean has great practical significance and development potential. Internationally, areas with a water depth of more than 1,000 meters are usually defined as deep sea, with an average depth of about 3,347 meters. The deep sea area accounts for 92.4% and 65% of the ocean and earth's area.

[0003] With economic growth and technological advancement, human activities in the deep sea are increasing. However, the deep sea environment is extremely harsh, resulting in frequent deep sea accidents. Important navigation equipment, weapons and equipment, submarines and other equipment often sink to the bottom of the sea, causing huge losses. Therefore, fast and efficient deep sea emergency detection, search and salvage capabilities are essential for national deep sea activities. In order to prevent the loss of contact with submersibles and possible mechanical failures, it is urgent to develop a model that can accurately predict the change of submersible position over time. Unlike the land and sea surface environment, the deep sea environment has complex ocean currents, changes in seawater density and seabed topography, which can cause the submersible to deviate from the expected position. In this context, safety accidents of deep-sea submersibles occur frequently, which brings great challenges to deep-sea resource exploration and scientific research. In order to reduce the losses of deep-sea accidents, timely find out the causes of the accidents and carry out rescue, fast and efficient deep sea emergency detection, search and salvage capabilities are particularly important. Summary of the invention

[0004] In order to solve the technical problems in the above background, the present invention realizes the search and positioning of the submersible in different situations by establishing a submersible position prediction model based on the federated Kalman filter and the Markov chain Monte Carlo method.

[0005] To achieve the above-mentioned purpose, the present invention provides a deep-sea faulty submersible positioning and search and rescue system, comprising: a seabed environment construction module, a submersible position prediction module and a submersible search module;

[0006] The seabed environment construction module is used to construct a seabed environment model of the sea area where the target submersible is located;

[0007] The submersible position prediction module is used to construct a position prediction model to predict the fault position of the target submersible in the seabed environment model when a fault occurs;

[0008] The submersible search module is used to search and rescue the target submersible based on the fault location.

[0009] Preferably, the seabed environment model includes a seabed terrain model and a seabed ocean current model of the target sea area, and the seabed environment construction module includes a terrain model construction unit and an ocean current model construction unit;

[0010] The terrain model construction unit is used to construct a seabed terrain model of the target sea area;

[0011] The ocean current model building unit is used to build a submarine ocean current model of the target sea area.

[0012] Preferably, the terrain model construction unit constructs the seabed terrain model by comprehensively analyzing terrain factors based on a digital elevation model, and the terrain factors include: elevation, slope and slope direction.

[0013] Preferably, the ocean current model building unit uses multiple Lamb vortices and Navier-Stokes equations to build a submarine ocean current model. First, the ocean current V C =(u C ,v C ) to simulate:

[0014]

[0015] Among them, u C represents the velocity component of the ocean current in the horizontal direction; v C It represents the velocity component of the ocean current in the vertical direction; represents the spatial coordinates, represents the center coordinate of the vortex, represents the strength of the eddy current; t represents time; y represents the y coordinate of the current current position; y0 represents the y coordinate of the eddy current center; x represents the x coordinate of the current current position; x0 represents the x coordinate of the eddy current center;

[0016] Based on the generated two-dimensional ocean current, the three-dimensional eddy current V C =(u C ,v C ,w C ) is modeled as a multi-layer two-dimensional eddy current, and the vertical component of each layer is calculated using the probability density function of the normal distribution:

[0017]

[0018] in, represents the spatial coordinate; w is the vortex radius covariance matrix; F represents the amplitude or intensity coefficient used to adjust the vertical component; det represents the matrix; T represents the transposed matrix; w C represents the vertical ocean current component at the spatial coordinate S;

[0019] The entire area is composed of multiple eddy current fields, and the superposition formula is as follows:

[0020]

[0021] in, For coordinates The total superposition of the eddy field at the location; eddy represents the eddy term, which is used to simulate the eddy part of the ocean current; rand represents the random number generation function, which is often used to simulate random processes or uncertainties.

[0022] Preferably, the submersible position prediction model includes a trajectory model and a fault position model, and the position prediction module includes a trajectory prediction unit and a fault position prediction unit;

[0023] The trajectory prediction unit is used to predict the motion trajectory of the submersible during normal navigation;

[0024] The fault location prediction unit is used to construct the fault location model and predict the fault location of the submersible based on the motion trajectory prediction of the submersible.

[0025] Preferably, the trajectory prediction unit comprises: a collection device, a federal Kalman filter and a prediction device;

[0026] The acquisition device is used to obtain parameter information of the target submersible;

[0027] The federal Kalman filter is used to perform fusion processing on the parameter information to obtain fusion information;

[0028] The prediction device is used to complete the motion trajectory prediction of the normal navigation of the submersible based on the fusion information.

[0029] Preferably, the submersible search module includes: a path planning unit, a search and rescue unit and a sonar search unit;

[0030] The path planning unit is used to plan the shortest search and rescue path based on the fault location;

[0031] The search and rescue unit is used to search and rescue the faulty submersible based on the shortest search and rescue path;

[0032] The sonar search unit is used to assist the search and rescue unit in searching and rescuing a faulty submersible.

[0033] The present invention also provides a method for locating and rescuing a faulty submersible in the deep sea, which is applied to the above system and comprises the following steps:

[0034] Construct a seabed environment model of the sea area where the target submersible is located;

[0035] Constructing a position prediction model to predict the fault position of the target submersible in the seabed environment model when a fault occurs;

[0036] Based on the fault location, search and rescue the target submersible.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention aims at the position prediction and search and rescue problems of deep-sea submersibles. By establishing a submersible position prediction model based on the federated Kalman filter and the Markov chain Monte Carlo method, the present invention realizes the search and positioning of the submersible in three situations: normal navigation, power loss of control and power failure. At the same time, through the submersible search model based on the genetic algorithm and the Archimedean spiral geometry model, the search and rescue of the faulty submersible is realized. The model greatly shortens the rescue time and improves the rescue success rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0040] Figure 1 A schematic diagram of the system structure of an embodiment of the present invention;

[0041] Figure 2 A schematic diagram of the overall framework of the system workflow of an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of the working process of the collection device according to an embodiment of the present invention;

[0043] Figure 4 A schematic diagram of the forces when a submersible fails and causes power outage according to an embodiment of the present invention;

[0044] Figure 5 A schematic diagram of the forces when the power of the submersible fails according to an embodiment of the present invention;

[0045] Figure 6 A schematic diagram of a search and rescue process of a submersible search module according to an embodiment of the present invention;

[0046] Figure 7 Schematic diagram of a genetic algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] 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 are within the scope of protection of the present invention.

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Embodiment 1

[0050] like Figure 1 As shown, it is a schematic diagram of the system structure of this embodiment, including: a seabed environment construction module, a submersible position prediction module and a submersible search module; the seabed environment construction module is used to construct a seabed environment model of the sea area where the target submersible is located; the submersible position prediction module is used to construct a position prediction model to predict the fault position in the seabed environment model when the target submersible fails; the submersible search module is used to search and rescue the target submersible based on the fault position.

[0051] Among them, the seabed environment model includes the seabed terrain model and the seabed ocean current model of the target sea area, and the seabed environment construction module includes a terrain model construction unit and an ocean current model construction unit; the terrain model construction unit is used to construct the seabed terrain model of the target sea area; the ocean current model construction unit is used to construct the seabed ocean current model of the target sea area. The submersible position prediction model includes a trajectory model and a fault location model, and the position prediction module includes a trajectory prediction unit and a fault location prediction unit; the trajectory prediction unit is used to predict the motion trajectory of the normally sailing submersible; the fault location prediction unit is used to construct a fault location model and predict the fault location of the submersible based on the motion trajectory prediction of the submersible. The trajectory prediction unit includes: an acquisition device, a federal Kalman filter and a prediction device; the acquisition device is used to obtain parameter information of the target submersible; the federal Kalman filter is used to fuse the parameter information to obtain fusion information; the prediction device is used to complete the motion trajectory prediction of the normal navigation of the submersible based on the fusion information. The submersible search module includes: a path planning unit, a search and rescue unit and a sonar search unit; the path planning unit is used to plan the shortest search and rescue path based on the fault location; the search and rescue unit is used to search and rescue the faulty submersible based on the shortest search and rescue path; the sonar search unit is used to assist the search and rescue unit in searching and rescuing the faulty submersible.

[0052] The following will describe in detail how the present invention solves the technical problems in practical work in conjunction with this embodiment.

[0053] like Figure 2As shown, it is the overall framework diagram of the system workflow of this embodiment. First, the submarine environment construction module is used to construct the submarine environment model of the sea area where the target submersible is located. In the deep-sea environment, the position of the submersible will be affected by various factors, such as ocean currents, wind and tides, temperature, water pressure, etc. In order to simplify the calculation, the factors with relatively weak influence are removed and the factors with the same influence mechanism are combined. Under the premise of the assumption of environmental stability, the four main influencing factors of ocean currents, seawater density, seabed topography and sensor error are retained. Before establishing the submersible position prediction model, it is first necessary to model the seabed topography, ocean current distribution and seawater density field for the underwater environment (this embodiment takes the Ionian Sea as an example).

[0054] The terrain model building unit is used to build the Ionian Sea seabed terrain model based on the Digital Elevation Model (DEM). When building the terrain digital elevation model, this embodiment comprehensively analyzes multiple terrain factors such as elevation, slope and slope aspect to obtain the final terrain feature results. The results obtained in this way can not only improve the accuracy of the hierarchical detail model in selecting elevation points, but also more comprehensively describe the terrain characteristics of the search and rescue sea area.

[0055] The General Bathymetric Chart of the Oceans (GEBCO) is a global DEM dataset based on the Global Earth System Project. With the help of GEBCO, this embodiment obtains a discrete DEM dataset of submarine terrain with a resolution of 15" within a rectangular range in the search and rescue sea area. After interpolation and fitting, the terrain construction unit reconstructs the submarine terrain environment under the search and rescue sea area.

[0056] When underwater vehicles are sailing in the deep sea, in addition to considering the impact of terrain and obstacles, they must also consider the impact of ocean currents. During the navigation of underwater vehicles, favorable ocean currents can drive the movement of the vehicle and reduce energy consumption, while unfavorable ocean currents will cause obstacles and increase energy consumption. For this reason, it is necessary to consider the impact of ocean currents on the path of the vehicle during normal navigation and fault navigation of underwater vehicles.

[0057] Based on this, in this embodiment, the ocean current model construction unit uses multiple Lamb vortices and Navier-Stokes equations to construct a submarine ocean current model. First, the ocean current V C =(u C ,v C ) to simulate:

[0058]

[0059] Among them, uC represents the velocity component of the ocean current in the horizontal direction; v C It represents the velocity component of the ocean current in the vertical direction; represents the spatial coordinates, represents the center coordinate of the vortex, represents the strength of the eddy; t represents time; y represents the y coordinate of the current ocean current position; y0 represents the y coordinate of the eddy center; x represents the x coordinate of the current ocean current position; x0 represents the x coordinate of the eddy center.

[0060] Based on the generated two-dimensional ocean current, the three-dimensional eddy current V C =(u C ,v C ,w C ) is modeled as a multi-layer two-dimensional eddy current. The vertical component of each layer is calculated using the probability density function of the normal distribution:

[0061]

[0062] in, represents the spatial coordinate; w is the vortex radius covariance matrix; F represents the amplitude or intensity coefficient used to adjust the vertical component; det represents the matrix; T represents the transposed matrix; w C Represents the vertical ocean current component at the spatial coordinate S.

[0063] The entire area is composed of multiple eddy current fields, and the superposition formula is as follows:

[0064]

[0065] in, For coordinates The total superposition of the eddy field at the location; eddy represents the eddy term, which is used to simulate the eddy part of the ocean current; rand represents the random number generation function, which is often used to simulate random processes or uncertainties.

[0066] The submersible position prediction module is used to construct a position prediction model to predict the fault location in the seabed environment model when a target submersible fails.

[0067] Firstly, the trajectory prediction unit is used to predict the motion trajectory of the normally navigating submersible.

[0068] like Figure 3The figure shows a schematic diagram of the working process of the acquisition device of this embodiment, which acquires parameter information of the target submersible through the inertial measurement unit, water temperature and water pressure sensor, GPS sensor, and visual sensor installed on the submersible. The parameter information includes: attitude, depth, position, and motion environment information. Then, the federal Kalman filter is used to fuse the data collected by the submersible IMU, water temperature and water pressure sensor, GPS sensor, and visual sensor. For each sensor, the federal Kalman filter generates an independent filter for it, and obtains the optimal estimation result of the local filtering under the parallel operation of each local filter, and then uses the main filter to realize the global fusion processing of each local optimal result, so as to output the global optimal estimation result. The navigation system of the submersible is characterized by arbitrariness and discreteness, and the Kalman filter state equation can be described by the following formula:

[0069] X(k+1)=φ(k)X(k)+B(k)U(k)+W(k)

[0070] Where: X(k) is the state vector of the submersible, φ(k) is the state transfer matrix, B(k) is the input control matrix, U(k) is the control variable, and W(k) is the noise.

[0071] Assuming the number of sensors configured in the submersible is N, the number of local filters required is also N, and each filter can complete the filtering process of its corresponding sensor. For the i-th local filter X i (k+1) can be described as:

[0072]

[0073] Among them, Z i (k) and H i (k) are the observation quantity and the measurement matrix, V i (k) is the noise.

[0074] For the Kalman filter, the time update equation is:

[0075]

[0076] P i (k+1 / k)=Φ(k)P i (k / k)Φ T (k)+Φ(k)W(k)Φ T (k)

[0077] The description formula of the measurement update equation is:

[0078]

[0079] P i (k+1 / k+1)=I·Pi (k+1 / k)-K i (k+1)H i (k+1)P i (k+1 / k)

[0080] Among them, S i (k+1) is the covariance matrix, After further prediction, the state value of the submersible is expressed as Its covariance is expressed as P i (k+1 / k), Kalman gain is represented by K i (k+1), the adjusted state of the submersible is expressed as The error covariance is denoted as P i (k+1 / k+1), the identity matrix is ​​denoted as I.

[0081] In Kalman filtering, As the initial point, by continuously advancing the form, each local filter estimates the local state of the sensor data collected, and obtains the local best estimation result When W(k) and V i When (k) is a white noise with a mean value of 0 and the local estimates are independent of each other, the multi-sensor information fusion result can be obtained.

[0082] After the fusion processing of the information collected by the multi-sensor of the submersible is realized, the prediction device can realize the track prediction of the submersible during normal navigation according to the previously established submersible itself and motion environment model. In this embodiment, assuming that the starting coordinates of the submersible are (36.99N18.01E, -3447m), the final predicted end coordinates are (36.81N18.17E, -3701m).

[0083] After the trajectory prediction is completed, the fault position prediction unit constructs a fault position prediction model based on the Markov chain Monte Carlo method to predict the position of the faulty submersible. Specifically, based on the Monte Carlo method, the faulty submersible is defined as a single particle, and the movement trajectory of the particle is regarded as a random process. This embodiment assumes that the position of the faulty submersible is expressed as a Markov process, and the transfer probability of floating objects from one position to the next position is modeled. Under the same rule, a large number of these particles are copied and it is assumed that the movement of these particles is unrelated. Then, according to the statistical properties of the distribution of these particles after Δt time, the probability area of ​​the drift of the faulty submersible can be analyzed. As the number of simulations increases, the probability area of ​​the faulty submersible will be more accurate, but at the same time, the amount of calculation will increase.

[0084] The submersible position prediction model essentially predicts the position of the faulty submersible by quantifying a series of uncertain factors (including the last known position of the faulty submersible, the type of search and rescue object, ocean currents, and its own residual power, etc.), quantifying a series of uncertain factors that cause the drift of the faulty submersible, and using this to predict the possible position of the faulty submersible. This is the main difficulty faced in the search process of deep-sea faulty submersibles.

[0085] Since the application scenario of the system in this embodiment is a deep-sea environment, the following assumptions are made:

[0086] 1. The effect of wave force is ignored in the position prediction model, and only the effect of deep ocean current force and the power of the faulty submersible itself are considered.

[0087] 2. As the deep ocean currents and the submersible's own power change, the motion of the faulty submersible also changes (infinite acceleration).

[0088] 3. In the position prediction model, only the initial fault position error, ocean current force error and self-power error are considered.

[0089] When a submersible fails, two situations may occur: power out of control and power failure. The kinematic models of the submersible are different in these two situations. In the modeling of these two models, for the convenience of calculation, the submersible is regarded as a particle and the kinematic equations of the submersible are established respectively.

[0090] Submersible malfunction caused power outage

[0091] When a submersible fails and loses power, the force on the submersible can be simplified to its own power F s , the buoyancy F f , the thrust F exerted by the ocean current on the submersible t , the resistance F of the submersible moving in water d and the gravity F acting on the submersible itself g ,like Figure 4 shown.

[0092] The kinematic equation of the submersible is expressed as:

[0093]

[0094] Where m is the mass of the submersible, and the resistance F d The size is expressed as:

[0095]

[0096] Where ρ is the density of water, ν is the speed of the submersible relative to the water, and C d is the drag coefficient, and A is the water-facing area of ​​the submersible.

[0097] The thrust F of the ocean current on the submersible t It can be solved by the mathematical model of deep-sea submersible interference under the action of ocean current.

[0098] Submersible malfunction caused power failure

[0099] When the submersible fails, the force on the submersible can be simplified to its own power F s =0, such as Figure 5 shown.

[0100] At this time, the kinematic equation of the submersible is expressed as:

[0101]

[0102] In order to optimize the prediction results of the submersible under the Ionian Sea after a failure (power loss and power outage), this embodiment predicts the position of the submersible based on the Markov Chain Monte Carlo method.

[0103] After the above assumptions and model simplification, the position prediction model is simplified to the superposition of the ocean current force velocity and the residual velocity of its own power, that is:

[0104]

[0105] Among them, x e (t) is the position of the faulty underwater vehicle at time t, x e0 is the initial position of the faulty submersible, V(t′) is the drift model of the faulty submersible, L(t′) is the residual power speed of the faulty submersible, μ W (t′) is the ocean current velocity around the faulty submersible.

[0106] Assume that the position change of the faulty submersible is expressed as a Markov process:

[0107] p(x t+1 / x t ,x t-1 ,x t-2 ,…,x1)=p(x t+1 / x t )

[0108] That is, the probability density function of the future state depends only on the current state, and has nothing to do with the special process by which the model system reaches this state. Based on the assumption that the position of the faulty submersible is a Markov process, the evolution of the drift trajectory satisfies the following formula:

[0109] d(x e )=V(x e ,t)dt+dε

[0110] Among them, xe is the position of the faulty submersible, V is the velocity function of the faulty submersible under the action of force; the factors affecting V include the residual power of the faulty submersible itself and the effect of ocean currents; dε is a random disturbance term with zero mean and known variance, representing the initial fault position error, ocean current force error and self-power error. In order to simplify the calculation, the ocean current error and self-power error are regarded as Gaussian distribution.

[0111] The power change of a faulty submersible after power outage is nonlinear, and the expression of the power decay curve is usually described by an exponential decay function, which is a common model that conforms to the decay behavior of many physical systems. This embodiment uses the general form of an exponential decay function to describe the power decay of a faulty submersible after power outage:

[0112] P(t)=P0×e -αt

[0113] Among them, P(t) is the power value at time t, P0 is the initial power value, α is the decay rate (usually a positive number), and t is time. This function indicates that as time increases, the power of the faulty vehicle decays at an exponential rate.

[0114] In the search process, the first step is to determine the initial fault time and position of the search target. If the initial error is very small, it can be considered that the fault position is at one point. From another perspective, if the initial position and fault time are known very little, the initial fault position error will be relatively large. Therefore, the initial fault position error determines the distribution of the initial particles, which will greatly affect the future search area. This embodiment uses the following parameters to define the distribution of the initial particles: the earliest possible time of the fault, the position (longitude and latitude) and the error range; the latest possible time of the fault, the position (longitude and latitude) and the error range. The distribution of the initial particles satisfies the circular normal distribution, and the standard deviation is:

[0115] δ=r / 2

[0116] Among them, the error range r satisfies the linear distribution between r0 and r1.

[0117] In this embodiment, it is assumed that the earliest position is (36.90N, 18.19E), the error range r0 is 8m, the latest position x1 is (37.9001N, 18.1901E), the error range r1 is 10m, and the simulation is based on Matlab. It is assumed that the number of environmental sampling of the position prediction model is 1000, the submersible loss time is 100s, and the submersible is initialized at 2500m below the sea. Next, the model is simulated and tested under the normal operation conditions and motion failure conditions of the submersible based on the Matlab simulation environment to obtain the position prediction result after the submersible fails.

[0118] After the location prediction model is built, the submersible search module is used to search and rescue the faulty submersible. The purpose of the search and rescue operation is to maximize the probability of successful search (POS), which depends on the probability of discovery (POD) and the probability of inclusion (POC). The specific relationship is as follows:

[0119] POS=POD×POC

[0120] Among them, the probability of inclusion (POC) refers to the probability of the search target being in the search area. The above-mentioned position prediction model is to determine the search area with the highest probability of inclusion (POC) and does not involve the probability of discovery. The submersible search module uses the optimal search algorithm based on the determined search area to maximize the probability of discovery of the search target.

[0121] like Figure 6 As shown in the figure, it is a schematic diagram of the search and rescue process of the submersible search module; in order to shorten the search and rescue time, first, the path planning unit uses a genetic algorithm to optimize a shortest search and rescue path that traverses all submersible prediction points. Then the search and rescue unit is released to search along the shortest path, identify the sonar signal, and obtain the coordinates of the submersible; in this embodiment, the search and rescue unit is an unmanned underwater vehicle (AUV), and finally the AUV fixes the submersible and pulls it up through the winch of the rescue ship to complete the rescue.

[0122] If the submersible cannot be found, another solution is adopted. The sonar search unit is used to conduct a spiral search to capture the submersible signal; then the AUV is released to identify the signal and obtain the coordinates; finally, the AUV is used to fix the submersible and it is pulled up by the rescue ship's winch.

[0123] Specifically, the expression of the minimum path is as follows:

[0124]

[0125] Where m is the number of nodes; d is an m×m node distance matrix, where d ij is the distance from the i-th node to the j-th node; HG is the Hamiltonian circuit, that is, minimizing the path length (objective function); x ij is a decision variable. If the i-th node to the j-th node is included in the solution range, then the decision variable x ij is 1 if the value is true, otherwise it is 0.

[0126] The initial deployment point is selected as a vertical water surface point that provides a positioning point for the position prediction model; the genetic algorithm (GA) is used to optimize the search pattern, such as Figure 7 shown.

[0127] When the faulty submersible cannot be found in Scheme 1, Scheme 2 is adopted, which is based on the submersible search model of Archimedean spiral. Under the towing of the main ship or rescue ship, a spiral search is performed at the center of the predicted sea area until the faulty submersible is found. The polar coordinate equation of Archimedean spiral is:

[0128] r=α+β·θ

[0129] Among them, α and β are real numbers, which affect the direction and curve distance of the spiral respectively. Its Cartesian coordinate system equation is as follows:

[0130] x s =rcosθ=(a+β·θ)cosθ

[0131] y s =rsinθ=(α+β·θ)sinθ

[0132] Among them, x s The horizontal coordinate of the rectangular coordinate system corresponding to the host ship or rescue ship in polar coordinates; s It represents the ordinate of the rectangular coordinate system corresponding to the main ship or rescue ship in polar coordinates; θ represents the direction from the polar axis to a point on the spiral line.

[0133] Embodiment 2

[0134] This embodiment also provides a method for locating and rescuing a faulty submersible in the deep sea, the steps comprising:

[0135] S1. Construct a seabed environment model of the sea area where the target submersible is located.

[0136] In the deep sea environment, the position of the submersible will be affected by various factors, such as ocean currents, wind and tides, temperature, water pressure, etc. In order to simplify the calculation, the factors with relatively weak influence are removed and the factors with the same influence mechanism are combined. Under the premise of the assumption of environmental stability, the four main influencing factors of ocean currents, seawater density, seabed topography and sensor error are retained. Before establishing the submersible position prediction model, it is first necessary to model the seabed topography, ocean current distribution and seawater density field for the underwater environment (this embodiment takes the Ionian Sea as an example).

[0137] This embodiment establishes the seabed terrain model of the Ionian Sea based on the Digital Elevation Model (DEM). When establishing the terrain digital elevation model, this embodiment comprehensively analyzes multiple terrain factors such as elevation, slope and slope aspect to obtain the final terrain feature results. The results obtained in this way can not only improve the accuracy of the level of detail model in selecting elevation points, but also more comprehensively describe the terrain characteristics of the search and rescue sea area.

[0138] The General Bathymetric Chart of the Oceans (GEBCO) is a global DEM dataset based on the Global Earth System Project. With the help of GEBCO, this embodiment obtains a discrete DEM dataset of submarine terrain with a resolution of 15" within a rectangular range in the search and rescue sea area. After interpolation and fitting, the terrain construction unit reconstructs the submarine terrain environment under the search and rescue sea area.

[0139] When underwater vehicles are sailing in the deep sea, in addition to considering the impact of terrain and obstacles, they must also consider the impact of ocean currents. During the navigation of underwater vehicles, favorable ocean currents can drive the movement of the vehicle and reduce energy consumption, while unfavorable ocean currents will cause obstacles and increase energy consumption. For this reason, it is necessary to consider the impact of ocean currents on the path of the vehicle during normal navigation and fault navigation of underwater vehicles.

[0140] Based on this, in this embodiment, multiple Lamb vortices and Navier-Stokes equations are used to construct a submarine current model. First, the ocean current V C =(u C ,v C ) to simulate:

[0141]

[0142] Among them, u C represents the velocity component of the ocean current in the horizontal direction; v C It represents the velocity component of the ocean current in the vertical direction; represents the spatial coordinates, represents the center coordinate of the vortex, represents the strength of the eddy; t represents time; y represents the y coordinate of the current ocean current position; y0 represents the y coordinate of the eddy center; x represents the x coordinate of the current ocean current position; x0 represents the x coordinate of the eddy center.

[0143] Based on the generated two-dimensional ocean current, the three-dimensional eddy current V C =(u C ,v C ,w C ) is modeled as a multi-layer two-dimensional eddy current. The vertical component of each layer is calculated using the probability density function of the normal distribution:

[0144]

[0145] in, represents the spatial coordinate; w is the vortex radius covariance matrix; F represents the amplitude or intensity coefficient used to adjust the vertical component; det represents the matrix; T represents the transposed matrix; w C Represents the vertical ocean current component at the spatial coordinate S.

[0146] The entire area is composed of multiple eddy current fields, and the superposition formula is as follows:

[0147]

[0148] in, For coordinates The total superposition of the eddy field at the location; eddy represents the eddy term, which is used to simulate the eddy part of the ocean current; rand represents the random number generation function, which is often used to simulate random processes or uncertainties.

[0149] The submersible position prediction module is used to construct a position prediction model to predict the fault location in the seabed environment model when a target submersible fails.

[0150] S2. Construct a position prediction model to predict the fault location in the seabed environment model when a target submersible fails.

[0151] First, the motion trajectory of the normally sailing submersible is predicted, and the parameter information of the target submersible is obtained through the inertial measurement unit, water temperature and water pressure sensor, GPS sensor, and visual sensor installed on the submersible. The parameter information includes: attitude, depth, position, and motion environment information. Then, the federal Kalman filter is used to fuse the data collected by the submersible's IMU, water temperature and water pressure sensor, GPS sensor, and visual sensor. For each sensor, the federal Kalman filter generates an independent filter for it. Under the parallel operation of each local filter, the optimal estimation result of the local filter is obtained, and then the main filter is used to realize the global fusion processing of each local optimal result, so as to output the global optimal estimation result. The navigation system of the submersible is arbitrary and discrete. The Kalman filter state equation can be described by the following formula:

[0152] X(k+1)=φ(k)X(k)+B(k)U(k)+W(k)

[0153] Where: X(k) is the state vector of the submersible, φ(k) is the state transfer matrix, B(k) is the input control matrix, U(k) is the control variable, and W(k) is the noise.

[0154] Assuming the number of sensors configured in the submersible is N, the number of local filters required is also N, and each filter can complete the filtering process of its corresponding sensor. i (k+1) can be described as:

[0155]

[0156] Among them, Z i (k) and H i (k) are the observation quantity and the measurement matrix, V i (k) is the noise.

[0157] For the Kalman filter, the time update equation is:

[0158]

[0159] P i (k+1 / k)=Φ(k)P i (k / k)Φ T (k)+Φ(k)W(k)Φ T (k)

[0160] The description formula of the measurement update equation is:

[0161]

[0162] P i (k+1 / k+1)=I·P i (k+1 / k)-K i (k+1)H i (k+1)P i (k+1 / k)

[0163] Among them, S i (k+1) is the covariance matrix, After further prediction, the state value of the submersible is expressed as Its covariance is expressed as P i (k+1 / k), Kalman gain is represented by K i (k+1), the adjusted state of the submersible is expressed as The error covariance is denoted as P i (k+1 / k+1), the identity matrix is ​​denoted as I.

[0164] In Kalman filtering, As the initial point, by continuously advancing the form, each local filter estimates the local state of the sensor data collected, and obtains the local best estimation result When W(k) and V i When (k) is a white noise with a mean value of 0 and the local estimates are independent of each other, the multi-sensor information fusion result can be obtained.

[0165] After the fusion processing of the information collected by the multi-sensor of the submersible is realized, the prediction device can realize the track prediction of the submersible during normal navigation according to the previously established submersible itself and motion environment model. In this embodiment, assuming that the starting coordinates of the submersible are (36.99N18.01E, -3447m), the final predicted end coordinates are (36.81N18.17E, -3701m).

[0166] After the trajectory prediction is completed, a fault location prediction model is constructed based on the Markov chain Monte Carlo method to predict the location of the faulty submersible. Specifically, based on the Monte Carlo method, the faulty submersible is defined as a single particle, and the movement trajectory of the particle is regarded as a random process. This embodiment assumes that the position of the faulty submersible is expressed as a Markov process, and the transfer probability of floating objects from one position to the next position is modeled. Under the same rule, a large number of these particles are replicated and it is assumed that the movement of these particles is unrelated. Then, according to the statistical properties of the distribution of these particles after Δt time, the probability area of ​​the drift of the faulty submersible can be analyzed. As the number of simulations increases, the probability area of ​​the faulty submersible will be more accurate, but at the same time, the amount of calculation will increase.

[0167] The submersible position prediction model essentially predicts the position of the faulty submersible by quantifying a series of uncertain factors (including the last known position of the faulty submersible, the type of search and rescue object, ocean currents, and its own residual power, etc.), quantifying a series of uncertain factors that cause the drift of the faulty submersible, and using this to predict the possible position of the faulty submersible. This is the main difficulty faced in the search process of deep-sea faulty submersibles.

[0168] Since the application scenario of the method in this embodiment is a deep-sea environment, the following assumptions are made:

[0169] 1. The effect of wave force is ignored in the position prediction model, and only the effect of deep ocean current force and the power of the faulty submersible itself are considered.

[0170] 2. As the deep ocean currents and the submersible's own power change, the motion of the faulty submersible also changes (infinite acceleration).

[0171] 3. In the position prediction model, only the initial fault position error, ocean current force error and self-power error are considered.

[0172] When a submersible fails, two situations may occur: power out of control and power failure. The kinematic models of the submersible are different in these two situations. In the modeling of these two models, for the convenience of calculation, the submersible is regarded as a particle and the kinematic equations of the submersible are established respectively.

[0173] Submersible malfunction caused power outage

[0174] When a submersible fails and loses power, the force on the submersible can be simplified to its own power F s , the buoyancy F f , the thrust F exerted by the ocean current on the submersible t , the resistance F of the submersible moving in water d and the gravity F acting on the submersible itself g ,like Figure 4 shown.

[0175] The kinematic equation of the submersible is expressed as:

[0176]

[0177] Where m is the mass of the submersible, and the resistance F d The size is expressed as:

[0178]

[0179] Where ρ is the density of water, ν is the speed of the submersible relative to the water, and C d is the drag coefficient, and A is the water-facing area of ​​the submersible.

[0180] The thrust F of the ocean current on the submersible t It can be solved by the mathematical model of deep-sea submersible interference under the action of ocean current.

[0181] Submersible malfunction caused power failure

[0182] When the submersible fails, the force on the submersible can be simplified to its own power F s =0, such as Figure 5 shown.

[0183] At this time, the kinematic equation of the submersible is expressed as:

[0184]

[0185] In order to optimize the prediction results of the submersible under the Ionian Sea after a failure (power loss and power outage), this embodiment predicts the position of the submersible based on the Markov Chain Monte Carlo method.

[0186] After the above assumptions and model simplification, the position prediction model is simplified to the superposition of the ocean current force velocity and the residual velocity of its own power, that is:

[0187]

[0188] Among them, x e (t) is the position of the faulty underwater vehicle at time t, x e0is the initial position of the faulty submersible, V(t′) is the drift model of the faulty submersible, L(t′) is the residual power speed of the faulty submersible, μ W (t′) is the ocean current velocity around the faulty submersible.

[0189] Assume that the position change of the faulty submersible is expressed as a Markov process:

[0190] p(x t+1 / x t ,x t-1 ,x t-2 ,…,x1)=p(x t+1 / x t )

[0191] That is, the probability density function of the future state depends only on the current state, and has nothing to do with the special process by which the model system reaches this state. Based on the assumption that the position of the faulty submersible is a Markov process, the evolution of the drift trajectory satisfies the following formula:

[0192] d(x e )=V(x e ,t)dt+dε

[0193] Among them, x e is the position of the faulty submersible, V is the velocity function of the faulty submersible under the action of force; the factors affecting V include the residual power of the faulty submersible itself and the effect of ocean currents; dε is a random disturbance term with zero mean and known variance, representing the initial fault position error, ocean current force error and self-power error. In order to simplify the calculation, the ocean current error and self-power error are regarded as Gaussian distribution.

[0194] The power change of a faulty submersible after power outage is nonlinear, and the expression of the power decay curve is usually described by an exponential decay function, which is a common model that conforms to the decay behavior of many physical systems. This embodiment uses the general form of an exponential decay function to describe the power decay of a faulty submersible after power outage:

[0195] P(t)=P0×e -αt

[0196] Among them, P(t) is the power value at time t, P0 is the initial power value, α is the decay rate (usually a positive number), and t is time. This function indicates that as time increases, the power of the faulty vehicle decays at an exponential rate.

[0197] In the search process, the first step is to determine the initial fault time and position of the search target. If the initial error is very small, it can be considered that the fault position is at one point. From another perspective, if the initial position and fault time are known very little, the initial fault position error will be relatively large. Therefore, the initial fault position error determines the distribution of the initial particles, which will greatly affect the future search area. This embodiment uses the following parameters to define the distribution of the initial particles: the earliest possible time of the fault, the position (longitude and latitude) and the error range; the latest possible time of the fault, the position (longitude and latitude) and the error range. The distribution of the initial particles satisfies the circular normal distribution, and the standard deviation is:

[0198] δ=r / 2

[0199] Among them, the error range r satisfies the linear distribution between r0 and r1.

[0200] In this embodiment, it is assumed that the earliest position is (36.90N, 18.19E), the error range r0 is 8m, the latest position x1 is (37.9001N, 18.1901E), the error range r1 is 10m, and the simulation is based on Matlab. It is assumed that the number of environmental sampling of the position prediction model is 1000, the submersible loss time is 100s, and the submersible is initialized at 2500m below the sea. Next, the model is simulated and tested under the normal operation conditions and motion failure conditions of the submersible based on the Matlab simulation environment to obtain the position prediction result after the submersible fails.

[0201] S3. Based on the fault location, search and rescue the target submersible.

[0202] After the location prediction model is built, the search and rescue of the faulty submersible begins. The purpose of the search and rescue operation is to maximize the probability of successful search (POS), which in turn depends on the probability of discovery (POD) and the probability of inclusion (POC). The specific relationship is as follows:

[0203] POS=POD×POC

[0204] Among them, the probability of inclusion (POC) refers to the probability of the search target being in the search area. The above-mentioned position prediction model is to determine the search area with the highest probability of inclusion (POC) and does not involve the probability of discovery. The submersible search module uses the optimal search algorithm based on the determined search area to maximize the probability of discovery of the search target.

[0205] In order to shorten the search and rescue time, firstly, a genetic algorithm is used to optimize a shortest search and rescue path that traverses all predicted points of the submersible. Then the search and rescue unit is released to search along the shortest path, identify the sonar signal, and obtain the coordinates of the submersible; in this embodiment, the search and rescue unit is an unmanned underwater vehicle (AUV), and finally the AUV fixes the submersible and pulls it up through the winch of the rescue ship to complete the rescue.

[0206] If the submersible cannot be found, another solution is adopted. The sonar search unit is used to conduct a spiral search to capture the submersible signal; then the AUV is released to identify the signal and obtain the coordinates; finally, the AUV is used to fix the submersible and it is pulled up by the rescue ship's winch.

[0207] Specifically, the expression of the minimum path is as follows:

[0208]

[0209] Where m is the number of nodes; d is an m×m node distance matrix, where d ij is the distance from the i-th node to the j-th node; HG is the Hamiltonian circuit, that is, minimizing the path length (objective function); x ij is a decision variable. If the i-th node to the j-th node is included in the solution range, then the decision variable x ij is 1 if the value is true, otherwise it is 0.

[0210] The initial deployment points are selected as vertical water surface points that provide positioning points for the position prediction model; a genetic algorithm (GA) is used to optimize the search pattern.

[0211] When the faulty submersible cannot be found in Scheme 1, Scheme 2 is adopted, which is based on the submersible search model of Archimedean spiral. Under the towing of the main ship or rescue ship, a spiral search is performed at the center of the predicted sea area until the faulty submersible is found. The polar coordinate equation of Archimedean spiral is:

[0212] r=α+β·θ

[0213] Among them, α and β are real numbers, which affect the direction and curve distance of the spiral respectively. Its Cartesian coordinate system equation is as follows:

[0214] x s =rcosθ=(a+β·θ)cosθ

[0215] y s =rsinθ=(α+β·θ)sinθ

[0216] Among them, x s The horizontal coordinate of the rectangular coordinate system corresponding to the host ship or rescue ship in polar coordinates; sIt represents the ordinate of the rectangular coordinate system corresponding to the main ship or rescue ship in polar coordinates; θ represents the direction from the polar axis to a point on the spiral line.

[0217] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A deep-sea faulty submersible positioning and search and rescue system, characterized in that: include: Seabed environment construction module, submersible position prediction module and submersible search module; The seabed environment construction module is used to construct a seabed environment model of the sea area where the target submersible is located; The submersible position prediction module is used to construct a position prediction model to predict the fault position of the target submersible in the seabed environment model when a fault occurs; The submersible search module is used to search and rescue the target submersible based on the fault location.

2. The deep-sea faulty submersible positioning and search and rescue system according to claim 1 is characterized in that: The seabed environment model includes a seabed terrain model and a seabed ocean current model of the target sea area, and the seabed environment construction module includes a terrain model construction unit and an ocean current model construction unit; The terrain model construction unit is used to construct a seabed terrain model of the target sea area; The ocean current model building unit is used to build a submarine ocean current model of the target sea area.

3. The deep-sea faulty submersible positioning and search and rescue system according to claim 2 is characterized in that: The terrain model construction unit constructs the seabed terrain model by comprehensively analyzing terrain factors based on the digital elevation model. The terrain factors include: elevation, slope and slope direction.

4. The deep-sea faulty submersible positioning and search and rescue system according to claim 2, characterized in that: The ocean current model building unit uses multiple Lamb vortices and Navier-Stokes equations to build a submarine ocean current model. First, the ocean current V C =(u C ,v C ) to simulate: Among them, u C represents the velocity component of the ocean current in the horizontal direction; v C It represents the velocity component of the ocean current in the vertical direction; represents the spatial coordinates, represents the center coordinate of the vortex, represents the strength of the eddy current; t represents time; y represents the y coordinate of the current current position; y0 represents the y coordinate of the eddy current center; x represents the x coordinate of the current current position; x0 represents the x coordinate of the eddy current center; Based on the generated two-dimensional ocean current, the three-dimensional eddy current V C =(u C ,v C ,w C ) is modeled as a multi-layer two-dimensional eddy current, and the vertical component of each layer is calculated using the probability density function of the normal distribution: in, represents the spatial coordinate; w is the covariance matrix of the eddy radius l; F represents the amplitude or intensity coefficient used to adjust the vertical component; det represents the matrix; T represents the transposed matrix; w C represents the vertical ocean current component at the spatial coordinate S; The entire area is composed of multiple eddy current fields, and the superposition formula is as follows: in, For coordinates The total superposition of eddy fields at ; eddy represents the eddy term; rand represents the random number generation function.

5. The deep-sea faulty submersible positioning and search and rescue system according to claim 1, characterized in that: The submersible position prediction model includes a trajectory model and a fault position model, and the position prediction module includes a trajectory prediction unit and a fault position prediction unit; The trajectory prediction unit is used to predict the motion trajectory of the submersible during normal navigation; The fault location prediction unit is used to construct the fault location model and predict the fault location of the submersible based on the motion trajectory prediction of the submersible.

6. The deep-sea faulty submersible positioning and search and rescue system according to claim 5, characterized in that: The trajectory prediction unit includes: a collection device, a federal Kalman filter and a prediction device; The acquisition device is used to obtain parameter information of the target submersible; The federal Kalman filter is used to perform fusion processing on the parameter information to obtain fusion information; The prediction device is used to complete the motion trajectory prediction of the normal navigation of the submersible based on the fusion information.

7. The deep-sea faulty submersible positioning and search and rescue system according to claim 1, characterized in that: The submersible search module includes: a path planning unit, a search and rescue unit and a sonar search unit; The path planning unit is used to plan the shortest search and rescue path based on the fault location; The search and rescue unit is used to search and rescue the faulty submersible based on the shortest search and rescue path; The sonar search unit is used to assist the search and rescue unit in searching and rescuing a faulty submersible.

8. A method for locating and rescuing a faulty submersible in the deep sea, the method being applied to the system according to any one of claims 1 to 7, characterized in that the steps include: Construct a seabed environment model of the sea area where the target submersible is located; Constructing a position prediction model to predict the fault position of the target submersible in the seabed environment model when a fault occurs; Based on the fault location, search and rescue the target submersible.

Images